Long-term variability and effects on larval fish distribution of the Gulf of Mexico Loop Current and Rings

UNIVERSIDAD DE LAS PALMAS DE GRAN CANARIA DEPARTAMENTO DE FÍSICA TESIS DOCTORAL Long-term variability and effects on larval fish distribution of the Gulf of Mexico Loop Current and Rings (Variabilidad a largo plazo de la Corriente de Lazo y Anillos, y efectos en la distribuciones de larvas de peces en el Golfo de México) DAVID LINDO ATICHATI ENERO DE 2012 LAS PALMAS DE GRAN CANARIA UNIVERSIDAD DE LAS PALMAS DE GRAN CANARIA SUBDIRECCiÓN DE TERCER CICLO Y POSTGRAoO 47/2011 -12 Reunido el dia de la fecha. el Tribunal nombrado por el Excmo. Sr. Rector Magfco. de esta Universidad, ella aspirante expuso esta TESIS DOCTORAL. Terminada la lectura y co ntestadas por ella Doctorando /a las objeciones formuladas por los sei'lores miembros del Tribunal, éste calificó dicho trabajo con la nota de Af' \'O c.u,"" ｌａ｜ｊｨｾ＠ Las Palmas de Gran Ca naria, a 27 de marzo de 2012 Ella PresIdente/a: Orla. Ola ., Santiago Hernández León, 1 Ella Vocal: Or la. Ola. Maria Eisa Vázquez Otero, . Aili" ') El la Secretariola: Or la. Ola. Angel Rodriguez Santa na, r I 1/ lL o Ｎ ｾ｟Ｎｊ＠ Ella Vocal: Or la. Ola. José Luis Pelegri L/opart, Ella Vocal: Or la. Ola. Eugenio Fraile Nu ez. A .t--.¿/ ｾ＠ ｾＺＺＺＺＭ El Doctorando: Da vi d Lindo Atlchati, Anexo I D/Dª......SALVADOR GALVÁN HERRERA....SECRETARIO/A DEL DEPARTAMENTO DE.............FÍSICA................. DE LA UNIVERSIDAD DE LAS PALMAS DE GRAN CANARIA, CERTIFICA, Que el Consejo de Doctores del Departamento en su sesión de fecha.............................tomó el acuerdo de dar el consentimiento para su tramitación, a la tesis doctoral titulada “LONG-TERM VARIABILITY AND EFFECTS ON LARVAL FISH DISTRIBUTION OF THE GULF OF MEXICO LOOP CURRENT AND RINGS” presentada por el/la doctorando/a D DAVID LINDO ATICHATI y dirigida por el Doctor D PABLO SANGRÀ INCIARTE. Y para que así conste, y a efectos de lo previsto en el Artº 73.2 del Reglamento de Estudios de Doctorado de esta Universidad, firmo la presente en Las Palmas de Gran Canaria, a..........................de.....................................de dos mil….doce.......... A Olivia o Quélén-Maurice… Acknowledgements This thesis was possible thanks to the funding provided by the Atlantic Oceanographic and Meteorological Laboratory (AOML) of the National Oceanic and Atmospheric Administration (NOAA) through the project Fish Stock Assessment From Satellite Observations. The grant awarded by “Obra Social La Caixa” gave me the opportunity to take PhD courses of the graduate program in Meteorology and Physical Oceanography at the University of Miami (RSMAS). First and foremost, I would like to express my deepest gratitude to my dissertation advisor, Dr. Pablo Sangrà, for giving me the opportunity to conduct this dissertation under his supervision, encouragement, outstanding scientific support, and friendship. I am grateful to him for hosting me during my visit at UCLA, and for offering me his temper and know-how. I would like to thank the co-advisor of mi dissertation Dr. Gustavo Goni, director of the physical oceanography division at NOAA/AOML, who has also been my supervisor there for the last years. He has taught me how to work with altimetry data and guided me during the meticulous process of becoming an author. The co-advisor of my dissertation Dr. Barbara Muhling has played a key role in my dissertation. I would like to extend to her my deepest gratitude for introducing me to the fascinating world of larval-fish with so much patience and accuracy. Thanks Barb for showing me how to be succinct and precise. I am getting there! I am very grateful to the Physics Department of the ULPGC, the administrative services of the Facultad de Ciencias del Mar, and the Servicio de Investigación y Tercer Ciclo for making all the PhD procedures easier. In particular, I would like to mention Dr. Alonso Hernández-Guerra for his wise advise when choosing my advisor, and María Dolores Morales Sosa for brilliantly managing my academic paperwork and defense proceedings. I sincerely thank Dr. Silvia Garzoli for inviting me to collaborate with and learn from the PHOD team at NOAA/AOML, and for always having her door open to guide me. I also acknowledge Dr. Arthur Mariano for being my supervisor at University of Miami (RSMAS). I warmly thank Dr. Nelson Melo-González (NOAA/AOML) for being always there “en la lucha” and for offering me the helpful poetry of José Martí. I greatly appreciate the effort of Dr. Francis Bringas and Jay Harris (NOAA/AOML) for helping me with the programming, and Dr. Francisco Machín for his support during the last months. I sincerely thank Dr. Francisco Alemany of the Instituto Español de Oceanografía (IEO) for the productive discussion we had. Dr. Frank Muller-Karger deserves a special mention for hosting me at the University of South Florida (USF), allowing me to teach graduate lectures, training me on how to write proposals, and above all for showing me that it is possible to be an excellent scientist, professor, and person. A big THANK YOU! to my fellows at NOAA/AOML, NOAA/SEFSC, and USF/IMARS Dr. Ryan Smith, Dr. Elisabeth Jones, Dr. John Lamkin, Sennai Habtes, and Dr. Mitch Roffer. Outside science I thank my sister Sonia and my mother for understanding me and supporting me wherever I am and have been. In all sincerity and conviction, I thank my father, for passing on me his courage, strength, and optimism. To Francisca, Tomeu, Abuela, Pedro, Cristina, and Uli, for sharing with me “the secret”. To Miquel Àngel for his unconditional friendship all life long. Last but not least, I would like to specially thank, My wife, Queta, my Majorcan little girl. Thanks for being by my side during our PhD journey. One among the many other trips we will make, from now on with our baby. Agradecimientos Esta tesis ha sido posible gracias a la financiación del Atlantic Oceanographic and Meteorological Laboratory (AOML) de la National Oceanic and Atmospheric Administration (NOAA) a través del proyecto Fish Stock Assessment From Satellite Observations. La beca de postgrado de la “Obra Social La Caixa” me dio la oportunidad de poder tomar las asignaturas del doctorado en Meteorología y Oceanografía Física de la University of Miami (RSMAS). En primer lugar, quisiera expresar mi más profunda gratitud a mi director de tesis, el Dr. Pablo Sangrà, por haberme dado la oportunidad de hacer la tesis bajo su supervisión, por la confianza en mi trabajo, su sobresaliente apoyo científico, y por su amistad. Por abrirme su casa en durante mi visita a UCLA, por brindarme su serenidad y su buen hacer. Mi agradecimiento más sincero al co-director de mi tesis el Dr. Gustavo Goni, director de la división de oceanografía de AOML, quien también ha sido mi supervisor durante los últimos años. Gracias a él he aprendido a trabajar con altimetría satelital y he sido guiado a través del minucioso proceso de publicar. La co-directora de mi tesis, la Dra. Barbara Muhling, ha sido y es una pieza clave para mí. Mi más profundo agradecimiento por descubrirme el fascinante mundo de las larvas, siempre con gran paciencia y precisión. Y por enseñarme a ser sucinto. Thanks heaps Barb! Mi agradecimiento al Departamento de Física, a los servicios administrativos de la Facultad de Ciencias del Mar, y al Servicio de Investigación y Tercer Ciclo por facilitar con tanta agilidad todos los trámites. En especial al Dr. Hernandez-Guerra por su sabia recomendación para la elección de mi tutor, y a María Dolores Morales Sosa por gestionar a la perfección mi documentación académica y la lectura de mi tesis. Tengo que agradecer enormemente a la Dra. Silvia Garzoli por invitarme a colaborar con el equipo de PHOD en Miami y por tener siempre sus puertas abiertas para orientarme. Al Dr. Arthur Mariano por ser mi supervisor en University of Miami (RSMAS). Por supuesto al Dr. Nelson Melo-González de NOAA/AOML, por estar siempre a mi lado en la lucha y por darme ese toque poético de José Martí que tanto me ha ayudado. Debo agradecer el esfuerzo al Dr. Francis Bringas y Jay Harris (NOAA/AOML) al ayudarme en la elaboración de programas y al Dr. Francisco Machín por su apoyo en los últimos meses. Al Dr. Francisco Alemany del Instituto Español de Oceanografía (IEO) un sincero agradecimiento por la fructífera discusión que tuvimos. Especial mención al Dr. Frank Muller-Karger por facilitar mi estancia en University of South Florida (USF), por permitirme impartir clases de doctorado, por enseñarme a escribir propuestas, y sobretodo por mostrarme que se puede ser un excelente científico, profesor, y persona. Un enorme ¡GRACIAS! a mis compañeros de NOAA/AOML, NOAA/SEFSC y USF/IMARS Dr. Ryan Smith, Dr. Elisabeth Jones, Jay Harris, Dr. John Lamkin, Sennai Habtes, y al Dr. Mitch Roffer. Fuera de la ciencia, a mi hermana Sonia y a mi madre por comprenderme y apoyarme allá donde estuviese. A mi padre por regalarme su valentía, fuerza y optimismo. A Francisca, Tomeu, s’Abuela, en Pedro, Cristina y n’Uli, per compartir es secret. A en Miquel Àngel por su incondicional amistad desde siempre. Y por último me gustaría agradecer especialmente, A mi mujer Queta, sa meva nina mallorquina, por querer acompañarme con una sonrisa en nuestro viaje de doctorado. Uno de tantos muchos, en los que ahora seremos tres. i Abstract The Gulf of Mexico (GOM) is an enclosed sea susceptible to climate change and anthropogenic pressures, as years of intense development and exploitation have resulted in significant change to the fisheries resources. It is the preferred spawning habitat for several commercially-important pelagic fish species (e.g. the Atlantic bluefin tuna). This basin is also characterized by a complex and highly variable system of currents and eddies, which affects the physical environment where fish spawn (Teo and Block 2010). The mesoscale circulation in the GOM is dominated by the Loop Current (LC) and rings shed by this major current. These powerful oceanic features carry anomalies in the physical, biological, and chemical properties of the region, and they affect -either directly or indirectly through their smaller-scale subsidiariesjust about every aspect of oceanography of the Gulf. For these reasons, the proposed investigation is focused on the satellite monitoring of the temporal and spatial variability of the LC and rings, and on the regulation of the larval fish distribution of some species by mesoscale oceanic features. The overarching aim of this dissertation is to investigate, assess and analyze linkages between ocean circulation and the spatial and temporal distribution of larval fish in the Gulf of Mexico. This connection is explored in here through the use of satellite derived observations and in situ biological sampling. The main goals of this research are to (1) monitor and describe the spatial and temporal variability of LC intrusions (northward), LC retreats (southward), and ring detachments; and (2) assess the influence of mesoscale ocean features on the distribution of larval fish spawned in the northern GOM in spring months from 1993 to 2007. The data that will be used in this study include, but are not limited to, sea surface temperatures (in situ and from satellites), sea surface height from satellite altimetry, and ocean surface color. The biological data correspond to in situ samples collected in the GOM north of 24°N during spring (April to June) NOAA/NMFS surveys between 1993 and 2007. Larval fish data are available from the National viii List of figures Figure 2.1. Intra-Americas Sea (IAS, orange rectangle) encompassing the Caribbean Sea (CS, red rectangle) and the Gulf of Mexico (GOM, yellow rectangle, and region of study). Bathymetric data was obtained from ETOPO1 (Amante et al. 2009), and contours in grey indicate isobaths from 500 to 8000 m with contours intervals of 1000 m. ............................................................................................. 11 Figure 2.2. Principal oceanographic features of the Intra-Americas Sea (IAS), encompassing the Gulf of Mexico and the Caribbean Sea, with schematic cartoons showing the Guiana Current (GC), North Brazil Current Ring (NBCR), Caribbean Current (CC), Yucatan Current (YC), Loop Current (LC), Loop Current Ring (LCR), Florida Current (FC), and Gulf Stream (GS). Bathymetric data was obtained from ETOPO1 (Amante et al. 2009), and contours in grey indicate isobaths from 500 to 8000 m with contours intervals of 500 m. ........... 13 Figure 2.3. Mean transports (Sv) through main passages in the Intra-Americas Sea from combined wind/MOC (Meridional Overturning Circulation) forced model simulation with observed transports (Johns et al. 2002; Smith 2010) are presented. Bathymetric data was obtained from ETOPO1 (Amante et al. 2009), and contours in grey indicate isobaths every 1000 m. ......................................... 16 Figure 2.4. Bottom topography in the Gulf of Mexico (GOM), data retrieved from ETOPO1 1-minute Global Relief (Amante et al. 2009) and contours in grey indicate isobaths from 500 to 8000 m with contours intervals of 200 m. The map illustrates the seven distinct geographical regions in which the GOM is divided: (a) GOM Basin, (b) Northeast GOM, (c) South Florida Continental Shelf and Slope, (d) Campeche Bank, (e) Bay of Campeche, (f) Eastern Mexico Continental Shelf and Slope, and (g) Northern GOM. ........................................ 19 Figure 2.5. Sea surface temperature (SST) on April 25 2011, showing the Loop Current extended into the Gulf, and a warm ring forming from the strangulation of the current. Note the appearance of cyclonic eddies along the edges of the Loop Current, in particular two cyclones located in the ‘bottle neck’ of the ix current (approximately centred at 23°N, 88°W and 24.5°N, 85.5°W). Data was obtained from the Advanced Very High Resolution Radiometer (AVHRR), available with a resolution of 2 days on an 18 km grid. Black arrows in the background represent satellite-derived geostrophic currents. ............................. 23 Figure 2.6. Domain of the Hurlburt and Thomson model superimposed on a map of the Gulf of Mexico. The deepest water is at 3000 m and the shallowest water is 400 m deep, with contour interval of 250 m. The locations of the inflow and outflow ports are also indicated. The model is driven by the inflow through the southern port (160 km wide, Yucatan Straits) and compensated by outflow though the eastern port (150 km wide, Florida Straits). The β-plane approximation and hence taking into account the variation of the local rotation rate is used (�=�� only), but not the earth sphericity. ..................................... 29 Figure 2.7. Diagram illustrating the “momentum imbalance paradox”. A northward meridional channel carrying water with density � empties into an otherwise stagnant ocean with density (�+∆�). Along the front it is assumed that the thickness h=0. The streamlines in the channel are assumed to remain parallel to the channel walls until the coastline is reached (i.e., section AB). The hypothetical steady configuration is not possible on both a � and �-plane. In the PN scenario, a steady inviscid outflow cannot exist because the long-shore momentum flux of the downstream current is not balanced. .............................. 31 Figure 2.8. Diagrams showing the (a) PN resolution of the “momentum imbalance paradox” and (b) the depth contours of the �-plane flow. Because of the imbalance paradox sketched in Figure 2.7, the flow of the downstream current is not balanced. As a result, there are two possibilities. On a �-plane (a), the anticyclonic ring growth is ultimately arrested and, as a result, rings are periodically shed on the left-hand side (looking offshore). Through the � effect, these eddies are forced to propagate to the left. Pichevin and Nof (1997) obtained their analytical solution by equating the momentum flux through EF to the momentum flux through CD. Nof (2005) added that we should not deal with the momentum flux through CD. Instead, the integrated Coriolis force associated with the slowly growing base ring -the one that is still no detached and whose center migrates slowly offshoreis equated to the momentum flux through CD. x The base ring, which is the eddy in contact with the source, should be distinguished from the already detached eddies downstream. The assumption that the eddies are “kissing” each other as they propagate westward is made in both PN and Nof 2005. (b) Conversey on a �-plane (b), the eddy grows forever and the downstream current mass flux is smaller than the incoming mass flux. ....... 33 Figure 2.9. Larval fish (Salmo salar, Atlantic salmon) egg hatching, and start of the external feeding. The Alevin (larva) has grown around the remains of the yolk sac. In about 24hrs it will be a fry without yolk sac and it will start external feeding from zooplankton. Image provided by Dr. Uwe Kils Institute of Marine and Coastal Sciences, Rutgers University. .......................................................... 41 Figure 2.10. Diagram of the recruitment process during early life stages of marine fish, showing the temporal evolution of abundances of eggs, yolk-sac larvae, larvae, and juveniles. Sources of nutrition (green), probable sources of mortality (blue), and hypothesized mechanisms of control (red) are included. Reproduced from Houde 1987. ................................................................................................ 45 Figure 3.1. Example of general circulation through the Gulf of Mexico from satellite altimetry fields, on May 17 1995. Fields illustrate the LC and the ring shed in terms of (a) sea height anomaly (SHA) and (c) sea surface height (SSH). SSH fields are obtained by adding (b) mean dynamic topography (MDT) to (a) SHA fields. ................................................................................................................... 56 Figure 3.2. Fields of altimetry-derived SSH gradient in the GOM with contours of SSH (in cm, black lines) superimposed, for (a) May 26, 1993, and (b) June 02 1993. Fields illustrate (a) the location of the LC front, and (b) the detection of a LC ring and the LC front. The northernmost and westernmost locations of the LC are tagged with a square in both panels. .............................................................. 57 Figure 3.3. Time series of (a) LC northward penetration (red line) and (b) LC westward penetration (green line), from November 1992 to December 2009. Black circles indicate the time of separation of LC rings. The signal was filtered using a Butterworth filter of order 6 and cutoff frequency 1/7 rad/s (black line). ............................................................................................................................. 59 xi Figure 3.4. (a) Monthly Mean location of the LC northward penetration (blue circles) and (b) annual mean location of the LC northward penetration. Black bars indicate one standard deviation over the 17 years of data analyzed, from January 1993 to December 2009. ...................................................................................... 60 Figure 3.5. Number of LC rings shed during each year, from January 1993 to December 2009 as obtained according to the methodology described in section 2. ............................................................................................................................. 63 Figure 3.6. Monthly sea height residuals (SHR), estimated as the difference between the mean SHA for each specific month and the mean SHA value for that same month since November 1992. The black line indicates a mean SHR increase of 2.78 cm per decade over the 17 years of data analyzed. ..................................... 64 Figure 4.1. General circulation in the Gulf of Mexico on October 22, 2010. Thin arrows represent geostrophic currents derived by combining satellite altimeter observations and numerical model results. Thick arrows contour anticyclonic (Loop Current/LC and rings/AR) and cyclonic features (rings/CR). .................. 71 Figure 4.2. Comparison between SST and SSH satellite derived fields. Fields on the left panels illustrate a ring separation event in terms of (a) SST, (b) SHA, (c) SSH, and (d) gradient of SSH on May 6, 1998. Fields on the right panels illustrate a ring already shed by the LC in terms of (e) SST, (f) SHA, (g) SSH, and (h) gradient of SSH on June 24, 1998. .......................................................... 74 Figure 4.3. Example of the five regions of circulation and boundaries defined using the mesoscale classification algorithm for June 24, 1998. The background field is the satellite derived SSH. The legend illustrates the mean SSH and standard deviation of SSH for each region of circulation. Colors in the legend correspond to the colors of the satellite derived SSH field. ................................................... 77 Figure 4.4. Station locations of the NOAA/SEFSC SEAMAP annual larvae survey and percentage of larval fish captured in (17%) and out (83%) of the LC ROI from 1993 to 2007. The background color is the satellite derived SSH for May 1, 1996. The black solid line represents the mean LC front. The dashed lines illustrates northernmost latitude, westernmost longitude and easternmost xii longitude of the LC front in the GOM north of 23°N, which were 26°N, 87°W, 84°W. ................................................................................................................... 82 Figure 4.5. Monthly mean location of the LC northward penetration. Bars indicate one standard deviation over the 16 years of data (January 1993 to December 2008). ................................................................................................................... 83 Figure 4.6. (a) Anomalies of the mean northernmost location of the LC in relation to spring larval concentration of larvae captured in a 2° box centered at 87°W and 26°N . Black line shows mean larval concentration of captures of T. thynnus, E. alleteratus, Auxis spp., Thunnus spp., and Coryphaena spp., from May 1993 to June 2007 with exception of 2003 and 2004. (b) Regression analysis of mean larval concentration vs. anomalies in the LC excursions. (c) Location examined represented with a black square. .......................................................................... 84 Figure 4.7. Mean larval concentration of captures (red circles) and proportion of captures (blue circles) for larvae of (a) T. thynnus, (b) E. alleteratus, (c) Auxis spp., (d) Thunnus spp., and (e) Coryphaena spp. in relation to satellite derived observations of sea surface height (SSH), from 1993 to 2007. SSH values were binned to 10 cm intervals. Error bars represent one standard error. Colored blocks represent SSH intervals of cyclonic regions (CR), cyclonic boundaries (CB), common waters (CW), anticyclonic boundaries (AB) and anticyclonic regions (AR), calculated from the mean and standard deviation of the SSH field for each of the 5 regions ...................................................................................... 87 Figure 5.1. Bottom topography in the Gulf of Mexico (GOM), data retrieved from ETOPO1 1-minute Global Relief (Amante et al. 2009). The map illustrates section A-B, used to estimate the difference in the mean sea level. .................... 