Using a Seabird Top Predator to Assess the Adequacy of the Berlengas Marine

2014 DEPARTAMENTO DE CIÊNCIAS DA VIDA FACULDADE DE CIÊNCIAS E TECNOLOGIA UNIVERSIDADE DE COIMBRA

Using a seabird top predator to assess the adequacy of the Berlengas Marine

Protected Area

adequacy of adequacy

the Berlengas Marine Protected Area Berlengas the Using a seabird predator top a the assess to Using

Joana Sofia Costa Neves Pais de Faria

2014 JoanaFaria de Pais

DEPARTAMENTO DE CIÊNCIAS DA VIDA FACULDADE DE CIÊNCIAS E TECNOLOGIA UNIVERSIDADE DE COIMBRA

Using a seabird top predator to assess the adequacy of the Berlengas Marine Protected Area

Dissertação apresentada à Universidade de Coimbra para cumprimento dos requisitos necessários à obtenção do grau de Mestre em Ecologia, realizada sob a orientação científica do Professor Doutor Jaime Albino Ramos (Universidade de Coimbra) e do Doutor Vítor Hugo Paiva (Universidade de Coimbra).

Joana Sofia Costa Neves Pais de Faria 2014

Acknowledgements First and foremost, I would like to express my sincere gratitude to my dedicated supervisors Professor Jaime Albino Ramos and Vítor Hugo Paiva for all the guidance, support, patience, reviews and of course for all the moments of good disposition. I want to particular highlight all the great technical support provided by Vítor in all the technology