96 Figure 5.2. Time series of monthly Loop Current migrations (thin red line) determined from altimetry derived fields, and monthly Sea Level Pressure Anomalies (thin blue line) determined from the pressure difference between two locations north and south of the Loop Current, from November 1992 to December 2009. Vertical black lines indicate the time of separation of LC rings. The signals were filtered using a Butterworth filter of order 6 and cutoff xiii frequency 1/7 rad/s (thick red line and thick blue line for LC migrations and pressure anomalies respectively). ........................................................................ 98 Figure 5.3. (a) Time series of monthly Loop Current migrations determined from altimetry derived fields from November 1992 to December 2009. (b) Real part of the complex CWT of dataset in (a). Bright red indicates large positive amplitude and dark blue indicates large negative amplitude. (c) Monthly Sea Level Pressure Anomalies determined from the pressure difference between two locations north and south of the Loop Current, from November 1992 to December 2009. (d) Real part of the complex CWT of dataset in (c). Bright red indicates large positive amplitude and dark blue indicates large negative amplitude. (e) Semblance S (calculated from Eq. (20), with n=1). Bright red corresponds to a semblance of +1, green to a semblance of zero, and dark blue to a semblance of −1.Wavelength ranges from zero to 48 units, which in time scale corresponds from zero to 4 years (48 months). ............................................................................................... 99 Figure 5.4. Forecasted currents in the Gulf of Mexico for February 10, 2012. Thin arrows and color scale represent current direction and magnitude, and are obtained from the operational 1/32° global NLOM (Navy Layered Ocean Model). The model presently assimilates SSH from 3 satellite altimeters (ENVISAT, JASON-1 interleaved and JASON-2) and SST from satellite infrared imagery. Three cyclonic eddies are clearly revealed, one on the eastern boundary of the Loop Current and two on the western boundary of the Loop Current. ... 101 Figura 7.1. Mar Intra-Americano (rectángulo naranja) abarcando el Mar Caribe (rectángulo rojo) y el Golfo de México (rectángulo amarillo y región de estudio aquí). Los datos batimétricos fueron obtenidos de ETOPO1 (Amante et al. 2009), y los contornos en gris indican isóbatas desde 500 a 8000 m con intervalos de contorno de 1000 m. .......................................................................................... 120 Figura 7.2. Principales estructuras oceanográficas del Mar Intra-Americano, abarcando el Golfo de México y el Mar Caribe, con dibujos esquemáticos de la Corriente de Guyana (GC), el anillo de la Corriente Norte de Brasil (NBCR), la Corriente del Caribe (CC), la Corriente de Yucatán (YC), la Corriente de Lazo (LC en la figura), el anillo de la Corriente de Lazo (LCR), la Corriente de la Florida (FC), y la Corriente del Golfo (GS). Los datos batimétricos fueron xiv obtenidos de ETOPO1 (Amante et al. 2009), y los contornos en gris indican isóbatas desde 500 a 8000 m con intervalos de contorno de 500 m. ................. 121 Figura 7.3. Transportes medios (Sv) a través de los principales pasajes en el Mar Intra-Americano a partir de combinar simulaciones numéricas con forzamientos de vientos con observaciones de transportes (Johns et al. 2002; Smith 2010). Los datos batimétricos fueron obtenidos de ETOPO1 (Amante et al. 2009), y los contornos en gris indican isóbatas cada 1000 m. ............................................... 122 Figura 7.4. Topografía del fondo del Golfo de México (GDM), datos batimétricos obtenidos de ETOPO1 (Amante et al. 2009) y los contornos en gris indican isóbatas desde 500 a 8000 m con intervalos de contorno de 1000 m.. El mapa ilustra las siete regiones geográficas en las que se divide el GDM: (a) Cuenca del GDM, (b) Noreste del GDM, (c) Plataforma continental del Sur de la Florida, (d) Banco de Campeche, (e) Bahía de Campeche, (f) Plataforma continental del este del GDM, y (g) Norte del GDM. ....................................................................... 124 Figure 7.5. Temperatura superficial del mar (SST) el 25 de abril de 2011, mostrando la Corriente de Lazo (CL) extendida en el Golfo, y un caliente anillo anticiclónico formándose a partir de la estrangulación de la corriente. Nótese la aparición de remolinos ciclónicos a lo largo de los márgenes exteriores de la CL, en particular dos ciclones localizados en el “cuello de botella” de la corriente (aproximadamente centrados en 23°N, 88°W y 24.5°N, 85.5°W). Los datos fueron obtenidos del Advanced Very High Resolution Radiometer (AVHRR), disponible con una resolución de 2 días en una malla de 18 km. Las flechas negras en el fondo de la imagen representan corrientes geostróficas derivadas de datos satelitales. ................................................................................................. 125 Figura 7.6. Diagrama que ilustra el “momentum imbalance paradox”. Un canal abierto hacia el norte por el que fluye agua de densidad � se vacía en un océano estanco con densidad (�+∆�). A lo largo del frente se asume que el espesor es h=0. Se asume que las líneas de corriente en el canal permanecen paralelas a las paredes del canal hasta que se alcanza la línea horizontal de la desembocadura (i.e., sección AB). La hipotética configuración de estado estacionario no es posible en plano-� ni el plano-�. En el escenario de PN un flujo continuo no xv puede existir debido a que el momento del flujo de corriente no está en equilibrio. ........................................................................................................... 129 Figura 7.7. Diagramas que muestran (a) la resolución de PN del “momentum imbalance paradox” y (b) los contornos flujo del plano-�. A causa de la paradoja esbozada en la Figura 7.6, el flujo de la corriente entrante no está en equilibrio. Como resultado, hay dos posibilidades. En un plano-� (a), el crecimiento del anillo anticiclónico es finalmente detenido y, consecuentemente, se desprenden anillos periódicamente hacia la izquierda. A través del efecto �, estos remolinos se ven forzados a propagarse hacia la izquierda. Pichevin y Nof (1997) obtienen su solución analítica mediante la identificación del momento del flujo a través de EF con el momento del flujo a través de CD. (b) Por el contrario, en un plano-�), el remolino crece eternamente y el flujo de masa de corriente que sale es menor que el flujo de masa de corriente entrante. ........................................................ 130 Figura 7.8. Diagrama del proceso de reclutamiento durante los primeros estadios de vida de los peces marinos, mostrando la evolución temporal de las abundancias de huevos, larvas con saco vitelino, larvas, y juveniles. Se incluyen las fuentes de nutrición (verde), las fuentes probables de mortalidad (azul), y el hipotético mecanismo de control (rojo). Figura reproducida de Houde 1987. ................... 136 Figura 7.9. Comparación entre campos satelitales de SST y SSH. Los campos en los paneles de la izquierda ilustran eventos de separación de anillos en términos de (a) SST, (b) SHA, (c) SSH, y (d) gradiente de SSH a 6 de mayo de 1998. Los campos en los paneles de la derecha ilustran un anillo desprendido de la Corriente de Lazo en términos de (e) SST, (f) SHA, (g) SSH, y (h) gradiente de SSH a 24 de junio de 1998. ............................................................................... 138 Figura 7.10. Campos de gradiente de SSH derivados de altimetría en el GDM con contornos de SSH superpuestos (en cm, líneas negras), para (a) 26 de mayo de 1993, y (b) 02 de junio de 1993. Los campos ilustran (a) la posición del frente de la CL, y (b) la detección del anillo de la CL y el frente de la CL. Las posiciones más al norte y más al oeste están marcadas con un cuadrado en ambos paneles. ........................................................................................................................... 140 xvi Figura 7.11. Ejemplo de las cinco regiones de circulación y sus fronteras definidas usando un algoritmo de clasificación de las estructuras de mesoescala para el 24 de junio de 1998. El campo de fondo es la SSH satelital. La leyenda ilustra la SSH media y la desviación estándar de la SSH para cada región de circulación. Los colores de la leyenda se corresponden aproximadamente con los colores del campo de SSH para cada región identificada. ................................................... 143 Figure 7.12. Series temporales de (a) la penetración hacia el norte de la Corriente de Lazo (línea roja) y (b) la penetración hacia el oeste de la Corriente de Lazo (línea verde), desde noviembre 1992 hasta diciembre 2009. Los círculos negros indican el momento de separación de los anillos de la CL. La señal fue filtrada con un filtro Butterworth de orden 6 y frecuencia de corte 1/7 rad/s (línea negra). ..... 147 Figura 7.13 (a) Ubicación media mensual de la posición más al norte de la CL (círculos azules) y (b) posición media anual de la posición más al norte de la CL (círculos azules). Las barras negras ndican una desviación estándar sobre 17 años de datos analizados, desde enero de 1993 hasta diciembre de 2009. ................ 148 Figura 7.14. Número de anillos desprendidos durante cada año, de enero de 1993 a diciembre de 2009, obtenidos de acorde a la metodología descrita en el apartado 4.3 de esta sección. ............................................................................................ 151 Figura 7.15. Residuos mensuales de la anomalía de la altura superficial del mar (SHR), estimados como la diferencia entre la SHA media mensual y la SHA media para ese determinado mes desde Noviembre de 1992. La línea negra indica la línea de tendencia, que muestra un incremento de 2.78 cm por década. ........................................................................................................................... 151 Figure 7.16. Localización de las estaciones de las campañas anuales del programa NOAA/SEFSC SEAMAP y porcentaje de larvas de peces capturados dentro (17%) y fuera (83%) de la región de influencia de la Corriente de Lazo de 1993 a 2007. El color de fondo es la altura superficial del mar para 1 de mayo de 1996. Los líneas sólidas negras representan el frente medio de la CL. Las líneas discontinuas ilustran la latitud más al norte, la longitud más al oeste y la longitud más al este del frente de la CL en el norte del Golfo de México al norte de 23°N, que son respectivamente 26°N, 87°W, 84°W. ................................................... 153 xvii Figura 7.17. (a) Anomalías de latitud media del la Corriente de Lazo en relación con concentraciones de larvas capturadas en una región simbolizada por una caja de 2° centrada en 87°W y 26°N. La línea negra muestra la concentración media de las capturas de T. thynnus, E. alleteratus, Auxis spp., Thunnus spp., y Coryphaena spp., de mayo 1993 a junio 2007 con excepción de 2003 y 2004. (b) Análisis de regresión de la concentración media de larvas vs. las anomalías anomalías en las intrusiones de la CL. (c) La región examinada, simbolizada por un cuadrado negro. ............................................................................................ 154 Figura 7.18. Concentraciones medias de capturas (círculos rojos) y proporción de capturas (círculos azules) (a) T. thynnus, (b) E. alleteratus, (c) Auxis spp., (d) Thunnus spp., y (e) Coryphaena spp en relación con observaciones satelitales de SSH, de 1993 a 2007. Los rectángulos coloreados representan los intervalos de SSH de regiones ciclónicas (CR), frentes ciclónicos (CB), aguas comunes (CW), frentes anticiclónicos (AB) y regiones anticiclónicas (AR), calculadas a partir de la media y desviación estándar del campo de SSH para cada una de las 5 regiones. ............................................................................................................. 157 Figura 7.19. Series temporales de las migraciones mensuales de la Corriente de Lazo (línea roja fina) determinadas mediante campos de altimetría, y anomalías mensuales de la presión de la superficie del mar (línea azul delgada) determinadas a partir de ubicaciones al norte y sur de la Corriente de Lazo, de noviembre de 1992 a diciembre 2009. Las líneas verticales negras indican el momento de separaciones de anillos de la Corriente de Lazo. Las señales fueron filtradas utilizando un filtro Butterworth de orden 6 y frecuencia de corte 1/7 rad/s (línea roja gruesa y línea azul gruesa para migraciones de la Corriente de Lazo y anomalías de presión respectivamente). ................................................ 159 2. Evidence of coupling between atmospheric pressure/wind field and LC migrations ............95 2.1 Introduction ......................................................................................................................95 2.2 Atmospheric pressure variability and LC migrations .......................................................96 3. Westward propagating long-lived eddies generated by the LC ...........................................101 3.1 Introduction ....................................................................................................................101 3.1 Westward propagating long-lived eddies generated by the LC ......................................102 4. Mesoscale features affecting primary productivity in the GOM ..........................................105 Chapter VI: General discussion and conclusions .................................................................. 107 1. Synthesis of results and general discussion ..........................................................................109 2. Conclusions ..........................................................................................................................110 Chapter VII: Resumen en español (Spanish summary) ........................................................ 114 1. Introducción .........................................................................................................................116 2. Objetivos de la investigación ...............................................................................................117 3. Oceanografía física y ecosistemas marinos en el Golfo de México .....................................119 3.1 Introducción ....................................................................................................................119 3.2 Circulación en el Mar Intra-Americano ..........................................................................121 3.3 Oceanografía del Golfo de México ................................................................................123 3.3.1 Topografía ...............................................................................................................124 3.3.2 Circulación general .................................................................................................125 3.3.2 ¿Hay un único mecanismo para las intrusiones de la Corriente de Lazo y los desprendimientos de anillos? ...........................................................................................128 3.3.3 Consideraciones aguas arriba ..................................................................................131 3.4 Biología marina y pescaderías en el Golfo de México ...................................................132 3.4.1 Niveles biológicos y factores limitantes en el Golfo de México ............................133 3.4.2 Comunidades de larvas de peces en el Golfo de México ........................................135 4. Planteamiento y metodología ...............................................................................................137 4.1 Localización del frente del la Corriente de Lazo ............................................................139 4.2 Detección del desprendimiento del anillo y periodo de vida ..........................................141 4.3 Identificación de estructuras oceanográficas locales ......................................................141 4.4 Muestreo de larvas y análisis de datos ............................................................................143 4.5 Análisis de probabilidades ..............................................................................................145 4.6 Análisis multivariado ......................................................................................................146 5. Síntesis de resultados ...........................................................................................................147 5.1 Las excursiones hacia el norte y hacia el oeste de la Corriente de Lazo ........................147 5.2 Eventos de desprendimientos de anillos .........................................................................149 5.3 Residuos de la anomalía de la altura superficial del mar ...............................................151 5.4 La Corriente de Lazo y las distribuciones de larvas .......................................................152 5.5 Efectos de las estructuras de mesoescala en las distribuciones de larvas .......................155 6. Futuras líneas de investigación ............................................................................................158 6.1 Evidencias de un acoplamiento ente el campo de presiones/viento y las migraciones de la Corriente de Lazo .........................................................................................................158 6.2 Remolinos generados por la Corriente de Lazo que se propagan hacia el oeste ............160 7. Discusión ..............................................................................................................................162 8. Conclusiones ........................................................................................................................164 Nomenclature ............................................................................................................................ 168 References .................................................................................................................................. 171 Chapter I: Introduction 4 Chapter I 1. Introduction The overarching aim of this study is to investigate the link between early life stages of fishes and the variability of the mesoscale oceanic structures in the Gulf of Mexico (GOM). This connection is explored in here through the use of satellite-derived observations and in situ biological sampling. There are many examples in the literature of correlations between changes in physical oceanographic parameters and changes in fish resources (Ortner et al. 1984; Mann 1993; Bakun 2006). Commonly, these correlations hold for a few years, then break down. Although this does not necessarily mean that the correlation was invalid, it calls for a better understanding of the mechanism of interaction in the complex system between the physics and the biology before we can understand the time dependence of these correlations. The GOM is a semi-enclosed sea bordered by the United States, Mexico and Cuba. The economic livelihoods of this international community depend on the many goods and services provided by the Gulf of Mexico; fisheries being one of the most important services. Considered to be a jewel among natural resources of the western hemisphere, the GOM wetlands are the source of an important seafood and shellfish industry, and its offshore waters support rich commercial and recreational fisheries, being the site for spawning and habitat of commercially relevant pelagic and benthic fish species (Shipp 1999; Rabalais et al. 1999). However, the GOM is susceptible to climate change and anthropogenic pressures, as years of intense development and exploitation have resulted in significant change to the fisheries resources. Major problems facing the Gulf, including habitat modification, pollution, and overexploitation, are generating uncertain impacts on this large marine ecosystem. Intensive fishing is the primary force driving biomass changes in the GOM, with climatic variability the secondary driving force (Sherman 2003). These two main pressures on the ecosystems may be addressed in part by having a synoptic comprehension of the physical-biological interactions in the GOM. The region is characterized by a complex and highly variable circulation, in time and space, with an intense mesoscale activity that is dominated by two main features: the Loop Current (LC) and the rings shed by the LC. These powerful oceanic features carry anomalies in the physical, biological, and chemical properties of the region, and they affect -either directly or indirectly through their smaller-scale subsidiariesjust about every aspect of oceanography of the Gulf. For these reasons, the proposed investigation is focused on the Introduction 5 satellite monitoring of the temporal and spatial variability of the LC and rings, and on the regulation of the larval fish distribution of some species by mesoscale oceanic features. Several studies show that physical and biological conditions for both larvae and adult fish in the GOM exhibit high spatial and temporal variability (Müller-Karger et al. 1991), which are likely linked to the patterns of spawning (Ortner et al. 1984; Bakun 2006). Variability in larval abundances of some pelagic species has been linked in more recent studies to environmental parameters -such as water temperature, salinity, water depth, and day lengthand plankton distribution in the GOM region (Muhling et al. 2010; Richardson et al. 2010). Since some ecosystems are highly affected by changes in environmental conditions and plankton distribution (Teo and Block 2010), variability of ocean properties and mesoscale structures in the GOM is expected to have a direct effect on ecosystems in this region. In particular, recent studies show that a detailed knowledge of the temporal and spatial variability of mesoscale structures in the eastern GOM is fundamental for understanding the environmental conditions that influence distributions of the larvae of different fish species, spawning sites, larval growth and subsequent variability in larval and juvenile survival (Richards et al. 1993; Bakun 2006). However, there seems to be a lack of information in how the ecosystems respond to short and long-term variability patterns in the region, which could help in dealing with the climatic change pressures in the GOM.! All the previous aspects considered, the main goal of this dissertation is to investigate, assess and analyze the link between fisheries resources and the temporal and spatial variability in ocean properties and mesoscale activity in the GOM. On this regard, this work will provide a better description of the mesoscale dynamics in the LC and associated ring field, and will shed some light on the physical regulation of the distribution of early life stages of some fish species in the GOM. Research tools and results obtained in this work will be useful to make assessments and connectivity studies using similar methodologies and tools upstream in the Atlantic Ocean. Essentially, then main 3 sections of this work are the following: • The first section (Chapter II) describes the main characteristics of the Physical Oceanography and Ecosystems in the Gulf of Mexico. Firstly, the general and mesoscale circulation of the deep and coastal GOM is described, as well as the interconnectivity of the flow processes in the tropical Atlantic Ocean and Caribbean Sea. Then, the physical mechanisms driving the dynamics of the GOM are 6 Chapter I investigated, described, and evaluated. On the other hand, the state of the art on the biological levels, primary production and nutrients in the GOM is reviewed. The population dynamics of pelagic larval fishes with relevant commercial importance in the GOM are presented, with emphasis on the dynamics of their early life stages. Then, the current comprehension on the environmental factors affecting ichthyoplankton distribution patterns and environmental conditions favorable for larval fish survival is investigated. Finally, a detailed description of the satellite derived data sources its relevance for biological end ecosystem studies is described. • The second section (Chapter III and IV) is the longest in extension and make up the core of this work. Chapter III provides an extensive description of the long term variability of the Loop Current, associated ring field, and sea height anomaly in the GOM, and proposes a novel methodology to identify regions of circulation based on fields of sea surface height derived from satellite altimetry. On Chapter IV, spatial and temporal distributions of larval fishes are related to environmental conditions and mesoscale oceanographic structures. Using satellite altimetry and in situ larval fish densities, the novel methodology used in Chapter III to identify regions of different circulation is used in Chapter IV to estimate links between mesoscale features and larval distributions of some fish taxa in the GOM, during spring from 1993 to 2007. As a result, some light is shed on the spatial and temporal distributions of larval fishes and its correlations to mesoscale oceanographic structures • Finally, the third section encloses a work that is in preparation that explores the effect of the atmospheric coupling on the migrations of the Loop Current, and the potential role that westward propagating eddies play on the Loop Current migrations. The innovative character of this latter work is reflected in using the footprint that cyclonic and anti-cyclonic eddies leave in sea surface height, temperature, and chlorophyll a, basically by applying Okubo-Weiss parameter. The discussion and conclusions obtained from this work will allow to make better assessments of the effect of longterm ocean changes and trends in the fisheries of the Gulf (Chapter VI), and stress the need for a sustained observing system able to resolve mesoscale features, which are important for the upper ocean temperature response to the changing climate, and continuous sampling of larvae of economically important species. Chapter II: Physical oceanography and ecosystems in the Gulf of Mexico 10 Chapter II 1. Introduction Although Europeans credited Christopher Columbus with the discovery of the Americas, the ships in his four voyages never reached the Gulf of Mexico (GOM). Instead, Columbus sailed into the Caribbean around Cuba and Hispaniola. The first European exploration of the GOM was by Amerigo Vespucci in 1497. He followed the coastal landmass of Central America before returning to the Atlantic Ocean via the Straits of Florida between Florida and Cuba. Vespucci described the voyage across the GOM in his letters, and it helped to produce the oldest known European cartographic representation of the New World. The first description of the ocean circulation in the GOM date from 1578, when William Bourne described a system in which the principal westward motion at the southern end of Africa merged with that of the central Atlantic, and some movement of water had to be diverted northward along the east coast of South America, into the Gulf of Mexico, and then eastward with the current between Florida and Cuba back to Europe (Peterson et al. 1996). Similarly to the first descriptions of the ocean circulation in the GOM, the study of the distribution of fishes in the region also started in the exploration of the Americas, when captain Bernard Romans developed a chart of East and West Florida native fishes (Romans and Concise 1775). Today, the Gulf of Mexico is considered as one of the most productive marine ecosystems in the world, and an important global reservoir of biodiversity. Knowledge of marine biodiversity in the Gulf of Mexico varies by region and taxon. On a Gulf-wide basis and considering the whole marine life, 3,302 species are censed in the GOM (Galtsoff 1954), from which 986 are listed to be species of fish distributed in 197 families (Froese and Pauly, 2011). This biodiversity is supported on the physical oceanographic processes that modulate the physiology, metabolism, behavior, and early growth rates of marine life (Blaxter 1991), and therefore it is crucial to know the physical oceanographic processes that drive the region in order to explore the physical regulation of fisheries resources in the GOM. The study area of this work is the enclosed region defined by 20°N to 31°N, and 80°W to 97°W. However, the GOM is a geographic region that together with the Caribbean Sea (CS), the Straits of Florida, and the adjacent western North Atlantic is comprised in a wider geographic region called the American Mediterranean Sea (Sverdrup et al. 1942) or Intra-Americas Sea (IAS) (Figure 2.1). A better understanding of the physical processes of the waters that flow in and through the IAS, especially its circulation variability, is important Physical oceanography and ecosystems in the Gulf of Mexico 11 because (1) it plays an important role as a conduit for mass, heat, salt and other tracers in the Atlantic circulation system (Schmitz and Richardson 1991); (2) the IAS contains the second largest body of very warm (≥ 28.5°) water on Earth: the western hemisphere warm pool (Wang and Enfield 2001), which is a significant heat source for the atmosphere; and (3) the waters of the IAS also support rich commercial fisheries and a major oil and gas industry (Tang et al. 2006). Even though the focus of this study is the GOM, recent studies have indicated that the GOM and the CS are dynamically inter-dependent; therefore certain aspects of the circulation in the Caribbean will also be discussed. Figure 2.1. Intra-Americas Sea (IAS, orange rectangle) encompassing the Caribbean Sea (CS, red rectangle) and the Gulf of Mexico (GOM, yellow rectangle, and region of study). Bathymetric data was obtained from ETOPO1 (Amante et al. 2009), and contours in grey indicate isobaths from 500 to 8000 m with contours intervals of 1000 m. Extensive physical oceanographic field and modeling studies have examined the GOM during the last 50 years (Amante et al. 2009). The main attractions have been the warm Loop Current and the large anticyclonic rings repeatedly shed by this current (Vukovich et al. 1979; Vukovich and Maul 1985; Yukovich 1986; Maul and Vukovich 1993; Elliott 1982; Lewis and Kirwan Jr 1987; Forristall et al. 1992; Sturges and Leben 2000; Nof 2005; ZavalaHidalgo et al. 2006; Lugo-Fernández 2007; Alvera et al. 2009; Lugo-Fernández and Leben 12 Chapter II 2010). However, compared to the knowledge that has accumulated about physical processes in the GOM, very little is known about the biological oceanography of the region, and most biological oceanographic studies in the GOM have been geographically restricted (e.g., Ortner et al. 1984; Biggs et al. 1984, 1988). Despite the importance of the region to fisheries, seasonal variability in larval fish assemblages in the GOM has been examined in relatively few studies, mainly focused on estuarine assemblages (Raynie and Shaw 1994; Tolan et al. 1997) or on relatively short-term interactions between assemblages and specific oceanographic features, such as the Mississippi River plume (Sogard et al. 1987; Govoni et al. 1989) or the Loop Current (Richards et al. 1993). Other studies have used ichthyoplankton survey data from the National Marine Fisheries Service's (NMFS's) gulf-wide Southeast Monitoring and Assessment Program (SEAMAP), but these studies are typically focused on a single species (Scott et al. 1993; Lyczkowski-Shultz and Ingram 2003; Lyczkowski-Shultz et al. 2007). Although these latter studies provided information on multiple species, no analyses were presented on the influence of mesoscale ocean features, whose associated energy tend to be higher than that of the main oceanic flow (Wunsch 1981), on the distribution patterns of fish larvae spawned in the GOM. In the next three sections, a more detailed description of the circulation in the IAS and the GOM is presented. First, the circulation in the IAS is studied since it is critical for the understanding the GOM. Then the oceanography of the GOM is extensively presented, from its topography to the possible mechanisms to explain the Loop Current and ring field variability. The importance of mesoscale activity is stated and its variability is analyzed in order to conjecture the ocean processes that could influence some larval fish assemblages in the GOM. This gives us room to explore the grounds of marine biology and fisheries of the region, from the lowest biological levels to larval fish assemblages. Finally, the state of the art on satellite remote sensing is exposed and also its importance as a suitable tool to couple the physics and the biology of the oceans. Physical oceanography and ecosystems in the Gulf of Mexico 19 years before present). Bounded on the north by the United States, the south by Mexico and the east by Cuba, the surface area of the Gulf is 564,200 km2 with a maximum east-west dimension of 1,575 Km and a maximum north-south dimension of about 900 Km. The bottom topography of the GOM is illustrated in Figure 2.4. A narrow to moderately wide continental shelf (0-200 m depth) located along the boundaries of the Gulf constitutes about 35% of its area. Almost half of the basin is shallow continental shelf water, and about one-fourth of the Gulf is very deep (3,000 m). The deepest area, the Sigsbee Deep, is approximately 3,850 m deep. Figure 2.4. Bottom topography in the Gulf of Mexico (GOM), data retrieved from ETOPO1 1-minute Global Relief (Amante et al. 2009) and contours in grey indicate isobaths from 500 to 8000 m with contours intervals of 200 m. The map illustrates the seven distinct geographical regions in which the GOM is divided: (a) GOM Basin, (b) Northeast GOM, (c) South Florida Continental Shelf and Slope, (d) Campeche Bank, (e) Bay of Campeche, (f) Eastern Mexico Continental Shelf and Slope, and (g) Northern GOM. The bathymetry has strong influence on the circulation in the GOM, as the Gulf's entrances are constricted by two wide and shallow continental shelves (Campache Bank, located North of Yucatan Peninsula, and the West Florida Shelf), with steep escarpments on 20 Chapter II their seaward edge. The GOM is usually divided into seven distinct geographical provinces (Antoine 1972): a) Gulf of Mexico Basin This portion of the Gulf of Mexico contains the Sigsbee Deep and can be further divided into the continental rise, the Sigsbee Abyssal Plain, and the Mississippi Cone. The Sigsbee Abyssal Plain is a deep, flat portion of the Gulf bottom located northwest of Campeche Bank. In this relatively uniform area of the Gulf bottom, the Sigsbee Knolls and other small salt domes represent the only major topographical features. The DeSoto Canyon borders the Mississippi Cone to the east, and the Mississippi Trough bounds it to the west (Ewing et al. 1958). b) Northeast Gulf of Mexico This region of the Gulf bottom extends from just east of the Mississippi Delta near Biloxi to the eastern side of Apalachee Bay. The Florida Escarpment separates the Florida Platform from the Gulf Basin and also forms the southeastern side of the DeSoto Canyon. Some theories suggest that the canyon is the result of erosion caused by oceanic currents, possibly the Loop Current (Nowlin 1971). c) West Florida Continental Shelf and Slope A submerged portion of the larger emergent Florida Peninsula, this region of the Gulf of Mexico extends along the coast from Apalachee Bay to the Straits of Florida and includes the Florida Keys and Dry Tortugas. The pressure imposed upon the West Florida Shelf by the Loop Current is found to give rise to a southward flowing jet along the shelf edge (Hetland et al. 1999). Evidence suggests that this basin was at one time enclosed by a barrier reef system (Antoine and Ewing 1963; Ewing et al. 1966; Sheridan et al. 1966). In the Straits of Florida the Jordan Knoll appears to be composed of remnants from this ancient reef system. Evidence suggests that this reef may have once extended across the straits, adjoining the Florida reefs with those of northern Cuba. Physical oceanography and ecosystems in the Gulf of Mexico 21 d) Campeche Bank Campeche Bank is located to the north of the Yucatan Peninsula (Ordonez 1936). The bank extends from the Yucatan Straits in the east to the Tabasco-Campeche Basin in the west and includes Arrecife Alacran. The region shows many similarities to the south Florida platform and some evidence suggests that the two ancient reef systems may have been continuous (Antoine and Ewing 1963). e) Bay of Campeche The Bay of Campeche is an isthmian embayment extending from the western edge of Campeche Bank to the offshore regions just east of Veracruz (~96° W). The Sierra Madre Oriental forms the southwestern border, and the associated coastal plain is similar to the Texas-Louisiana coast in the northern Gulf. The bottom topography is characterized by long ridges parallel to the exterior of the basin. Cyclonic eddies form near the western edge of the Loop Current, at the northeast shelf break of the Campeche Bank (ZavalaHidalgo et al. 2003a). Similar to the northern Gulf, large quantities of oil are produced here. f) Eastern Mexico Continental Shelf and Slope Located between Veracruz to the south and the Rio Grande to the north, this geological province spans the entire eastern shore of Mexico. The relative complexity of the bottom structure increases from south to north (Bryant et al. 1968). g) Northern Gulf of Mexico The northern Gulf of Mexico extends from Alabama to the U.S.-Mexico border. North to south, the province extends from 200 miles inland of the present day shoreline to the Sigsbee escarpment. Emergent topographic features on the continental slope are the Flower Garden Banks off the Texas/Louisiana coast, and the pinnacles region offshore of the Mississippi/Alabama coast (Halbouty 1979). 22 Chapter II 3.2 General circulation The most prominent and energetic components of the general circulation of the GOM are the Loop Current (LC), mesoscale eddies, and rings that are shed from the current (Figure 2.5). These powerful oceanic features affect, either directly or indirectly through their smallerscale subsidiaries, just about every aspect of oceanography of the Gulf. The LC in the Gulf is part of the Gulf Stream System, the very energetic western boundary current regime in the North Atlantic Ocean. The current is highly variable in position and strength with time, and can intrude northward into the northeastern GOM, forming an intense clockwise flow even as far northward as the Mississippi river delta or the Florida continental shelf (Huh et al. 1981; Wiseman and Dinnel 1988). The LC can retreat to have an almost direct path to the Straits of Florida – which is usually called port-to-port configuration - after shedding a LC ring. The LC ring is an anti-cyclonic warm-core ring, which is formed and pinched off from the LC, and then propagates westward (Elliott 1982). Water exits the IAS through the Straits of Florida as the FC between Florida and the Bahamas. Physical oceanography and ecosystems in the Gulf of Mexico 23 Figure 2.5. Sea surface temperature (SST) on April 25 2011, showing the Loop Current extended into the Gulf, and a warm ring forming from the strangulation of the current. Note the appearance of cyclonic eddies along the edges of the Loop Current, in particular two cyclones located in the ‘bottle neck’ of the current (approximately centred at 23°N, 88°W and 24.5°N, 85.5°W). Data was obtained from the Advanced Very High Resolution Radiometer (AVHRR), available with a resolution of 2 days on an 18 km grid. Black arrows in the background represent satellite-derived geostrophic currents. Anticyclonic rings are shed from the LC in the eastern Gulf at irregular intervals (Vukovich 1988a; Sturges and Leben 2000; Leben 2005), which has been reported to range from 4 to 18 months (Sturges and Leben 2000). At this point it must be spelled out the difference between mesoscale eddies and rings. Although it is usually found in the literature that LC rings are designated indistinctively as rings or eddies, it must be clear here that eddies and rings are quite different structures. Shortly after formation they are rings, with little or no relative vorticity in its central core, while later on the core has gained enough angular velocity (through radial diffusion from the ring) to properly talk of an eddy (Auladell et al. 2010). For example, the LC ring represents a wrapped-up piece of the major current that describes a loop, and that is why during its initial stages –before that diffusion takes placethe ring is composed by a frontal region that retains the intense current of the loop on its edge, and a central core region where waters are roughly stagnant. On the other hand, whereas mesoscale eddies show radii comparable to the Rossby radius of deformation (~40 24 Chapter II km, (Chelton et al. 1998)), the radius of the ring is much larger (~150 km, (Oey et al. 2005)). A simple diffusion-advection model with an effective diffusion coefficient (Sangrà et al. 2007) is appropriate to grossly simulate the temporal evolution of angular velocity of eddies that start in solid-body type rotation (Auladell et al. 2010). The diffusion-advection equation to study the angular velocity evolution, which is derived from cross-differentiation of the momentum isopycnic equations in cylindrical coordinates under the �-plane and axialsymmetry approximations, is the following: !" !" +�! !" !" =! ! !!(!"#) !!! (1) where � is angular velocity, �! is radial velocity, � is an effective horizontal diffusion coefficient, � is time, and � radial position. On the basis of all the explained above and the nature of its generation, rotation, robustness, and long-lived features (Fuglister 1971; Olson 1991), LC Anticyclonic rings will be named rings hereafter along this dissertation. Here it suffices to define rings as intense eddies or vortices that represent a wrapped-up piece of a major ocean current. These large rings shed by the LC have radii of approximately 150 km, swirl speeds of 1.8-2 m s-1, around 800 m depth (similarly to the LC), translate westward into the GOM over space scales of hundreds to thousands of kilometers and exist for periods lasting from months to years (Oey et al. 2005). Rings move through the GOM carrying anomalies similar to the contrasts in physical, biological, and chemical properties observed across the major ocean currents. As a matter of fact, the ring field associated with the LC affects almost every aspect of the circulation in the entire GOM, being considered as the most energetic events in the circulation of the GOM (Sturges and Leben 2000). Once the rings are propagating into the western Gulf, cyclonic features may occasionally cleave these anticyclonic rings into smaller eddies (Biggs et al. 1996). The shedding process can take several days to a few weeks, and often, after a ring has separated, it reattaches to the LC (Sturges et al. 1993). The frequency of LC ring shedding has been studied by many authors (Maul and Vukovich 1993; Sturges and Leben 2000; Leben 2005), stating that the interval between Loop Current ring shedding varies in the range of 6 to 17 months (Molinari 1980), with primary periods of 6 and 11 months (Sturges and Leben 2000). Physical oceanography and ecosystems in the Gulf of Mexico 25 A diverse group of complex processes variably participate in the composite mechanism of ring shedding. For a generalized mechanism of first detachment, either partial or complete, cyclone pairs, one on each side of the LC, play a key role. Detachment may occur after an extended LC is pinched-down or necked-down by cyclones, but not every necking-down is followed by detachment (Schmitz et al. 2005). The formation of these cyclones in the vicinity of the LC ring during the separation stage has also been studied (e.g., Vukovich and Maul 1985; Cherubin et al. 2005). Westward propagation is also involved. The initial observational description of first detachment was published approximately 40 years ago (Cochrane 1972) when peripheral cyclones on the boundary of the LC were observed. It was called first mode of detachment. Another prototypical view of detachment is observed to occur when the LC has an east-west orientation on its tip (Schmitz et al. 2005). This tends to occur after a first detachment or near-detachment is followed by subsequent reattachment(s), and it is usually called second mode of detachment. Pulling apart by westward propagation is the principal initial source for this second mode of detachment, with cyclones slipping in behind the embryonic ring separation to bring about final visual detachment. In these cases, peripheral cyclones on the northern side of the LC often participate in the detachment process, normally in conjunction with a cyclone on the western side of the LC. Mixtures of these two modes of detachment may also occur, involving both cyclonic intrusion and westward propagation in roughly equal proportion (Schmitz 2005). The mechanisms of ring separation will be described in further on a separate section. However, one well-known component process, the penetration (sometimes referred to as an intrusion or northern excursion) of the LC approximately northwest into the GOM, is a basic requirement for the future separation of rings. LC rings can be strongly reattached, involving the recapture of a previously detached ring by the LC. In addition, it is observed that reattachments can be made by neck filling after near detachments, along with weak reattachments by a touching-like contact between the LC and other anticyclonic features. An exhaustive list of previous studies of interest with respect to elements of the composite ring shedding process and reattachment can be found on the references (Cochrane 1972; Elliott 1982; Fratantoni et al. 1998; Molinari et al. 1978; Sturges and Leben 2000; Sturges et al. 1993; Vukovich 1988; Vukovich and Maul 1985; ZavalaHidalg et al. 2003; Zavala-Hidalgo et al. 2006). Regarding the deep circulation in the Gulf, a basin scale cyclonic circulation has been defined in the deepest layers in the GOM (DeHaan and Sturges 2005). Although the upper 26 Chapter II layer mean flow that has been described above -when describing the LC and ringsis anticyclonic, mean flow near the edges of the GOM below 2000 m is cyclonic. The mechanism underling this cyclonic deep flow is based on deep topographic Rossby waves that are rectified by bottom friction (Mizuta and Hogg 2004). Direct observations from drifters in the intermediate depths of the GOM at 900 dbar (approximately at depths of 900 m) also show a mean cyclonic circulation along the northern and western edges of the GOM. 3.3 Coastal flow regimes Coastal circulation in the GOM may be characterized dynamically as mostly wind and buoyancy forced on the inner and middle continental shelves. For wide shelves, coastal circulation regimes over the inner and middle shelves tend to be decoupled dynamically from flows over the continental slope and rise. Both downwelling and upwelling coastal regimes are found in the GOM, typically a seasonal phenomena due to observed regional wind fields. Many of the adjacent shelf circulation systems in the GOM interact with each other. The mechanism of cross-shelf-slope exchange in the region is related to the presence of the LC ring and a cyclonic byproduct of its topographic interaction. This exchange, firstly reported in 1972 (Nowlin Jr 1972), is due to the onshore/offshore movement of water between the shelf and the deep GOM as induced by this type of cyclone-anticyclone pair over the continental slope (Brooks and Legeckis 1982). Pairs of sub-mesoscale cyclonic eddies with a presence on the continental slope are also a well established mechanism for moving water off the shelf system in the northern GOM (Biggs and Müller-Karger 1994; Biggs et al. 2005; Hamilton and Lee 2005). Eddy vorticity flux over the continental slope and rise is typically greater than the contribution from wind-stress curl (Ohlmann et al. 2001; Ohlmann and Niiler 2005). Obviously, circulation patterns on the continental shelf also depend on shelf geometry as well as river distribution and the nature of offshore flow and forcing (Weisberg et al. 2005). Physical oceanography and ecosystems in the Gulf of Mexico 27 3.4 Is there a unique mechanism for Loop Current intrusion and ring sheddings? Resolving, understanding, and monitoring the upper ocean mesoscale field and its vertical thermal structure, including an accurate knowledge of the position LC rings and LC fronts, appear to be critical elements for a variety of applications: in tropical cyclones intensification studies and forecasts (Leipper and Volgenau 1972, Shay et al. 2000, Kaplan et al. 2009, Goni et al. 2010); in oil spill response management (Kolch et al. 1991, Kaiser and Pulsipher 2007); in navigation; in search and rescue operations; and in understanding the environmental conditions that influence distributions of the larvae of different fish species and spawning sites (Ortner et al. 1984, Richards et al. 1993, Bakun 2006). Also the detached LC rings may have a measurable impact on the heat budget of the GOM since a LC ring can extend to a maximum depth of 1000 m and the volume of a LC ring has been estimated to be as much as 7% of the total volume of the GOM (Elliot, 1982). This leads to the question: What physical processes might drive the LC intrusion, trigger the ring shedding, and set the initial state of the LC after ring separation? The mechanism supporting LC and ring shedding has been widely studied, notably by Hurlburt and Thompson (1980) on their classic modeling paper (henceforth HT), and more recently interpreted using the “momentum imbalance paradox” idea of Pichevin and Nof (1997) (henceforth PN) (see also Nof and Pichevin, 2001; Nof, 2005). These studies showed that the ring shedding can be captured by a single layer reduced gravity (1.5) layer model without the need to consider the interaction with the variable bottom topography. When discussing circulation features and model resolutions, it is relevant to have a clear idea of the first-mode baroclinic radius of deformation in the region. The baroclinic Rossby radius of deformation plays a fundamentally important role in large-scale ocean circulation theory, since it describes the horizontal scales of meso-scale processes and it represents an important measure of the ocean dynamics. The internal (baroclinic) Rossby radius of deformation is the ratio between the phase speeds of the long internal waves to the Coriolis parameter. Basically, it is the horizontal (length) scale at which rotation effects become as important as buoyancy (or gravity waves) effects (e.g., Gill 1982); in other words it defines the length scale of baroclinic variability longer than which internal vortex stretching is more important than relative vorticity and it is intimately related to the dominant 28 Chapter II length scale of unstable waves in a stratified shear flow. Outside near-equatorial latitudes (about ± 10° from the equator), the first baroclinic Rossby radius can be formally defined as the distance, �!, that first baroclinic gravity waves of speed �! propagate over time �!! (Gill 1982): �!=!! ! (2) where � is the planetary vorticity. Values of the first baroclinic Rossby radius in the ocean vary with latitude, ranging from a few kilometers at high latitudes to greater than 100 km near the equator. A standard procedure for evaluating the baroclinic Rossby radii of deformation is to use a linearized quasi-geostrophic potential vorticity equation. Employing the Wentzel-Kramers-Brillouin (WKB) method (Gill 1982; Chelton et al. 1998), the phase speed of the first baroclinic gravity wave is �!=�!!�� ! !!!�� (3) where � is a buoyancy frequency, computed following the procedure outlined in section d of appendix B in Chelton et al. (1998), and �(�,�) is the depth of the ocean. Calculation for the GOM gives a first baroclinic Rossby of deformation of approximately 40 km (Chelton et al. 1998). 3.4.1 The Hurlburt and Thompson model Hurlburt and Thompson (1980) (henceforth HT) pioneering work in modeling the LC and rings in the GOM is today a benchmark against which many later model experiments are measured. HT experimented with 1-layer barotropic, 1.5-layer reduced-gravity, and 2-layer models of an idealized rectangular Gulf of Mexico basin with inflow (Yucatan Channel) and outflow (Florida Straits) ports (Figure 2.6). Physical oceanography and ecosystems in the Gulf of Mexico 35 (Schmitz and Richardson 1991). This water of South Atlantic origin is in the Florida Current concentrated in two comparatively fresh water masses. There is an additional significant contribution of deeper, mostly upper Antarctic Intermediate Water (Schmitz and Richardson 1991b; Schmitz Jr 2004). The rest of the upper ocean water entering the CS, the GOM though the LC, and then flowing to the Atlantic within the Florida Current is of North Atlantic origin. These water masses have been studied in the GOM by different authors, and it has been suggested that the it is related with the variability in the flow in Windward Passage and the Old Bahama Channel (Johns et al. 2002; Hamilton et al. 2005). The two passageways providing the entry and exit routes for transport in and out of the GOM, flow through the Yucatan Channel between Yucatan and Cuba and flow though the Straits of Florida between Florida and Cuba, are intimately connected. Assuming incompressibility in the mass conservation law, the equation for the balance of volume in the Gulf of Mexico may be written as !!!"# !" =� !+� !+�+(�−�) (4) where � !"# is the total volume of water in the Gulf of Mexico, � ! is the volume transport across Yucatan Channel, � ! is the volume transport across the Florida Straits, � is the runoff and �−�!is the volume transport due to precipitation minus evaporation. Estimates of �+(�−�) made by Etter [1983] show that these terms are very small, on the order of 0.1 percent of � ! and � !. Since altimetry derived estimates of !!!"# !" indicate that it is a small term (Bunge et al. 2002), large terms in equation (4) (TY and TF) almost balance out, the transport into the GOM through the Yucatan Current is expected to be balanced by the transport out through the Straits of Florida. The deep flow between the Caribbean and the Gulf has been studied, finding a 3-layer flow in the waters below the sill depth (approximately of 700 to 800 m) of the Straits of Florida. First there is inflow to the Gulf from the Caribbean, through the Yucatan Channel, at depths of approximately 800 m down to 1200 m. There is also an inflow from the Caribbean in the layers just above the sill at 2000 m. What it is surprising is that there is a return flow of approximately 1 Sv at depths of 1300 to 1900 m, from the GOM into the Caribbean (Rivas et al. 2005). 36 Chapter II Many efforts have been made in the past to connect the intrusion of the Loop Current into the GOM and the associated ring shedding with the flow structure and transport through the Yucatan Channel (Maul et al. 1985; Candela et al. 2003; Bunge et al. 2002; Oey 1996; Ezer et al. 2003). The time dependent exchange at the Yucatan Strait between the GOM and the CS, within both the upper and deep ocean, was first suggested in 1977 when cooscillation between the Cayman Sea and the GOM were found to be connected with time dependent growth of the LC (Maul 1977, 1978). That is, the growth of the LC as it expands to the north displaces an approximately equivalent volume of Gulf water, which is observed to flow back into the CS at depths below 800 m (Bunge et al. 2002). The Maul View has become a key contemporary issue (Bunge et al. 2002; Candela et al. 2002, 2003; Oey et al. 2003; Oey 2004; Oey et al. 2005; Rivas et al. 2005; Chérubin et al. 2005). The intrusion of the Loop Current into the GOM is measured differently in different studies. For example, Vukovich (1995) used monthly averaged distance between the northern boundary of the Loop Current and the 30°N latitude line to study the Loop Current ring shedding frequency. Bunge et al. (2002) evaluated the variability of the Loop Current by the surface extension of the Loop Current, which was inferred from a series of radiometer images using graphic software to manually define the boundaries of the current. Ezer et al. (2003) generated a time series of variations of the Loop Current extension defined as the area averaged sea surface elevation over the region of the Loop Current (89°W to 83°W, 21°N to 27°N). The Gulf-Caribbean connectivity is explored in Sturges (1992), who demonstrated interactions of the natural shedding frequency with the frequencies of variability of other oceanographic forcing fields, such as the Yucatan Channel inflow, the Florida Current and the North Brazil Current variability, as well as the synoptic meteorological forcing variability. Murphy et al. (1999), showed that Caribbean eddies that squeeze through the Yucatan Channel can affect the timing of Loop Current shedding. Oey et al. (2003) concluded that wind-induced transport fluctuations through the Greater Antilles Passages cause shedding at shorter periods, while Caribbean eddies (anticyclones) cause shedding at longer periods. Oey (2004) argued that the potential vorticity flux anomaly at the Yucatan Channel may serve as a determining factor. Although previous studies consitute a big endeavor to better understand the behavior of the LC and detached ring field by exploring the upstream conditions, the variability in the LC system -including the Yucatan Current and Florida Currentassociated with ring shedding remains not fully understood. Physical oceanography and ecosystems in the Gulf of Mexico 37 4. Marine biology and fisheries of the Gulf of Mexico Ichthyoplankton (from Greek: �����, ikhthus, "fish"; and ��������, planktos, "drifter") refers to fish eggs, newly hatched eggs (fry), young fish, and adults of small fish, whose sizes are from 153 µm long to 5,000 µm or longer (Froese and Pauly 2011). Planktonic organisms -any drifting organisms (animals, plants, archaea, or bacteria) that inhabit the pelagic zone of oceans, seas, or bodies of fresh waterare typically aggregated in the vertical and horizontal dimensions on a variety of temporal and spatial scales. This aggregation is influenced by biological characteristics of these organisms and by physical features of their habitat (Ortner et al. 1978; Owen 1981; Mackas et al. 1985). The propagation of plankton, their behavior, and their buoyancy interact with the physical hydrographic structure and hydrodynamic motion of the ocean to affect aggregation. Among the planktonic organism, ichthyoplankton is referred to the eggs and larvae of fish. Although they are typically considered inert particles, fish larvae are interactive components of the ecosystem (Cowen 2002; Fuiman 2002). Most fishes have a pelagic stage, whose duration varies between weeks to months (Brothers et al. 1983; Victor 1986). During this pelagic stage, important changes occur in short time periods. For example, most fishes increase their weight 5 orders of magnitude throughout their life and three of them occur during this pelagic stage (Werner and Gilliam 1984; Houde 1987; Miller et al. 1988). The sharp biomass increase during the short larval period evidences the importance of this stage as a recruitment modulator (Houde 1987). Thus, growth and processes that favor larval survival are crucial for recruitment success (Cushing 1975; Cowan Jr and Shaw 2002). Thus, it is important to study ichthyoplankton because the abundances of eggs and larvae of several species have been demonstrated to be good indicators of the transient spawning population size of the adults. Determining the abundance of eggs and larvae in an area is usually less expensive to do than sampling the adults. For species such as sardine and anchovy, for instance, egg and larval counts are good indicators of population size. Thus, we can use the egg and larval data to monitor trends in population abundance of the adults. We are able to tell when populations are declining, often more rapidly than we could if we were just 38 Chapter II monitoring adults. For species that are not captured by a fishery, monitoring their population trends by monitoring their eggs or larvae can provide an indication of a healthy or stressed ecosystem. It is unlikely that we would have an idea of the abundance, growth or decrease of these species in any other way, so ichthyoplankton monitoring becomes essential. 4.1 Biological levels and limiting factors in the Gulf of Mexico Plankton typically flow with ocean currents. While some forms are able to move independently and they can swim hundreds of meters vertically in a single day -a behavior called diel (or diurnal) vertical migrationtheir horizontal position is primarily determined by the surrounding currents. This is in contrast to nekton organisms that can swim against the ambient flow and control their position (e.g., squid, fish, and marine mammals). Within the plankton, holoplankton spend their entire life cycle as plankton (e.g., most algae, copepods, salps, and some jellyfish). By contrast, meroplankton are only planktic for part of their lives (usually the larval stage), and then graduate to either a nektic or benthic (sea floor) existence. Examples of meroplankton include the larvae of sea urchins, starfish, crustaceans, marine worms, and most fish. Planktonic organisms are primarily divided into 3 trophic level groups: phytoplankton, zooplankton, and bacterioplankton2. 4.1.1 Phytoplankton variability in a non-oligotrophic region Phytoplankton (from Greek phyton, or plant) are autotrophic, prokaryotic or eukaryotic algae that live near the water surface where there is sufficient light to support photosynthesis. Among the more important groups are the diatoms, cyanobacteria, dinoflagellates and coccolithophores. Although phytoplankton are too small to be individually seen with the unaided eye, when present in high enough numbers, they may appear as a green discoloration of the water due to the presence of chlorophyll-a within their cells. They are agents for 2 Bacterioplankton refer to the bacterial component of the plankton that drifts in the water column. bacteria and archaea, which play an important role in remineralising organic material down the water column. They will not be further described in this work. Physical oceanography and ecosystems in the Gulf of Mexico 39 primary production, the creation of organic compounds from carbon dioxide dissolved in the water, and a process that sustains the aquatic food web (Peters and Marrasé 2000). Phytoplankton obtain energy through the process of photosynthesis and must therefore live in the euphotic zone of the ocean, account for half of all photosynthetic activity on Earth (Boyce et al. 2010). The GOM had in the past been described as an oligotrophic system (Ortner et al. 1984). However, satellite data and in situ observations (Muller-Karger et al. 1991; Gilbert et al. 1996; Gilbes et al. 1996, 2002; Muller-Karger and Fuentes-Yaco 2000; Biggs and Ressler 2001; Belabbassii et al. 2005; Biggs et al. 2008) helped demonstrate that the GOM experiences intermediate to high phytoplankton concentrations both over the shelf and in certain offshore areas, and that some of these patterns are seasonal. Some changes in inshore areas are related to wind-driven coastal upwelling (Chuang et al. 1982; Schroeder et al. 1987; Yang and Weisberg 1999; Muller-Karger and Fuentes-Yaco 2000; Weisberg et al. 2000) and river plumes (Gilbes et al. 1996; Walker 1996; Del Castillo et al. 2000; Hu et al. 2003). In deeper water, variation in chlorophyll-a is affected by seasonal convective mixing, divergenceand convergence-associated cyclonic and anticyclonic eddies (Biggs and MullerKarger 1994) and the off-margin entrainment dispersal of riverine outflows (Biggs et al. 2008). Using Coastal Zone Color Scanner (CZCS) images, Muller-Karger et al. (1991) found that variability in pigment concentration seaward of the shelf was synchronous throughout the Gulf of Mexico, with the highest values in winter (from December to February) and the lowest during summer (from May to July). Also using CZCS images, Gilbes et al. (1996) observed an episodic plume with high pigment concentration developed each spring extending along the West Florida Shelf from an origin region within the North Eastern GOM. This plume persisted for 1–6 weeks in a pattern that extended >250 km southward along the shelf. After the plume dissipates, low pigment concentrations are generally observed during the summer along the outer West Florida Shelf, while coastal concentrations increase as a result of higher river discharge by Florida rivers (Gilbes et al. 1996). Occasionally in summer, surface waters from the Mississippi River can be entrained either by deep water off the edge of the West Florida Shelf or by slope eddies into the eastern edge of the Loop Current, and therefore extend for hundreds of kilometers to the southeast of the delta (Hu et al. 2003, 2005). The high spatial and temporal variability of pigment concentration in the GOM increases from east to west and from south to north, due to the difference in thermal stratification across the region, the intrusion of nutrient-poor waters from the western 40 Chapter II Caribbean, the occurrence of cold fronts and extra-tropical low pressure systems, and the nutrient distribution from the Mississippi river (Melo-González et al. 2000). 4.1.2 Zooplankton, from near-shore to central Gulf Zooplankton (from Greek zoon, or animal), are small protozoans or metazoans (e.g., crustaceans and other animals) that feed on other plankton and telonemia. Some of the eggs and larvae of larger animals, such as fish, crustaceans, and annelids, are included here. Zooplankton feed on bacterioplankton, phytoplankton, other zooplankton (sometimes cannibalistically), detritus (or marine snow) and even nektonic organisms. As a result, zooplankton are primarily found in surface waters where food resources (phytoplankton or other zooplankton) are abundant. Just as any species can be limited within a geographical region, so is zooplankton. However, species of zooplankton are not dispersed uniformly or randomly within a region of the ocean. Instead ‘patches’ of zooplankton species (this also applies to phytoplankton) exist throughout the ocean. Though few physical barriers exist above the mesopelagic3 zone, specific species of zooplankton are strictly restricted by salinity and temperature gradients; while other species can withstand wide temperature and salinity gradients (Lalli and Parsons 1997). Zooplankton patchiness can also be influenced by biological factors, as well as other physical factors. Biological factors include breeding, predation, concentration of phytoplankton, and vertical migration (Lalli and Parsons 1997). The physical factor that influences zooplankton distribution the most is the mixing of the water column (upwelling and downwelling along the coast and in the open ocean) that affects nutrient availability and, in turn, phytoplankton production (Lalli and Parsons 1997). Through their consumption and processing of phytoplankton and other food sources, zooplankton play a role in aquatic food webs, as a resource for consumers on higher trophic levels (including fish). Thus, zooplankton are the initial prey item for almost all fish larvae as they switch from their yolk sacs to external feeding (Figure 2.9), and it is considered as the principal food of larval fish (Turner 1984; Govoni et al. 2010). Fish rely on the density and distribution of zooplankton to match that of new larvae, which can otherwise starve. Natural factors (e.g., current variations) and man-made factors (e.g., river dams) can strongly affect zooplankton, 3 The mesopelagic zone limits are from 200 m to around 1000 m. Although some light penetrates this second layer, it is insufficient for photosynthesis. At about 500 m the water also becomes depleted of oxygen. Physical oceanography and ecosystems in the Gulf of Mexico 41 which can in turn strongly affect larval survival, and therefore breeding success. Figure 2.9. Larval fish (Salmo salar, Atlantic salmon) egg hatching, and start of the external feeding. The Alevin (larva) has grown around the remains of the yolk sac. In about 24hrs it will be a fry without yolk sac and it will start external feeding from zooplankton. Image provided by Dr. Uwe Kils Institute of Marine and Coastal Sciences, Rutgers University. Studies of vertical distribution of zooplankton in the eastern GOM showed that the zooplankton community is diverse, 21 genera individually exceeding 1% of the biomass in the 0 to 1000 m layer (Hopkins 1982). Grazers (herbivores, detritivores, omnivores) are 66% of the 0 to 1000 m standing stock and carnivores 34%, their biomass in the epipelagic4 zone above the base of the thermocline (150 m) at night increasing from 46 to 57%. Zooplankton biomass available as forage for higher trophic levels is most concentrated in the upper 50 m, whereas, paradoxically, the zooplanktivorous micronekton, the myetophid fishes in particular, are centered deeper, primarily between 50 and 150 m (Hopkins 1982). Regarding the variability of zooplankton in the GOM, analysis of zooplankton community composition 4 The epipelagic zone is the illuminated zone at the surface of the sea where there is enough light for photosynthesis. Almost all the primary production in the ocean occurs here. The epipelagic zone limits are from the surface mean sea level down to around 200 m. 42 Chapter II and abundance in and beneath the Mississippi River outflow, and at similar depths at control stations in regions beyond the influence of the river off western Florida and in the central Gulf, showed that copepod5 abundances were significantly higher in the nearshore regions than in the central GOM (Ortner et al. 1989). 4.1.3 Are light or nutrients conditioning phytoplankton success? The dominant factor limiting growth of phytoplankton and zooplankton varies from region to region in the world's oceans. On a broad scale, growth of phytoplankton in the oligotrophic tropical and subtropical gyres is generally limited by nutrient supply, while light often limits phytoplankton growth in subarctic gyres. Like many other aquatic plants, photosynthetic marine organisms (i.e., phytoplankton) rely upon sunlight and chlorophyll-a to absorb visible light from the sun as well as nitrogen (N), nearly all phosphorus (P), and silica (Si) to generate food and promote growth and reproduction. However, the amount of light penetrating the ocean surface tends to decrease with increasing water depth, therefore photosynthesis can only take place within a small band near the surface of the water (called the photic zone). In addition, nutrient availability often varies significantly from place to place. For example, in the open ocean, nutrient levels are often very poor causing primary production to be very low. In contrast, near shore waters such as estuaries and marshes are often rich in nutrients, allowing primary production to be very high. In some instances, nearshore ecosystems have an excess of nutrients due to runoff and other terrestrial sources. Excess nutrients can cause an over-stimulation of primary production, depleting oxygen levels and causing eutrophic conditions to occur in coastal habitats. Environmental variability at multiple scales influences the nutrient and light available for phytoplankton, and as these organisms form the base of the marine food web, this variability in phytoplankton growth influences higher trophic levels. For example, at interannual scales phytoplankton levels temporarily plummet during El Niño periods, influencing populations of zooplankton, fishes, sea birds, and marine mammals. On broad scales, phytoplankton and zooplankton have patchiness that is consistent with the local hydrodynamic features (i.e., passive response), while zooplankton shows more spatial 5 Copepods are usually the dominant members of the zooplankton (Boxshall and Halsey 2004). Physical oceanography and ecosystems in the Gulf of Mexico 43 variability than phytoplankton at finer scales and some other mechanisms such as the swimming behavior (i.e. active response) must be invoked to explain zooplankton patchiness (Levin 1992). The behavioral mechanisms (Bollens and Frost 1991; Rios-Jara and Gonzalez 2000) and physical processes (Peterson et al. 1996; Yen et al. 1998) affecting the patchiness of zooplankton are likely to be speciesand size-specific. Ichthyoplankton may also be affected by those factors, although not with the same intensity since the relationship between zooplankton biomass and ichthyoplankton commonly is not straightforward (de Ciechomski and Sanchezf 1983; Cowan Jr and Shaw 1991; FloresCoto et al. 2009). Observed heterogeneity in distribution patterns of the ichthyoplankton can be explained by different processes acting in a continuum of scaling dimensions that range from El Niño events (Yoklavich et al. 1996), circulation patterns, fronts (Cowen et al. 1993), river plumes (Govoni et al. 1989), flood and ebb tides(Raynie and Shaw 1994), physical properties of the water column (Rakocinski et al. 1996), and finer scale turbulence (Muelbert et al. 1994). In the GOM, the depth of the mixed layer -embodying both light limitation and nutrient availabilityhas been suggested to be the single most important factor controlling the seasonal variation of chlorophyll-a concentrations in offshore waters. Since algal biomass is higher when the surface mixed layer is deeper, this is strong evidence that primary production in the region is controlled by variations in upward nutrient flux (Muller-Karger et al. 1991) There is enough illumination in the mixed layer all year round, and only in the northern GOM and during winter months light could appear as a factor to take into consideration (Rabalais et al. 2001). For example, phytoplankton growth rate limitation due to underwater light attenuation has been shown to be substantially greater than growth rate limitation due to nonoptimal nutrient concentrations in the Louisiana inner shelf, but the degree of nutrient limitation increases with increasing distance from the Mississippi Delta (Bierman et al. 1994). Nutrients loadings in the GOM have increased during the last century, especially during the rapid growth of population, agriculture, and fertilizer production beginning in the 1950’s (Nixon 1995). However, phytoplankton productivity in the Northern GOM region has been shown to be inorganic phosphorus limited in the spring and early summer during high runoff periods (Ammerman 1992; Dortch and Whitledge 1992; Lohrenz et al. 1999; Rabalais et al. 2007; Sylvan et al. 2006; Turner et al. 2007), and silicate limited in the spring has also been documented (Dortch and Whitledge 1992; Redalje and Fahnenstiel 1994; Lohrenz et al. 1999). Phytoplankton may also be light limited at times on the Louisiana shelf (Rowe and 44 Chapter II Chapman 2002). It is for those reasons that it is important to monitor the ocean structures that control the upwelling and downwelling processes (Biggs and Müller-Karger 1994; Toner et al. 2003), as well as the wind field, responsible of enhancing eddy pumping of nutrients into surface (McGillicuddy et al. 1998; Falkowski et al. 1991). 4.2 Larval fish assemblages in the Gulf of Mexico A larval fish assemblage can be defined as a suite of species whose larvae are collected in the same area at the same time (Miller 2002). Therefore, a larval assemblage is by definition transient and does not necessarily imply or require evolutionary convergence or biological interactions (e.g., competition, predation). The species association within an assemblage only denotes that they have come to a similar solution for a precise stage of their lives (Miller 2002). Mortality is critical during the larval stage of a fish, because larvae have to overcome survival rates below 1% (Houde 1987; Chambers 1997). Nowadays, predation is considered as the main mortality factor influencing recruitment (Hunter 1981; Bailey and Houde 1989) , and mortality is modulated by other factors: (1) Temperature that influences physiology, metabolism, behavior, and early life growth rates (Blaxter 1991). It operates on all time and space scales (from cell level to large ocean basins); (2) Physical processes (diffusion, advection and dispersion), whose importance was already acknowledged in some recruitment hypotheses (Cushing 1975; Heath et al. 1988); (3) Large amounts of prey are associated with increased survival rates and with faster larval growth (Zenitani et al. 2007). However, turbulence can increase encounter rates in environments with low prey densities (MacKenzie et al. 1994); and (4) Nutritional condition and growth rates influence predation through larval size and growth rates (Meekan and Fortier 1996; Vigliola and Meekan 2002): “the faster you get bigger the better” (Houde 1987). Many factors influencing larval behavior and physiology are size-dependent (Hunter 1981). Thus, smaller larvae, with limited energy storage (yolk sac), are more susceptible to starvation and, consequently, to be predated. Therefore, recruitment variability is now considered to be the result of an integration of processes operating on different time and space scales acting throughout the pre-recruit fish life-stage (Figure 2.10). That is to say, recruitment success is not determined during a particular ontogenetic stage and it depends on the species, populations and environmental factors (Houde 2008). Loop Current excursions and ring field detachments during 1993-2010 53 1. Introduction The objectives of this chapter are 1) to monitor and describe the spatial and temporal variability of LC intrusions (northward), LC retreats (southward), and ring detachments, and 2) to explore a possible link between this variability and the changes in the upper ocean thermal structure in the GOM. In order to do that, satellite altimetry observations are used in this work to determine the dynamic structure in the region. The temporal and spatial variability of the main mesoscale features in the GOM are addressed here in terms of the northward and westward intrusion of the LC, and using a characterization of ring shedding events. A peer-reviewed paper focused on these aspects of the LC excursions and ring field detachments is under revision: Lindo-Atichati, D., F. Bringas, and G. Goni, 2011: Loop Current excursions and ring field detachments during 1993-2010. Submitted to Remote Sensing Letters. The upper ocean circulation in the Gulf of Mexico (GOM) is characterized by the fluctuations of the Loop Current (LC), which irregularly sheds anticyclonic rings that travel in a northwest direction into the GOM. The LC forms an intense anticyclonic flow, which expands northwestward in accordance to Rossby wave dynamics (Hurlburt and Thompson 1980) and can extend northward into the GOM to 28ºN, in the vicinity of the shelf break of the West Florida Shelf (WFS) at about 250 km off Mobile Point (Molinari and Mayer 1982). Although the LC intrusion tends to form more frequently in the spring, it may occur in any season and with periods varying from 6 to 17 months with an average period of 10-11 months (Maul and Vukovich 1993). The large, warm-core anticyclonic rings generally propagates westward at speeds of approximately 4 km day-1 with a standard deviation of 3 km day-1, and have lifetimes of days to approximately a year (Vukovich 2007). These rings have radii of about 150 km, may reach depths of 800 m (Oey et al. 2005) and are generated irregularly, with an average shedding period of 11 months and a standard deviation of 4 months (Vukovich 2007). Whereas many authors have studied the frequency of LC ring shedding, the mechanism for eddy detachment remains unknown. However, it has been proposed that when 54 Chapter III the LC grows, the westward Rossby wave speed (which is ∝!−��!, where � is the Rossby radius based on the matured deep Loop) overcomes the growth rate, then the LC goes to an unstable configuration and it usually sheds a ring (Nof 2005). An accurate knowledge of the position and dynamics of mesoscale ocean features, mainly LC rings and LC fronts, is critical for a variety of applications: for understanding the environmental conditions that influence habitat of marine fish larvae (Bakun 2006); oil spill response management (Kaiser and Pulsipher 2007); navigation; search and rescue operations; and hurricane forecasting (Shay et al. 2000). Moreover, northward excursions of the LC have been suggested to be anticorrelated with shoreward excursions of a resilient Cross-Shelf Transport Barrier (CSTB) on the WFS, which provide a large degree of isolation and is consequential for controlling red tides development on the southern WFS (Olascoaga 2010). Satellite altimetry allows the measurement of mesoscale activity over the World Ocean from 1993. More than 18 years of continuous monitoring of the GOM circulation using altimetry help to understand the upper ocean dynamics and the vertical thermal structure at a spatial and temporal resolution that resolves ocean mesoscale features and fronts (Goni and Wainer 2001; Le Traon et al. 1998). Although studies using thermal satellite imagery continue to be invaluable due to the unsurpassed spatial resolution of satellite infrared data, they are limited to winter months due to the lack of thermal contrast over the GOM. In addition, clouds often obscure the thermal satellite imagery, making it difficult to locate and track dynamic mesoscale features. For this reason, satellite altimetry fields, obtained from two to three satellites, are used in this investigation for year round monitoring of the mesoscale circulation in the GOM from January 1993 to December 2009. This chapter is organized as follows. In section 2 the data and methods used in this work are described. Section 3 presents results and analysis of the variability of the LC and anticyclonic rings in the GOM. Finally, section 4 summarizes the main findings of this work. Loop Current excursions and ring field detachments during 1993-2010 55 2. Data and methodology The altimetry data used herein are the optimally interpolated gridded sea! height! anomaly! (SHA) fields produced by Archiving, Validation and Interpretation of Satellite Oceanographic data (AVISO) according to an improved objective analysis method (Le Traon et al. 1998). These fields have a spatial resolution of 0.25 degrees, and temporal resolution of 1 week. The AVISO SHA fields are anomalies computed with respect to the 1993–1999 mean from the direct altimetry observations. It has been found from numerical modeling studies that SHA fields from two independent altimeters are needed for adequate spatial and temporal coverage to properly position mesoscale features and (Goni and Wainer 2001; Le Traon et al. 1998). For example JASON-1 altimeter, which measures the sea height anomaly (SHA) along satellite ground tracks that are approximately separated by 3 degrees zonally and repeats its tracks every 9.91 days, does not allow complete identification of mesoscale features, such as warm core rings (Goni et al. 2011). It has also been found from numerical modeling studies with the Real-Time Ocean Forecast System (RTOFS) that sea height anomaly fields from two independent altimeters are needed for adequate spatial and temporal coverage to properly position mesoscale features and fronts (Goni et al. 2011). The altimetric observations used here to produce gridded fields are obtained from two or three satellites throughout the period from January 1993 to December 2009. The investigation of the dynamics of mesoscale features in the GOM is performed using sea surface height (SSH) fields derived from SHA data and a mean dynamic topography (MDT) of the ocean (Rio and Hernandez 2004) (Figure 3.1), i.e.: ��� =��� +��� (5) where SSH, SHA, and MDT are measured in cm. 56 Chapter III Figure 3.1. Example of general circulation through the Gulf of Mexico from satellite altimetry fields, on May 17 1995. Fields illustrate the LC and the ring shed in terms of (a) sea height anomaly (SHA) and (c) sea surface height (SSH). SSH fields are obtained by adding (b) mean dynamic topography (MDT) to (a) SHA fields.! The temporal and spatial variability of the main mesoscale features in the GOM is addressed herein in terms of the northward and westward intrusion of the LC, and using a characterization of ring shedding events. The LC and the ring field are characterized by their anticyclonic motion and SSH values that are larger than those of their surrounding waters (Figure 3.1(c)). Selected contours of constant SSH values are used here to objectively define the locations of the LC front and the shedding of the rings. 2.1 Location of LC front The location of the LC front is determined as the contours of maximum horizontal gradient of SSH. Specifically, the northernmost and westernmost locations of the LC are detected by characterizing the maximum latitude and longitude of the SSH contours corresponding to the location of the maximum gradient of SSH (Figure 3.2). The methodology used herein to locate the LC front and its northernmost and westernmost locations consists of the following steps: 1) The absolute value of the SSH gradient is computed using zonal and meridional derivatives of the SSH field. 2) The SSH gradient is mapped over the corresponding SSH contour map. 3) The value of SSH contours corresponding to the location of maximum SSH gradient defines the LC front. To avoid confusion between the LC front and the boundary of a ring, Loop Current excursions and ring field detachments during 1993-2010 57 which may have maximum SSH gradients as well, the selected SSH contours belonging to a location of maximum SSH gradient must describe an opened path. 4) The northernmost latitude of these selected contours defines the LC northernmost location. 5) The westernmost longitude of these selected contours defines the LC westernmost location. For example, for the GOM conditions on May 26, 1993 (Figure 3.2(a)), the LC northernmost location is 28°N (upper grey square) and the LC westernmost location is 89.75° W (left grey square). ! Figure 3.2. Fields of altimetry-derived SSH gradient in the GOM with contours of SSH (in cm, black lines) superimposed, for (a) May 26, 1993, and (b) June 02 1993. Fields illustrate (a) the location of the LC front, and (b) the detection of a LC ring and the LC front. The northernmost and westernmost locations of the LC are tagged with a square in both panels. 2.2 Detection of ring shedding The LC ring shedding events are also detected using the methodology described in 2.1. A LC ring is considered to be separated from the LC at the surface when the following two conditions are satisfied: 1) when all the SSH contours belonging to the higher values of SSH gradient describe a closed path (Figure 3.2(b)), and 2) when the first condition last for a period of time longer than four weeks, which is the observed maximum period of time for a LC ring to reattach. The life span of a ring is determined herein as the period of time since the 58 Chapter III ring is shed until the values of the enclosed dynamic height contours decrease to reach similar values to the surrounding waters, typically between 160 cm and 140 cm. Results obtained here are compared with those obtained in previous works of the LC northward penetration (Zavala-Hidalgo et al. 2006) and ring separation (Sturges and Leben 2000; Alvera et al. 2009) that cover shorter study periods. The fact that rings identified and LC excursions studied in this work are in agreement with the findings of previous studies confirms the consistency of the novel methodology used here and complements the valuable previous studies. The results showed here are more complete, cover longer time span, and define objectively the locations of the LC front and the shedding of the rings. 3. Results and discussion 3.1 Northern and western LC excursions Weekly time series of the northernmost and westernmost position of the LC during the period from November 1992 to December 2009 are presented in Figure 3.3. These signals are filtered using a Butterworth filter (black lines in Figure 3.3). This filter is designed with order 6 and angular cutoff frequency 1/7 rad s-1, which means that the boundary in the filter response is 7 weeks and the maximum filter slope is 20 degrees week-1. Loop Current excursions and ring field detachments during 1993-2010 59 Figure 3.3. Time series of (a) LC northward penetration (red line) and (b) LC westward penetration (green line), from November 1992 to December 2009. Black circles indicate the time of separation of LC rings. The signal was filtered using a Butterworth filter of order 6 and cutoff frequency 1/7 rad/s (black line). The northernmost location of the LC oscillates between 24ºN and 28.5ºN, with a mean value (mean ± SD) of (26.5 ± 1)ºN and marked seasonal variability (Figure 3.3(a)). The amplitude of the oscillations exhibits maximum values with meridional motions of 4 degrees of latitude from 1993 to 2003, and meridional motions of 2.5 degrees of latitude from 2004 to 2009. Monthly mean values of the LC northward excursions between 1993 and 2009 indicate that the location of the LC in summer (July through August) is significantly more to the north than in the fall season, with winter and spring having values closer to the mean (Figure 3.4(a)). Similar results are found in previous studies that, using data from 47 cruises in the eastern GOM and monthly fields of temperature at 200 m from 1970 to 1976, have shown that on average the penetration of the LC into the GOM increases during the winter and spring, reaching a maximum in the early summer (Behringer et al. 1977). The northward penetration of the LC also exhibits year-to-year variability (Figure 3.4(b)), showing a mean maximum value of 27.5ºN in the annual northernmost location of the LC in 2005, a minimum value of 25ºN in the annual northernmost location of the LC in 1998, and an interannual mean of around 26.3ºN. Between 1993 and 2002 the mean location of the LC northward penetration is 26ºN, below the annual mean. Conversely, between 2003 and 2009 the mean 60 Chapter III location of the LC northward penetration is 27ºN, above the annual mean. The link between northward excursions and ring formation will be discussed in section 3.2. Figure 3.4. (a) Monthly Mean location of the LC northward penetration (blue circles) and (b) annual mean location of the LC northward penetration. Black bars indicate one standard deviation over the 17 years of data analyzed, from January 1993 to December 2009. The LC westernmost location oscillates between 91ºW and 85ºW, with a mean value of 88ºW and pronounced seasonal variability (Figure 3.3(b)). The amplitude of the oscillations exhibits maximum values with zonal motions of approximately 6º degrees of longitude from 1993 to 2003, and zonal motions of 4 degrees of longitude from 2004 to 2009. 3.2 Ring shedding events The location, lifetime, shape, size, kinetic energy, and available potential energy of LC rings can be characterized by undertaking a ring census similar to those carried out in other regions that used observations of infrared imagery (Brown et al. 1986) or a combination of climatological data, in situ data and satellite altimetry (Goni and Johns 2001). Following the Loop Current excursions and ring field detachments during 1993-2010 61 methodology described in section 2, all the LC ring shedding events between November 1992 and December 2009 (black circles in Figures 3.3(a) and (b)) are identified and described (Table 1). A total of 26 rings are identified as having been shed from the LC between November 1992 and December 2009 (Table 1 and Figure 3.5). The average number of rings formed per year is 1.5, with a standard deviation of 0.7. A marked increase in ring formation is detected beginning in 2003. While from 1993 to 2002 an average number of rings formed per year of 1 ring year-1 is observed, starting in 2003 the average number of rings formed per year is 2 ring year-1. Thirteen out of the 26 rings identified here are shed in the 3-month period from July to September, which is in agreement with the seasonality in the timing of ring shedding suggested in a previous work that monitored ring detachment events from October 1993 to February 2006 using satellite altimetry observations and a visual methodology for ring determination (Alvera et al. 2009). September and August, which are months with LC intrusions to the north, are also the months that exhibit more ring-separation events. There were no ring-separation events identified during the months of January, May, and June from November 1992 to December 2009. The average period between consecutive LC ring detachments is approximately 9 months from 1993 to 2003 with a standard deviation of 6 months, which also is in agreement with the 8.2 months period suggested by other authors (Alvera et al. 2009). The average period between consecutive LC ring detachments and its variability are reduced from 2004 to 2009, exhibiting average periods between consecutive shedding of 6 months with standard deviation of 1 month. 62 Chapter III Table 3.1. Compilation of ring-separation events, including dates of LC ring detachment, period that followed until the next detachment, dates of extreme LC intrusions (northward or westward), dates of extreme LC retreats (southward), and description of such extreme excursions (intrusions or retreats). Rings were identified using estimates of the gradient of SSH derived from satellite altimetry observations, from January 1993 to December 2010, using the methodology described in section 2. Ring Date Life span Extreme LC intrusion Extreme LC retreat Description Jul 7 to Jul 14 1993 LC intrudes to 92°W 1-93 1993 Jul 21 5 2-93 1993 Sep 8 5 3-94 1994 Aug 31 6 4-95 1995 Apr 26 5 5-95 1995 Sep 13 5 6-96 1996-Mar-20 7 7-96 1996 Aug 21 9 Dec 4 to Dec 25 LC retreats to 24 °N 8-97 1997 Sep 24 10 9-98 1998 Mar 04 11 Jun 23 to Sep 22 LC intrudes to 28 °N 10-99 1999 Sep 29 12 11-01 2001 Apr 11 7 Feb 20 to Mar 6 2002 LC intrudes to 93 °W 12-02 2002 Mar 13 1 13-02 2002 Apr 17 11 Apr 3 to Apr 17 2002 LC retreats to 24 °N May 7 to Aug 13 LC intrudes to 28 °N 14-03 2003 Aug 20 4 15-03 2003 Sep 24 10 16-03 2003 Dec 24 8 17-04 2004 Sep 01 11 Apr 13 to Jun 29 LC intrudes to 28 °N 18-05 2005 Aug 03 9 19-06 2006 Mar 08 5 Aug 9 to Aug 23 LC intrudes to 91 °W 20-06 2006 Sep 27 12 21-07 2007 Apr 11 4 22-07 2007 Nov 14 6 23-08 2008 Jul 2 8 24-08 2008 Dec 3 2 Feb 11 to Feb 25 LC intrudes to 91 °W 25-09 2009 Mar 4 8 26-09 2009 Sep 2 7 Mesoscale structures with significant effects on larval fish distribution 71 and Vukovich 1993). The LC returns to its southern location, where water from the Yucatan Current flows more directly toward Florida Strait, after the pinching off of the northern extension of the LC. This process forms large anticyclonic rings which then propagate westward at speeds of 2-5 km day-1, with a lifetime of days to approximately a year (Elliott 1982; Forristall et al. 1992; Shay et al. 1998). These anticyclonic rings have radii of approximately 150 km, swirl speeds of 1.8-2 m s-1, and reach around 800 m depth (Oey et al. 2005). The average time between consecutive sheddings is 9.5 months, with a range of 3 to 21 months (Sturges and Leben 2000; Zavala-Hidalgo et al. 2006). The annual fluctuations in LC flow have been attributed to wind forcing (Sturges and Evans 1983) and to variations in the inflow of water from the Caribbean (Lee et al. 1995; Oey et al. 2003; Alvera et al. 2009). Figure 4.1. General circulation in the Gulf of Mexico on October 22, 2010. Thin arrows represent geostrophic currents derived by combining satellite altimeter observations and numerical model results. Thick arrows contour anticyclonic (Loop Current/LC and rings/AR) and cyclonic features (rings/CR). Knowledge of the temporal and spatial variability of mesoscale structures in the GOM is fundamental for understanding the distribution of larvae of pelagic fish, their spawning 72 Chapter IV habitat, and food availability for larvae (Ortner et al. 1978, 1984; Davis and Wiebe 1985; Richards et al. 1993; Medina et al. 2002; Royer et al. 2004; Wilson et al. 2005; Bakun 2006). Both anticyclonic and cyclonic eddies have been shown to positively affect the abundance and distribution of plankton and fish larvae in other regions of the ocean (Nakata et al. 2000; Okazaki et al. 2002). Physical and biological conditions in the GOM exhibit high spatial and temporal variability (Müller-Karger et al. 1991). The intersection of particular conditions of water temperature, salinity, chlorophyll-a concentration and zooplankton abundance, depth of the water, and day of the year has been associated with a higher abundance of fish larvae in particular areas of the GOM during annual surveys conducted by the NOAA National Marine Fisheries Service (NMFS) in the GOM since 1982 (Muhling et al. 2010; Richardson et al. 2010). Therefore, adult fish appear to target specific habitats or oceanographic features to spawn. The NMFS SEAMAP surveys conducted in spring 1987 showed higher displacement volumes of plankton and higher densities of fish larvae in LC frontal areas (Richards et al. 1993). Mesoscale structures with significant effects on larval fish distribution 73 2. Materials and methods 2.1 Satellite data Boundaries of mesoscale features in the GOM has frequently been defined using sea surface temperature (SST) and ocean color (OC) observations from satellites, and a more limited number of hydrographic observations (e.g., Breaker 1981; Maul et al. 1984; Richards et al. 1989; Müller-Karger et al. 1991; Bigelow et al. 1999b; Royer et al. 2004). Satellite derived observations of sea surface height and surface height anomaly (SSH and SHA), SST and OC offer sufficient temporal and spatial resolution to study the evolution of the main surface mesoscale features in the region. Altimetry measurements have shown great utility in biophysical studies (e.g., Polovina and Howell 2005; Chiswell et al. 2003; Rudorff et al. 2009), and can be used to monitor surface ocean features year-round. For example, SST, SHA, SSH, and the gradient of SSH fields clearly show the structure of the LC and an anticyclonic ring in May 1998 (Figure 4.2a, b, c, & d). Conversely, during the month of June 1998 the LC and the ring cannot be easily observed in SST data (Figure 4.2e) due to the lack of marked gradients as the surface of the GOM warmed up. These features are clearly distinguishable using altimetry observations (Figure 4.2f, g, & h). ! SST fields were obtained from a number of satellite sources. Optimally-interpolated SST fields were obtained from microwave observations retrieved by the TMI and AMSR-E radiometers onboard the TRMM and Aqua satellites, respectively (http://trmm.gsfc.nasa.gov/, http://nsidc.org/data/amsre/). These daily fields have a spatial resolution of 0.25 degrees. SST observations from the Advanced Very High Resolution Radiometer (AVHRR) were available with a resolution of 2 days and 18 km equal-area grid (Ryan et al. 1996). The altimetric observations used in this study were obtained from between two and four satellites operating during the period covered by this study. Altimetry data were the optimally interpolated according to an improved objective analysis (Le Traon et al. 1998) with a resolution of 0.25 degrees and 1 week. 74 Chapter IV Figure 4.2. Comparison between SST and SSH satellite derived fields. Fields on the left panels illustrate a ring separation event in terms of (a) SST, (b) SHA, (c) SSH, and (d) gradient of SSH on May 6, 1998. Fields on the right panels illustrate a ring already shed by the LC in terms of (e) SST, (f) SHA, (g) SSH, and (h) gradient of SSH on June 24, 1998. Mesoscale structures with significant effects on larval fish distribution 75 Because of uncertainties in the geoid, altimeter data are generally estimated as variations about mean sea level. To obtain an estimate of the geostrophic component of the circulation it is therefore necessary to add a mean dynamic height field to the anomalies. The absolute geostrophic velocities derived from the satellite altimeter measuring SSH were obtained from the French AVISO (Archiving, Validation and Interpretation of Satellite Oceanographic data) using the mean dynamic topography (MDT) of the ocean (Rio and Hernandez 2004), as follows: ���!�� =���!�� +���!(��) (6) 2.2 Spatial variability of LC front and local oceanographic feature identification The LC is characterized by a sea surface height greater than surrounding waters, and its position can be determined from the horizontal gradient of SSH. Specifically, the northernmost, westernmost, and easternmost locations of the LC were measured by obtaining the maximum latitude, maximum longitude, and minimum longitude of the SSH contours corresponding to the location of the maximum gradient of SSH (red regions in Figure 4.2d and h, for example). The spatial variability of the LC was studied in terms of northward penetration of the front using the 768 weekly fields of SSH gradient in the GOM. One-way ANOVA was employed to test the seasonal variability of the northern intrusion of the LC. Our approach to study the distribution of larval fish in the LC included defining a region of influence (ROI hereafter) of the LC. The ROI is the area limited by the northernmost latitude, westernmost longitude and easternmost longitude of the LC front. These were compared with the weekly collections of the in situ fish larvae surveys from 26 April 1993 to 29 May 2007. To identify the main mesoscale features in the GOM outside the LC ROI, we classified every grid point in the altimeter fields to have one of the following characteristics: a) An anticyclonic region (AR): ��� ≥���!"# −�∙���� (7) 76 Chapter IV b) A cyclonic region (CR): ��� ≤���!"# +�∙���� (8) c) An anticyclonic region boundary (AB): ��� ≥�∙���!"#!!!!!and!!!!!���� ��� ≥�∙����� ��� (9) d) A cyclonic region boundary (CB): ��� ≤�∙���!"#!!!!!and!!!!!���� ��� ≥�∙����� ��� (10) e) Common Waters (CW) if none of the previous conditions was satisfied. They are defined as the background waters in between the boundaries of mesoscale features. Here, ���� is the standard deviation of SSH in the region and ����� ��� is the standard deviation of the absolute value of the gradient of SSH in the region. ���!"# and ���!"# are the minimum and maximum values of SSH in the GOM for any given date. The five dimensionless parameters m, n, p, q and r were tuned by comparing the circulation patterns classified using the relationships above at 758 sampling stations distributed on a onedegree grid in the northern GOM during summer 2009, using the actual satellite-derived SSH fields, their geostrophic currents for this period of time, and ocean color fields. The values obtained by the tuning for these parameters were: �=0.91 �=3.30 �=0.60 �=1.08 �=0.67 (11) An example of the results from this classification showing the five regions of circulation is shown in Figure 4.3. Mesoscale structures with significant effects on larval fish distribution 77 Figure 4.3. Example of the five regions of circulation and boundaries defined using the mesoscale classification algorithm for June 24, 1998. The background field is the satellite derived SSH. The legend illustrates the mean SSH and standard deviation of SSH for each region of circulation. Colors in the legend correspond to the colors of the satellite derived SSH field. 2.3 Biological data Samples were collected in the GOM north of 24°N during spring (April to June) NMFS SEAMAP surveys between 1993 and 2007. Larval fish data were available from the National Marine Fisheries Service Southeast Area Monitoring and Assessment Program (SEAMAP) database. Cruises were divided into 2 legs and were conducted throughout the United States Exclusive Economic Zone (EEZ) in the northern GOM only. Most of the sampling effort was focused on a one-degree grid of stations, and this grid was usually completed twice each year, with the exception of 2003 and 2004 when they were completed only once. Additional stations were sampled in 1994, 1995, 2005, and 2006 (Table 4.1). Between 41 and 137 hydrographic-plankton stations were carried out each year, with an average of 86 stations per survey in the 15 years covered by this study. At each station, plankton was collected with bongo nets, and CTD casts were completed. Abundance of larvae captured was standardized to larval concentrations (number of larvae per m2) by estimating volume filtered using a flowmeter fitted to the bongo net, and integrating to the towing depth at each station. 120 120 160 160 160 160 200 20 0 200 200 95˚W90˚W85˚W 80˚W 20˚N 25˚N 30˚N1998−06−24 Output from methodology Region SSH[cm] m[cm] AR 175 8 AB 159 2 CW 149 8 CB 135 3 CR 125 7 100 120 140 160 180 200 SSH [cm] 78 Chapter IV Table 4.1. Sampled SEAMAP stations between 1993 and 2007 across the northern GOM. Year Sampling dates Region sampled No of sampled stations with satellite data extracted (*) Net type 1993 26 April - June 15 24°N - 30°N / 84°W - 96°W 84 (14, 38, 26, 13, 16) Bongo 1994 28 April - 9 June 24°N - 29°N / 84°W - 96°W 137 (29, 75, 45, 25, 15) Bongo 1995 19 April - 7 June 24°N - 30°N / 84°W - 96°W 110 (26, 61, 42, 23, 21) Bongo 1996 17 April - 25 May 24°N - 29°N / 83.5°W - 96°W 84 (15, 17, 30, 17, 15) Bongo 1997 17 April - 9 June 24°N - 30°N / 84°W - 96°W 84 (23, 6, 35, 9, 7) Bongo 1998 26 April - 23 June 24°N - 29°N / 84°W - 96°W 68 (16, 44, 27, 15, 16) Bongo 1999 24 April - 31 May 24°N - 29°N / 83.5°W - 96°W 78 (18, 39, 21, 14, 8) Bongo 2000 20 April - 26 May 24°N - 30°N / 83.5°W - 95°W 85 (26, 56, 26, 16, 11) Bongo 2001 18 April - 29 May 24°N - 29°N / 83.5°W - 96°W 87 (24, 45, 20, 22, 9) Bongo 2002 19 April - 28 May 24°N - 29°N / 83.5°W - 96°W 79 (17, 46, 31, 16, 13) Bongo 2003 13 May - 30 May 24°N - 29°N / 83.5°W - 96°W 51 (9, 38, 15, 17, 2) Bongo 2004 13 May - 30 May 25°N - 29°N / 84°W - 96°W 41 (10, 24, 17, 12, 7) Bongo 2005 23 April - 17 July 24°N - 30°N / 83.5°W - 96°W 118 (21, 44, 39, 31, 29) Bongo 2006 23 April - 16 July 24°N - 30°N / 84°W - 97°W 113 (16, 70, 55, 21, 28) Bongo 2007 17 April - 29 May 24°N - 30°N / 84°W - 96°W 66 (21, 48, 17, 19, 10) Bongo * stations containing Coryphaenidae, Thunnus spp., Auxis spp., Thunnus thynnus, and Euthynnus alleteratus Bongo net tows were generally completed across the grid of stations in the GOM in late April and May, with sampling continuing into late June in some years. Sampling extended to July in 2005 and 2006 (Table 4.1). Bongo nets were fitted with 333 µm mesh, on two 61 cm diameter round frames, and were towed obliquely to 200 m depth or to just above the bottom at shallower stations (Richards et al. 1993). Nets were towed at 2-3 knots, and sampling took place during both day and night. Samples from bongo nets were sorted, and larvae identified to the lowest possible taxa at the Sea Fisheries Institute, Plankton Sorting and Identification Center, Gdynia and Szczecin, Poland. 2.4 Larval fish data analysis The spring plankton surveys were originally designed to target Thunnus thynnus larvae (T. thynnus hereafter). However, larvae of more than 500 taxa were recorded over the duration of the surveys. In this study, larvae of five taxa from three commercially important families were analyzed. In many cases, larvae from closely-related species were not distinguishable Mesoscale structures with significant effects on larval fish distribution 79 visually, and so larval groups were merged at genus or family level. Larvae of Coryphaenidae (Coryphaena spp. hereafter, likely incorporating Coryphaena hippurus and Coryphaena equiselis) were analyzed at genus level. Within the family Scombridae (tunas), Auxis larvae (Auxis spp. hereafter, likely incorporating larvae of Auxis rochei rochei and Auxis thazard thazard) were analysed at genus level, as were Thunnus larvae (Thunnus spp. hereafter, likely a mix of Thunnus albacares and Thunnus atlanticus, and not including T. thynnus). Larvae of both T. thynnus and Euthynnus alleteratus (E. alleteratus hereafter) were visually distinguishable from other tuna species, and were therefore analyzed at species level. Larval distributions were compared to the mesoscale ocean features identified as anticyclonic locations, anticyclonic boundaries, cyclonic locations, cyclonic boundaries or common waters using the altimeter data. The statistical methods to analyze the LC seasonality and variability of larval fish relative to the mesoscale circulation features are explained below. 2.5 Probability analyses Associations between larval catches of T. thynnus, E. alleteratus, Thunnus spp., Auxis spp., and Coryphaena spp., and the inner and outer regions of mesoscale features were determined by mean of a probabilistic analysis. The capture locations of each taxon were classified as one of five different categories (AR, AB, CR, CB, or CW). The probability of finding larvae of taxon i in an oceanic mesoscale feature j was calculated using the following quotient: � !�= !!" !!" !!" !!" !!! !!! !!!!! (12) Where cij is the number of individuals of taxon i captured in feature j, eij is the fishing effort, the denominator is the summation of the captures divided by the fishing effort in the five regions, i is the taxa (T. thynnus (1), E. alleteratus (2), Auxis spp. (3), Thunnus spp. (4), and Coryphaena spp. (5)), and j is the region (AR, AB, CR, CB, and CW). The captures of taxon i were standardized by eij, defined as the number of times that feature j was sampled finding a particular taxon i. This was necessary because sampling effort was not equal within 80 Chapter IV all water mass classes. For example, regions defined as CW were sampled more often simply due to the location of the stations during the SEAMAP surveys. 2.6 Multivariate analyses Permutational multivariate analysis of variance (PERMANOVA; Anderson 2001) was used to test the null hypothesis of no difference in 1) total abundances of all the five taxa among the five regions of circulation, 2) the species assemblages of the five taxa among the five regions of circulation, 3) individual abundances of each taxa among the five regions of circulation, and 4) the presence/absence (occurrence) of each taxa among the five regions of circulation. Each test was run on a fourth-root transformed Bray Curtis similarity matrix, in Primer-6 software, with 999 unique permutations to determine significance at p<0.05 (Clarke and Gorley 2006). Mesoscale structures with significant effects on larval fish distribution 87 Conversely, Thunnus spp. were more abundant at higher SSH than other species (Figure 4.7d). Figure 4.7. Mean larval concentration of captures (red circles) and proportion of captures (blue circles) for larvae of (a) T. thynnus, (b) E. alleteratus, (c) Auxis spp., (d) Thunnus spp., and (e) Coryphaena spp. in relation to satellite derived observations of sea surface height (SSH), from 1993 to 2007. SSH values were binned to 10 cm intervals. Error bars represent one standard error. Colored blocks represent SSH intervals of cyclonic regions (CR), cyclonic boundaries (CB), common waters (CW), anticyclonic boundaries (AB) and anticyclonic regions (AR), calculated from the mean and standard deviation of the SSH field for each of the 5 regions 88 Chapter IV 4. Discussion The average cycle of northward penetration of the LC showed maximum values in early summer. This is consistent with the findings of Behringer et al. (1977). While this seasonality was not statistically significant, there was a positive correlation with larval fish density, with higher mean larval concentrations in the eastern Gulf of Mexico when the Loop Current intruded farther north and vice versa. This could be due to a biological response by either adults or larvae to the LC feature, but also to purely physical mechanisms of northward displacement of larvae carried by the LC. Larvae were less abundant in the core of both cyclonic and anticyclonic mesoscale features, and more abundant near anticyclonic frontal areas (Table 4.3 and Figure 4.7). This suggested that larval concentrations within the LC frontal zone may be higher than within the main body of the current. Larvae may be more abundant in boundaries of mesoscale features simply because these represent convergence zones where favorable prey concentrations exist (zooplankton, or fish larvae). Fish larvae eat a variety of zooplankton, and some (e.g. scombrids) may eat each other (piscivory) once they are post-flexion. Alternatively, adult fish may actively spawn in the vicinity of such frontal regions. There is indeed some indication of spatial autocorrelation in the abundance of larvae from different species of fish throughout the SEAMAP time series. T. thynnus, E. alleteratus, and Auxis spp. larvae were most often present within the boundaries of anticyclonic features and in GOM common waters (this typically corresponds to SSH of 140 cm to 150 cm). Presence of T. thynnus was significantly different among zones (p = 0.038), as also was presence of Auxis spp. (p = 0.001). However, probabilities of finding larval-fish were not different among zones for E. alleteratus (p = 0.45), Thunnus spp. (p = 0.488), and Coryphaena spp. (p = 0.057). These results are similar to those from studies in the western Mediterranean (García et al. 2005; Alemany et al. 2010), in which tuna spawning grounds were related to anticyclonic features, and T. thynnus larvae were more abundant near frontal areas. Previous studies showed highest abundance of E. alleteratus and Auxis spp. eggs outside cyclonic eddies in open waters near southern Brazil (Matsuura and Sato 1981) and in the eastern Pacific (Klawe 1963). We found Thunnus spp. larvae to be distributed more broadly and over a wider range of SSH. This agrees with reports that that adult tropical tunas have broader habitat preferences in the GOM than T. thynnus (Weng et al. 2009; Teo Mesoscale structures with significant effects on larval fish distribution 89 and Block 2010), including higher tolerances for warm features (Teo and Block 2010; Muhling et al. 2010) such as the LC and warm LC rings. SSH is proportional to integrated vertical water temperature (Willis et al. 2004), and our results therefore suggest higher tolerances for warm waters for these tropical tuna species, which also is in agreement with previous studies showing that the present Thunnus spp., which included mainly larvae of Thunnus atlanticus, were more abundant in warmer waters than other species such as T. thynnus (Muhling et al. 2010). Some of the variability in larval abundances in GOM common waters, may be due to smaller-scale oceanographic features which were not apparent from an analysis which used only altimetry data. For example, small cyclonic eddies generated at irregular intervals tend to travel along the LC edge (Zavala-Hidalgo et al. 2003b), and may influence larval distributions. Larvae were typically very small, with mean lengths of approximately 4 mm in bongo net samples, and thus young age of around 7 days or less (Brothers et al. 1983; Oxenford and Hunte 1983; Macias et al. 2006; Muhling et al. 2010). They were present in waters with mean geostrophic current (mean ± SD) of 32 ± 28 cm s-1 in the northern GOM, and of 20 ± 18 cm s-1 north of 26°N and west of 87°W. This suggests that larval fish outside the ROI of the LC were captured approximately 120 km from where they were spawned (SD = 100 km). The larvae are unlikely to cross oceanographic boundaries during these early life stages. We believe that larvae outside the LC region were observed near to the spawning location in the northern GOM region. The year-to-year variability of the LC might be reflected in adult recruitment, with possibly higher adult recruitment of T. thynnus, E. alleteratus, Auxis spp., Thunnus spp., and Coryphaena spp. during years of high northward penetration of the LC. Mean geostrophic currents (mean ± SD) in the ROI of the LC were ~57 ± 31 cm s-1. Therefore, larvae captured in this region were transported a mean distance of 342 km from their spawning location (SD = 186 km), i.e. in the northwestern Caribbean. The largest fraction of the spawning habitat of T. thynnus and Auxis spp., however, was within the boundaries of anticyclonic eddies and in GOM common waters of the northern GOM, outside the ROI of the LC. In conclusion, our findings based on coarse spatial resolution altimeter data show that AB regions show higher concentrations of T. thynnus larvae than CB, CR, or AR. Similarly to recent studies in the northern GOM (Muhling et al. 2010) (Muhling et al. 2010), the mesoscale circulation is shown to introduce significant variability into time series which rely 90 Chapter IV on plankton surveys in the region. The larval fish distributions in GOM common waters may be investigated in future research by using other sources of satellite data to overlay on the altimetry. Tracing frontal zones from SST and OC data would give more detail regarding the location of larvae relative to smaller scale features. Chapter V: Future research 94 Chapter V 1. Introduction Now that we have presented the main characteristics of the Physical Oceanography and Fisheries in the Gulf of Mexico (Chapter II), provided an extensive description of the long term variability of the Loop Current, associated ring field, and sea height anomaly in the GOM (Chapter III), and assessed linkages between mesoscale circulation and the spatial and temporal distribution of larval fish in the GOM (Chapter IV), this chapter will explore some potential future research resulting from this dissertation. The results of this dissertation give us room for a wide scope of upcoming research. Three scientific questions that arise after this work are highlighted below: 1. Is there a significant atmospheric coupling between wind stress forcing and the LC migrations? 2. What may be the role of cyclonic mesoscale eddies shed by the LC on the resolution of the “momentum imbalance paradox” raised by PN (1997)? 3. How are primary producers correlated with mesoscale structures in the Gulf of Mexico? In this chapter we will shed some light essentially on the first and second questions, which describe our planned future research directions. The third question will only be introduced broadly, to highlight that including primary production data would complement the binomial ocean mesoscale features – larval-fish distribution section of this dissertation. Future research 95 2. Evidence of coupling between atmospheric pressure/wind field and LC migrations 2.1 Introduction As it has been shown in chapter II, the findings of Hurlburt and Thomson (1980), Pichevin and Nof (1997), and Nof (2005) provide significant insights into the dynamics of the Loop Current and ring shedding. In particular, the latter explains how the LC expands with a mass influx from the Yucatan Current, and how eddies are shed from the LC when, up to a certain large size, the westward Rossby wave speed overcomes the growth rate and the LC. Their valuable works indicate that understanding mass balances in the Gulf of Mexico may hold the key to also understanding the LC and eddy dynamics. The preliminary research presented in this section regarding the possible effects of atmospheric pressure/wind on the migrations of the Loop Current has been inspired by the recent modeling work of Chang and Oey (2010). In their paper they analyze transport balances, and study how these are modified by a local, steady westward wind blowing inside the Gulf of Mexico, intentionally excluding wind stress curl so that they can focus solely on eddy transports (Chang and Oey 2010). They hypothesize that the intensification of the prevailing zonal wind field creates an eastward transport in the central GOM, which delays the detachment of rings and strengthens them (see their Figure 3, p. 2485, and Figure 6, p. 2489). They show that this is caused by the variability in Ekman transport instead of Ekman pumping, since wind curl is excluded in their study. The modeling effort of the work of Chang and Oey (2010) on the effect of wind stress on the delay of LC ring shedding suggests that future study of atmospheric modulation of LC migrations would be beneficial. To date the atmospheric modulation of the LC migrations has not been studied from an observational approach, and this therefore represents a novel future research direction. 96 Chapter V 2.2 Atmospheric pressure variability and LC migrations In order to explore whether variability of Ekman transport has some effect on the migrations (northward and southward) of the Loop Current, the pressure gradients between the northern and southern locations of the loop current were measured. This first approximation based on pressure gradients was assumed to be a valid estimation due to the fact that the wind field in the region is mainly zonal (westward). Future work will extend the direct analysis to the wind stress and wind curl, which drive Ekman transport and Ekman pumping, respectively. Monthly means of surface winds computed from the FNMOC (Fleet Numerical Meteorology and Oceanography Center) 6-hourly analyzed wind fields on a 360 by 181 global spherical grid (1 degree) are calculated and posted to PFEL's (Pacific Fisheries Environmental Laboratory) Live Access Server a few days after the first of every month (Clancy 1992; Rosmond 1992). Monthly mean pressure differences between the northern and southern locations in the GOM were estimated. The northern and southern locations were located along 86.5ºW at 27.5ºN, one degree of latitude above the mean northernmost location of the LC (see Chapter III), and 21.5ºN just in the middle of the Yucatan Channel (Figure 5.1). Figure 5.1. Bottom topography in the Gulf of Mexico (GOM), data retrieved from ETOPO1 1-minute Global Relief (Amante et al. 2009). The map illustrates section A-B, used to estimate the difference in the mean sea level. Future research 97 The wind regime in the Caribbean (including the Yucatan basin) is determined by the North Atlantic Subtropical High (NASH). Located between Bermuda and the Azores, a high air pressure zone centered at around 30°N drives northeasterly trade winds on the southern side of its clockwise atmospheric circulation (Tomczak and Godfrey 2003). The winds over the Caribbean Sea vary owing to the northward migration of the Intertropical Convergence Zone (ITCZ) (Muñoz et al. 2008), from its southernmost position in winter (over the Amazon basin, approximately the Equator) to its northernmost position in summer (Costa RicaNicaragua border, ~ 11°N) (Poveda et al. 2006). Once they approach the Yucatan basin, the northeasterly trades intensify, forming the easterly Caribbean low-level jet (CLLJ) (between 12.5°– 17.5°N and 70°– 80°W) with wind speeds > 13 m s−1 at sea level pressures around 925 mb (Wang 2007; Wang and Lee 2007). To test the hypothesis that the wind regime affects the migrations of the Loop Current, we tested the monthly difference (or anomaly) in the sea level pressure between two locations north and south of the Loop Current, as explained above. Time series of monthly Loop Current migrations overlaid on time series of monthly Sea Level Pressure Anomalies (SLPA) have been determined for 17 years of data, from November 1992 to December 2009, and the time of separation of LC rings has been added (Figure 5.2). Although both time series data were sampled weekly, they were merged to monthly scales to facilitate easier analysis. In general, we observe an apparent negative correlation (or anticorrelation) between the LC migrations and the monthly SLPA. This indicates that when the gradient of pressure increases (and therefore the wind stress and Ekman transport increases), the LC retreats southward. Conversely, when the gradient of pressure decreases the LC intrudes northward. For example, after a ring has been shed (vertical black line on Figure 5.2) in March 1998, the LC migrates south from March 1998 to June 1998, then moves northward from June 1998 to August 1998, and finally moves south from August 1998 to November 1998. The LC then starts a strong and continuous northward intrusion into the Gulf of Mexico from November 1998 to July 1999. For the same period – from March 1998 to July 1999– the SLPA describes an inverse oscillation to the one shown by the LC, increasing SLPA when the LC retreats southward and decreasing SLPA when the LC migrates northward into the Gulf of Mexico. Generally, the visual characterization of this apparent negative correlation is very clear, although there are some periods of time when there is no correlation, such as from 2007 to 2009. 98 Chapter V Figure 5.2. Time series of monthly Loop Current migrations (thin red line) determined from altimetry derived fields, and monthly Sea Level Pressure Anomalies (thin blue line) determined from the pressure difference between two locations north and south of the Loop Current, from November 1992 to December 2009. Vertical black lines indicate the time of separation of LC rings. The signals were filtered using a Butterworth filter of order 6 and cutoff frequency 1/7 rad/s (thick red line and thick blue line for LC migrations and pressure anomalies respectively). Continuous wavelet transforms (CWT) of the time series presented above were used to analyze the variability of the dominant periods of the series, and a wavelet-based semblance analysis was used to correlate them (Figure 5.3). Semblance analysis allows the local phase relationships between the two datasets to be studied as a function of both scale (or wavelength) and time (Cooper and Cowan 2008). In the time series of LC migrations, the periods of 20 weeks and 12 weeks were the dominant periods (Figure 5.3b), with 20 weeks showing the highest amplitude. Similarly, in the time series of the gradient of pressure (or SLPA), there was also a clear signal in the period of 20 weeks (Figure 5.3d), although this signal was faded by the signal at 12 weeks, which in this case showed a much higher amplitude. The semblance analysis showed a negative correlation at all periods along most of the time interval (blue regions on Figure 5.3d). In particular the negative correlation between the LC migrations and the pressure gradient is higher than 80% in the 12 to 20 weeks range. Future research 105 4. Mesoscale features affecting primary productivity in the GOM Addressing the influence of mesoscale features on primary productivity in the GOM will be a valuable complement to the present work, since it will allow us to better understand correlations between mesoscale ocean features and larval-fish assemblages. We already know that a marked seasonality in phytoplankton concentration is present in the GOM, with maxima during winter and minima during summer (Muller-Karger et al. 1991). In addition, there is high spatial and temporal variability in pigment concentrations, which generally increase from east to west and from south to north, due to the difference in thermal stratification across the region, the intrusion of nutrient-poor waters from the western Caribbean, the occurrence of cold fronts and extra-tropical low pressure systems, and nutrient distribution from the Mississippi river (Melo-González et al. 2000). There is sufficient illumination for primary production in the GOM mixed layer all year round, and it has been suggested in previous studies that primary production in the region is controlled by variations in upward nutrient flux (Muller-Karger et al. 1991). Consequently, the preliminary hypothesis of the proposed research is that regional mesoscale ocean features, which control upwelling and downwelling processes, may have a significant effect on the distribution of phytoplankton through deepening the mixed layer and through vertical mixing in the region. The first approach that we will consider will use regional ocean color imagery overlaid on altimetry fields to study spatial and temporal variability of both mesoscale features and phytoplankton in the GOM. Chlorophyll-a images will be derived from the HERMES multisensor combined product (MODIS, MERIS, SeaWiFS) obtained from GlobColorWeb (ftp.fracri.com). Chapter VI: General discussion and conclusions General discussion and conclusions 109 1. Synthesis of results and general discussion The central goal of this dissertation work has been to study the temporal and spatial variability of the Loop Current and rings, and the regulation of larval fish distributions of some species by mesoscale oceanic features. To this end, we developed methodology to locate LC fronts, detect the shedding of LC rings, and we addressed the seasonal and interannual variability of Loop Current excursion, ring field detachments and sea height anomaly trends in the Gulf of Mexico from 1993 to 2009. Once the variability of mesoscale circulation structures in the GOM was defined, we assessed linkages between circulation and the spatial and temporal distribution of larval fish. The first part of the second section of the dissertation (Chapter III) showed four main findings. The first result was that northward penetration of the LC appeared to be seasonal, with a tendency to increase during the spring and show maximum values in the summer. In particular, the location of the LC from July through August was more to the north than in the fall season, with winter and spring having values closer to the mean. This result is based on the long time series of weekly LC migrations we derived, and it is in agreement with what previous studies -covering shorter life span and based on visual methods and in situ temperature samplinghad concluded. Our second finding was that starting in the 2003, the LC was located more to the north; 1.5 degree of latitude above the interannual mean. To date, this result is novel, and it may be closely related with our third result, which showed that also starting in 2003, there was an increase in the number of rings shed from the Loop Current. Finally, we determined the variability of the meridional and zonal oscillations of the LC, showing that meridional motions decreased 1.5 degrees of latitude, and that zonal motions decreased 2 degrees of longitude, from 2004 to 2009. Our proposed future research (Chapter V) is partially oriented to better understand the dynamics of the LC and shedding of rings. The second part of second section of the dissertation (Chapter IV) proposed to combine oceanography and biology to study the variability in spring larval fish collections in the GOM from 1993 to 2009. We obtained three main findings. The first result was that generally, higher larval concentrations occurred during years of high northward penetration. A regression analysis showed that oscillations of the LC about its mean latitude were 110 Chapter VI significantly correlated with larval concentrations. This could be due to a biological response by either adults or larvae to the LC feature, but also to purely physical mechanisms of northward displacement of larvae carried by the LC. The second finding was that larvae were less abundant in the core of both cyclonic and anticyclonic mesoscale features, and more abundant near anticyclonic frontal areas. Larvae may be more abundant in boundaries of mesoscale features simply because these represent convergence zones where favorable prey concentrations exist (zooplankton, or fish larvae). Fish larvae eat a variety of zooplankton, and some (e.g. scombrids) may eat each other (piscivory) once they are post-flexion. Alternatively, adult fish may actively spawn in the vicinity of such frontal regions. The third finding is that Thunnus thynnus -one of the most commercially important species of our workshowed significantly higher concentrations in the boundaries of anticyclonic features than in anticyclonic features, cyclonic features, and the boundaries of cyclonic features. Our future research (Chapter V) will investigate the larval fish distributions at a higher level of detail by using ocean color to assess phytoplankton concentration in the mesoscale ocean features of the region. In particular, we will analyze a recently available larval data from a synoptic survey conducted in 2010, which has a high number of Scombridae larvae and may allow the study of larval-fish assemblages in small eddies with a higher level of detail. 2. Conclusions The main conclusions that arise from this dissertation are: 1. Starting in 2003, the LC was located further to the north than the 1993-2009 mean, and there was a marked increase in the number of rings shed from the Loop Current. 2. The amplitude of the meridional oscillations of the Loop Current has decreased, exhibiting maximum values with meridional motions of 4 degree of latitude from 1993 to 2003, and meridional motions of 2.5 degrees of latitude from 2004 to 2009. Similarly, the amplitude of the zonal oscillations of the LC exhibited maximum values with zonal motions of approximately 6 degree degrees of General discussion and conclusions 111 longitude from 1993 to 2003, and zonal motions of 4 degree of longitude from 2004 to 2009. 3. The Loop Current northward penetration appears to be mostly seasonal (Figure 3.4 (a)), with maximum values of 0.8 degrees of latitude above the annual mean during June, July and August. 4. Monthly sea height residuals, estimated as the difference between the mean sea height anomaly for each specific month and the mean sea height anomaly value for that month since November 1992, showed a positive linear trend of (2.78±0.26) cm/decade during 1993-2009. It is hypothesized here that this increase in the sea height anomaly residuals is linked to the observed increase in mesoscale activity in the region for the same period. 5. The apparent seasonality of the Loop Current northward penetration was positively correlated to larval fish density, with higher mean larval concentrations in the eastern Gulf of Mexico when the Loop Current intruded farther north and vice versa. 6. Larvae of T. thynnus, E. alleteratus, Auxis spp., Thunnus spp., and Coryphaena spp. were less abundant in the core of both cyclonic and anticyclonic mesoscale features, and more abundant near anticyclonic frontal areas of the Gulf of Mexico. Specifically, larvae of T. thynnus, E. alleteratus, and Auxis spp. were most often present within the boundaries of anticyclonic features, and also in common waters of the Gulf of Mexico, defined as the background waters in between the boundaries of mesoscale features. 7. The distributions of T. thynnus larvae in the Gulf of Mexico were similar to the distributions of T. thynnus larvae in the western Mediterranean, in which tuna spawning grounds were related to anticyclonic features, and T. thynnus larvae were more abundant near frontal areas. 112 Chapter VI 8. Thunnus spp. larvae were distributed more broadly and over a wider range of sea surface height than T. thynnus, which confirms reports that adult tropical tunas have broader habitat preferences in the GOM than T. thynnus. 9. Analysis of geostrophic currents in the mesoscale features of the Gulf of Mexico and larvae size and age indicate that larvae outside the Loop Current region were likely to be near to the spawning location in the northern GOM region. 10. The mesoscale circulation was shown to introduce significant variability into time series that rely on plankton surveys in the region. 11. Preliminary investigations show a clear negative correlation between gradients of pressure and the migrations of the Loop Current, suggesting that the Loop Current dynamics might be modulated at some extent by atmospheric forcing. Preliminary investigations on westward propagating long-lived mesoscale eddies indicate that the number of westward propagating cyclonic eddies, likely originated in the Loop Current western boundary, is higher than the number of westward propagating anticyclonic eddies. This might be crucial in rethinking the “momentum imbalance paradox” Chapter VII: Resumen en español (Spanish summary) 116 Chapter VII 1. Introducción El resumen en español se organiza de la siguiente manera. La sección 2 presenta las principales metas y objetivos de la tesis. La sección 3 describe las principales características de la oceanografía física y de los ecosistemas del Golfo de México (GDM): por un lado haciendo especial hincapié en la dinámica de la Corriente de Lazo (CL) y el desprendimiento de anillos, y por otro resaltando los niveles biológicos y sus factores limitantes en el GDM. El planteamiento adoptado y la metodología utilizada en este trabajo de tesis se describe en la sección 4. Los principales resultados y una discusión general de las aportaciones originales de los capítulos centrales (Capítulos III y IV) se presenta en la sección 5. Las futuras líneas de investigación se exponen y discuten en la sección 6. Finalmente, la sección 7 concluye con un resumen de las principales conclusiones. Resumen en español (Spanish summary) 117 2. Objetivos de la investigación El objetivo general de este estudio es investigar la conexión entre las primeras fases de desarrollo de los peces, usualmente denominado ictioplancton, y la variabilidad de las estructuras de mesoescala en el Golfo de México. Este vínculo se explora en esta tesis mediante el uso de observaciones satelitales y muestreos biológicos in situ. Hay muchos ejemplos en la literatura de correlaciones entre los cambios de los parámetros físicos del océano y cambios en los recursos pesqueros (Ortner et al. 1984; Mann 1993; Bakun 2006). Comúnmente, estas correlaciones se mantienen vigentes durante unos años para luego dejar de serlo. Sin embargo esto no significa que la correlación fuera invalida, sino que requiere un mejor entendimiento del mecanismo de interacción en el complejo sistema entre la física y la biología antes de que podamos entender la dependencia temporal de dichas correlaciones. El Golfo de México es un mar semi-cerrado bordeado por los Estados Unidos, México y Cuba. El sustento de las economías de esta comunidad internacional depende de los bienes y servicios que ofrece el GMD; siendo las pescaderías uno de los servicios más importantes. Considerado como una joya entre los recursos naturales del hemisferio oeste, los pantanos del GDM son fuente de una importante industria de marisco, y sus aguas cercanas a la costa sostienen ricas pescaderías comerciales y recreacionales, siendo la zona de desove y hábitat de especies de peces pelágicos de relevancia comercial (Shipp 1999; Rabalais et al. 1999). No obstante, el GDM es susceptible al cambio climático y presiones antropogénicas, como demuestra el hecho que años de intenso desarrollo y explotación han resultado en cambios significativos de los recursos pesqueros. Los serios problemas que está afrontando el GDM, incluyendo modificación de hábitat, contaminación marina y sobre-explotación, están generando impactos que todavía no están determinados en el gran ecosistema marino que constituye el GDM. La región se caracteriza por una compleja y altamente variable circulación oceánica, con una intensa actividad de mesoescala que está dominada por dos estructuras principales: la Corriente de Lazo (CL) y los anillos desprendidos por la CL. Estas poderosas estructuras oceánicas transportan anomalías en las propiedades físicas, biológicas y químicas de la región, y afectan –ya sea directamente como indirectamente a través de estructuras secundariasprácticamente a cada aspecto de la oceanografía del Golfo. Es por estos motivos que la investigación propuesta se enfoca en la monitorización satelital de la variabilidad 118 Chapter VII espacio-temporal de la CL y anillos, así como en la regulación de las distribuciones de larvas de algunas especies llevada a cabo por estructuras oceánicas de mesoescala. Varios estudios muestran que las condiciones físicas y biológicas tanto para las larvas como para los peces adultos en el GDM exhiben una elevada variabilidad espacial y temporal (Müller-Karger et al. 1991), la cual está probablemente ligada a los patrones de desove (Ortner et al. 1984; Bakun 2006). La variabilidad en las abundancia larvarias de algunas especies pelágicas ha sido ligada en recientes estudios a parámetros ambientales –tales como la temperatura del agua, la salinidad, la profundidad del agua, y la duración de la luz del díay la distribución de plancton en la región del GDM (Muhling et al. 2010; Richardson et al. 2010). Dado que algunos ecosistemas son susceptibles a ser altamente afectados por cambios en las condiciones ambientales y en la distribución del plancton (Teo and Block 2010), la variabilidad de las propiedades oceánicas y las estructuras de mesoescala en el GDM se espera que tengan un efecto directo en los ecosistemas de esta región. En particular, recientes estudios muestran que un conocimiento detallado de la variabilidad temporal y espacial de las estructuras oceánicas de mesoescala al este del GDM es fundamental para entender las condiciones ambientales que influyen en las distribuciones de larvas de diversas especies de peces, sus lugares de desove, crecimiento larvario y consecuente variabilidad en la supervivencia larvaria y juvenil (Richards et al. 1993; Bakun 2006). Sin embargo, parece que existe una falta de información sobre cómo los ecosistemas responden a patrones de variabilidad a corto y largo plazo en la regiones, lo cual puede ayudar adaptarse a posibles presiones de cambio climático en el GDM. De acuerdo con todo lo expuesto, el principal objetivo de esta tesis es investigar, evaluar y analizar la conexión entre los recursos pesqueros y la variabilidad temporal y espacial de las propiedades oceánicas asociadas a la actividad de mesoescala en el GDM. A este respecto, este trabajo proporciona una mejor descripción de la dinámica de mesoescala de la CL y el campo de anillos asociado, y arroja luz sobre la regulación física de la distribución de algunas especies de larvas en el GDM. Resumen en español (Spanish summary) 125 3.3.2 Circulación general La componente más prominente y energética de la circulación general del GDM es la Corriente de Lazo, los remolinos de mesoescala, y los anillos que se han desprendido de la corriente. Estos tres componentes y la formación de un anillo anticiclónico por el estrangulamiento de la CL se pueden observar claramente en la Figura 7.5. Figure 7.5. Temperatura superficial del mar (SST) el 25 de abril de 2011, mostrando la Corriente de Lazo (CL) extendida en el Golfo, y un caliente anillo anticiclónico formándose a partir de la estrangulación de la corriente. Nótese la aparición de remolinos ciclónicos a lo largo de los márgenes exteriores de la CL, en particular dos ciclones localizados en el “cuello de botella” de la corriente (aproximadamente centrados en 23°N, 88°W y 24.5°N, 85.5°W). Los datos fueron obtenidos del Advanced Very High Resolution Radiometer (AVHRR), disponible con una resolución de 2 días en una malla de 18 km. Las flechas negras en el fondo de la imagen representan corrientes geostróficas derivadas de datos satelitales. Anillos anticiclónicos se desprenden de la CL en el este del GDM a intervalos irregulares (Vukovich 1988a; Sturges and Leben 2000; Leben 2005) en el rango de 4 a 18 meses (Sturges and Leben 2000). Llegados a este punto se debe distinguir bien entre remolinos de mesoescala y anillos. Aunque normalmente hallamos en la literatura que a los 126 Chapter VII anillos de la CL se les denomina indistintamente anillos o remolinos, es conveniente especificar que remolinos y anillos son estructuras diferentes. Justo después de su formación, estas estructuras son anillos, con poca o ninguna vorticidad relativa en su centro, mientras que más tarde el centro gana suficiente velocidad angular (a través de difusión radial del anillo) como para hablar propiamente de un remolino (Auladell et al. 2010). El anillo de la CL normalmente representa una fracción que se ha separado de la corriente principal tras describir un bucle o lazo, y es por ello que durante sus estadios iniciales –antes de que la difusión tenga lugar– el anillo está compuesto de una región frontal que retiene la intensa corriente del lazo en sus bordes, y una región central en su núcleo donde las aguas están estancadas. Por otro lado, mientras los remolinos de mesoescala muestran radios comparables al radio de deformación Rossby (~40 km, (Chelton et al. 1998)), el radio de los anillos es mucho más grande (~150 km, (Oey et al. 2005)). Un modelo simple de difusión y advección con un coeficiente de difusión efectivo (Sangrà et al. 2007) es apropiado para simular aproximadamente la evolución temporal de la velocidad angular de remolinos que comienzan con una tipo de rotación de sólido rígido (Auladell et al. 2010). La ecuación de difusión y advección para estudiar la evolución de la velocidad angular, que se deriva de las ecuaciones de momento en coordenadas cilíndricas bajo las aproximaciones del plano-� y simetría axial, es la siguiente: !" !" +�! !" !" =! ! !!(!"#) !!! (16) donde � es la velocidad angular, �!es la velocidad radial, � es un coeficiente de difusión horizontal, � es el tiempo, y � la posición radial. En base a lo expuesto anteriormente y a la naturaleza de su generación, rotación, robustez y su larga vida (Fuglister 1971; Olson 1991), los anillos anticiclónicos de la CL serán llamados anillos en esta tesis, y no remolinos. Estos grandes anillos desprendidos por la CL tienen radios de aproximadamente 150 km, velocidades de giro de 1.8-2 m s-1, alrededor de 800 m de profundidad (similarmente a la CL), se trasladan hacia el oeste en el GDM en una escala de cien a mil kilómetros y tienen un periodo de vida que puede variar de meses a años (Oey et al. 2005). En su trayecto hacia el oeste, los anillos transportan anomalías en las propiedades físicas, biológicas y químicas, afectando así a cada aspecto de la circulación del GDM. Una vez los anillos se están propagando hacia el oeste, estructuras ciclónicas pueden ocasionalmente partir los anillos dividiéndolos en remolinos más pequeños (Biggs et al. 1996). Resumen en español (Spanish summary) 127 El proceso de desprendimiento puede tomar entre días y varias semanas y a menudo, después de que el anillo se ha separado, se vuelve a unir a la CL (Sturges et al. 1993). La frecuencia de separación de anillos ha sido estudiada por varios autores (Maul and Vukovich 1993; Sturges and Leben 2000; Leben 2005), y se ha establecido que el intervalo de tiempo entre desprendimiento de anillos varia de 6 a 17 meses (Molinari, 1980), con periodos primarios de 6 a 11 meses (Sturges and Leben 2000). Varios procesos participan en el mecanismo de desprendimiento de anillos. Un mecanismo generalizado de desprendimiento, parcial o completo, también puede ser explicado mediante el rol que juegan pares de remolinos ciclónicos uno a cada lado de la CL juegan un papel clave. El desprendimiento puede ocurrir después de que una extendida CL sea estrangulada por remolinos ciclónicos, aunque no todo estrangulamiento está seguido por un desprendimiento (Schmitz et al. 2005). La primera descripción observacional fue publicada hace 40 años cuando se observaron ciclones periféricos en los límites de la CL (Cochrane 1972). A esto se le denominó el primer modo de desprendimiento. Otra visión del desprendimiento se observa cuendo la CL adquiere una orientación este-oeste en su extremo más superior (Schmitz et al. 2005). Este proceso tiende a ocurrir después de un primer desprendimiento y se le llama segundo modo de desprendimiento. Una combinación de estos dos modos también puede ocurrir, englobando ambos modos de desprendimiento al mismo tiempo (Schmitz 2005). Una lista exaustiva de estudios previos muy relevantes con respecto a los elementos que participan en el proceso de desprendimiento de anillos de la CL puede encontrarse en las referencias y están descritos en mayor detalle en el Capítulo II de esta tesis (Cochrane 1972; Elliott 1982; Fratantoni et al. 1998; Molinari et al. 1978; Sturges and Leben 2000; Sturges et al. 1993; Vukovich 1988; Vukovich and Maul 1985; Zavala-Hidalgo et al. 2003; Zavala-Hidalgo et al. 2006). 128 Chapter VII 3.3.2 ¿Hay un único mecanismo para las intrusiones de la Corriente de Lazo y los desprendimientos de anillos? En esta sección se muestran los mecanismos físicos que pueden liderar las intrusiones de la CL, disparar el desprendimiento de anillos, y establecer el estado inicial después de una separación de anillo. Cuando discutimos las estructuras de circulación oceánica es relevante tener clara la idea del radio de deformación de Rossby baroclínico (o interno), el cual juega un papel importante en la teoría de la circulación oceánica a gran escala puesto que describe las escalas horizontales de los procesos de mesoescala y representa una medida importante de la dinámica oceánica. El radio de deformación de Rossby baroclínico es el ratio entre las velocidades de fase de las ondas internas con respecto al parámetro de Coriolis, y fundamentalmente indica la escala horizontal (longitud) a la que los efectos de rotación llegan a ser tan importantes como las ondas de gravedad (flotabilidad). Los cálculos para el GDM dan un primer radio de deformación de Rossby baroclínico de aproximadamente 40 km (Chelton et al. 1998). El mecanismo que explica el comportamiento de la CL y el desprendimiento de anillos ha sido ampliamente estudiado, especialmente por Hurlburt y Thompson (1980) en su clásico artículo de modelización (de ahora en adelante HT), y más recientemente ha sido interpretado utilizando la idea del “momentum imbalance paradox” de Pichevin y Nof (1997)6 (de ahora en adelante PN) (ver también Nof y Pichevin, 2001; Nof 2005). Dichos estudios muestran que el desprendimiento de anillos puede ser capturado por un modelo de gravedad reducida de 1.5 capas sin necesidad de considerar la interacción con la topografía del fondo marino. En particular, el artículo de PN tiene un especial interés ya que analiza las consecuencias de un flujo Q que se mueve hacia el norte por un canal estrecho (i.e., de una anchura comparable al radio Rossby o inferior) y que desemboca en un océano abierto (Figura 7.6). 6 ver también (Nof y Pichevin, 2001) y (Nof, 2005) Resumen en español (Spanish summary) 129 Figura 7.6. Diagrama que ilustra el “momentum imbalance paradox”. Un canal abierto hacia el norte por el que fluye agua de densidad � se vacía en un océano estanco con densidad (�+∆�). A lo largo del frente se asume que el espesor es h=0. Se asume que las líneas de corriente en el canal permanecen paralelas a las paredes del canal hasta que se alcanza la línea horizontal de la desembocadura (i.e., sección AB). La hipotética configuración de estado estacionario no es posible en plano-� ni el plano-�. En el escenario de PN un flujo continuo no puede existir debido a que el momento del flujo de corriente no está en equilibrio. La visión tradicional de este fenómeno era que, debido a la fuerza de Coriolis, una corriente entrante como la mostrada gira hacia la derecha (mirando hacia fuera de la costa o canal) y forma una corriente de frontera (o limítrofe) occidental que fluye hacia el este. En este escenario, un frente (correspondiente a la interfase superficial) separa las aguas oceánicas y las aguas anómalas (las entrantes) (Figura 7.6). Al integrar la ecuación del momento en el eje x sobre un dominio rectangular exterior al flujo (área FEDC en la Figura 7.6), PN muestran que el momento integrado que se ejerce en el dominio por el aguan que sale del dominio a la derecha no puede ser equilibrada en un estado estacionario (lo que da una fuerza de flujo a lo largo de la costa no equilibrada). Esta paradoja en el balance del momento, “momentum imbalance paradox” como la llamaron los autores, puede resolverse si se permite que anillos se vayan desprendiendo hacia la izquierda (plano-�, Figura 7.7a) o bien si el flujo crece ininterrumpidamente (plano-�, Figura 7.7b). 130 Chapter VII Figura 7.7. Diagramas que muestran (a) la resolución de PN del “momentum imbalance paradox” y (b) los contornos flujo del plano-�. A causa de la paradoja esbozada en la Figura 7.6, el flujo de la corriente entrante no está en equilibrio. Como resultado, hay dos posibilidades. En un plano-� (a), el crecimiento del anillo anticiclónico es finalmente detenido y, consecuentemente, se desprenden anillos periódicamente hacia la izquierda. A través del efecto �, estos remolinos se ven forzados a propagarse hacia la izquierda. Pichevin y Nof (1997) obtienen su solución analítica mediante la identificación del momento del flujo a través de EF con el momento del flujo a través de CD. (b) Por el contrario, en un plano-�), el remolino crece eternamente y el flujo de masa de corriente que sale es menor que el flujo de masa de corriente entrante. Resumen en español (Spanish summary) 131 En resumen, el modelo de HT puede ser simple pero captura remarcablemente bien las características generales de la variabilidad de la CL: su extensión (crecimiento), el desprendimiento de los anillos, y las retiradas de la CL. Los periodos de desprendimientos predichos por PN (~10 meses) están en concordancia con los periodos predichos por HT (~12 meses). Además, las predicciones de PN con respecto al tamaño del anillo y la velocidad están de muy de acuerdo con las observaciones. El modelo también ayuda a desmentir duradera idea falsa de que la CL desprende remolinos en respuesta a una casi anual variación del flujo entrante a través del canal de Yucatán. El océano real funciona de una manera curiosa y compleja: los periodos de anillos observados cubren un rango de tiempo más extenso (de 3 a 18.5 meses); la CL puede extenderse y retirarse (escalas de ~ meses y 100 km) sin necesariamente desprender anillo alguno; un anillo puede temporalmente desprenderse para luego volverse a unir a la CL (escalas de tiempo de ~ semanas); meandros frontales y remolinos pueden desarrollarse e influir en el desprendimiento de los anillos; la plataforma y la vertiente del Yucatán, la capa profunda y las influencias del Caribe pueden ser significativas, etc. Los modelos presentes y futuros tienen el reto de capturar algunas de estas complicaciones. 3.3.3 Consideraciones aguas arriba El flujo neto que entra en el GDM es de origen del Atlántico Sur en aproximadamente un 40% (Schmitz and Richardson 1991b), y existe también una contribución considerable de agua profunda, mayoritariamente aguas intermedias Antárticas (Schmitz and Richardson 1991b; Schmitz Jr 2004). El otro componente importante de aguas entrantes en el Mar del Caribe y luego en el GDM a través de la Corriente de Lazo tiene su origen en el Atlántico Norte (Johns et al. 2002; Hamilton et al. 2005). Los dos pasajes que dan entrada y salida al transporte de masa en el GDM son el canal de Yucatán (entre Yucatán y Cuba) y el estrecho de la Florida (entre Cuba y la Florida). Asumiendo incompresibilidad en la ley de conservación de la masa, la ecuación del balance de volumen en el GDM puede escribirse como 132 Chapter VII !!!"# !" =� !+� !+�+(�−�) (17) donde � !"#es el volumen total de agua en el GDM, � ! es el volumen transportado a través del canal de Yucatán, � ! es el volumen transportado a través del estrecho de la Florida, � es la descarga de los ríos y (�−�) es el volumen transportado debido a la precipitación menos evaporación. Estimaciones de �+(�−�) por Etter (1983) muestran que el término es un 0.1% de � ! y � !, mientras que estimaciones satelitales muestran que !!!"# !" también es un término pequeño (Bunge et al. 2002) con lo que los grandes términos en la ecuación (17) (� ! y � !) están prácticamente en equilibrio. Esto significa que el transporte hacia el GDM a través de la Corriente de Yucatán está en equilibrio con el transporte a través del estrecho de la Florida. 3.4 Biología marina y pescaderías en el Golfo de México Los organismos planctónicos son cualquier organismo errante (sea animal, plantas, arqueas o bacterias) que habita en la zona pelágica de los océanos, mares, o aguas dulces. Estas agregaciones se encuentran influidas por las características biológicas de dichos organismos y por las características físicas de su hábitat (Ortner et al. 1978; Owen 1981; Mackas et al. 1985). Entre los organismos planctónicos, el ictioplancton se refiere a los huevos y larvas de peces. Aunque normalmente son considerados partículas inertes, las larvas de peces son componentes interactivos del ecosistema (Cowen 2002; Fuiman 2002). La mayoría de los peces tienen un estadio pelágico, el cual varia de semanas a meses (Brothers et al. 1983; Victor 1986) y durante el cual importantes cambios ocurren en cortos intervalos de tiempo, como por ejemplo un importante aumento de la biomasa, con un aumento de hasta 5 órdenes de magnitud (Werner and Gilliam 1984; Houde 1987; Miller et al. 1988). Este incremento de masa durante el periodo larvario evidencia la importancia de este estadio como un modulador del reclutamiento (Houde 1987). De esta manera, es importante estudiar el ictioplancton ya que las abundancias de huevos y la supervivencia de larvas es un indicador del tamaño de la población de adultos. Al ser usualmente más económico muestrear huevos y larvas que adultos, determinar la abundancia de ictioplancton deviene esencial. Resumen en español (Spanish summary) 133 3.4.1 Niveles biológicos y factores limitantes en el Golfo de México Mientras que algunas formas de plancton son capaces de moverse independientemente y pueden nadar cientos de metros verticalmente en un día –un comportamiento llamado migración vertical diurna– su posición horizontal está principalmente determinada por las corrientes secundarias. Los organismos planctónicos se dividen principalmente en tres grupos tróficos: fitoplancton, zooplancton y bacterioplancton. El fitoplancton está compuesto por organismos autótrofos, algas procariotas o eucariotas que viven cerca de la superficie del agua donde hay suficientemente luz como para permitir la fotosíntesis. Aunque el fitoplancton es demasiado pequeño para ser observado a simple vista, cuando se presenta en número suficientemente elevado puede aparecer como una decoloración del agua debido a la presencia de clorofila en las células del fitoplancton. Son responsables de la producción primaria la creación de compuestos orgánicos a partir del dióxido de carbono disuelto en el agua, y son parte de un proceso que sostiene la cadena alimenticia acuática (Peters and Marrasé 2000). El Golfo de México ha sido habitualmente descrito como un sistema oligotrófico (Ortner et al. 1984). Sin embargo la combinación de datos satelitales con observaciones in situ (Muller-Karger et al. 1991; Gilbert et al. 1996; Gilbes et al. 1996, 2002; Muller-Karger and Fuentes-Yaco 2000; Biggs and Ressler 2001; Belabbassii et al. 2005; Biggs et al. 2008) han ayudado a demostrar que el GDM experimenta de intermedias a altas concentraciones de fitoplancton tanto en la plataforma como en ciertas zonas no costeras. En las zonas costeras la variabilidad en el fitoplancton del GDM está relacionada con afloramientos derivados del viento (Chuang et al. 1982; Schroeder et al. 1987; Yang and Weisberg 1999; Muller-Karger and Fuentes-Yaco 2000; Weisberg et al. 2000) y plumas de río (Gilbes et al. 1996; Walker 1996; Del Castillo et al. 2000; Hu et al. 2003). En aguas más profundas, las variaciones de clorofila-a están afectadas por la mezcla convectiva estacional, la divergencia y convergencia asociada a remolinos ciclónicos y anticiclónicos (Biggs and Muller-Karger 1994) y la incorporación de aguas dispersadas de efluentes de ríos (Biggs et al. 2008). La elevada variabilidad espacial de concentración de pigmentos en el GDM aumenta de este a oeste y de sur a norte debido a la diferencia en la estratificación térmica en la región, la intrusión de aguas pobres en nutrientes desde el Mar del Caribe occidental, los acontecimientos de frentes 134 Chapter VII frios y sistemas tropicales de baja presión, y la distribución de nutrientes del río Misisipi (Melo-González et al. 2000). El zooplancton está compuesto por pequeños protozoos o metazoos que se alimentan de bacterioplancton, fitoplancton, otro zooplancton e incluso detritus. Como resultado, encontramos zooplancton en aguas superficiales donde hay recursos alimenticios en abundancia. Como ocurre con cualquier especie, también pueden estar limitados geográficamente. El factor físicos que influye principalmente en la distribución del zooplancton es básicamente la mezcla en la columna de agua (surgencia y contrasurgencia) a lo largo de la costa y en el océano profundo que afectan tanto a la disponibilidad de nutrientes como a la producción de fitoplancton (Lalli and Parsons 1997). A través del consumo y procesamiento de fitoplancton y otros recursos, el zooplancton juega un papel muy importante en la redes acuáticas de alimentos, como un recurso para consumidores de niveles tróficos superiores, incluyendo los peces. Así, el zooplancton es la presa inicial para casi todas las larvas de peces tan pronto cambian su saco vitelino por la alimentación externa, y es considerado como el principal alimento de las larvas de peces (Turner 1984; Govoni et al. 2010). Estudios de la distribución vertical de zooplancton en el este del Golfo de México revelan que la comunidad de zooplancton es diversa, 21 géneros que exceden el 1% de la biomasa de la capa de 0 a 1.000 m (Hopkins 1982). Con referencia a la variabilidad de zooplancton en el GDM, varios análisis de la composición de la comunidad de zooplancton y de la abundancia en la desembocadura del río Misisipi y en similares profundidades fuera de la influencia fluvial al oeste de la Florida y en el Golfo central, muestran que las abundancias de copépodos son significativamente más elevadas en zonas cercanas a la costa que en zonas centrales del GDM (Ortner et al. 1989). Como se observa en esta sección, el fitoplancton es determinante para varios niveles biológicos. Ahora cabe preguntarnos, cuál es el factor limitante del éxito del fitoplancton? El factor dominante en la limitación del crecimiento del fitoplancton y zooplancton varía de región en región en los océanos. A escala global, el crecimiento de fitoplancton en los oligotróficos giros tropicales y subtropicales está generalmente limitado por el aporte de nutrientes, mientras que la luz a menudo limita el crecimiento de fitoplancton en giros subárticos. La fotosíntesis depende tanto de la disponibilidad de luz (en la zona fótica) como de la disponibilidad de nitrógeno (N), y de casi todo el fósforo (P) y silicio (Si) para generar alimento y promover el crecimiento y reproducción. En el GDM, la profundidad de la capa de mezcla –que comprende la limitación de luz y la disponibilidad de nutrientesse ha sugerido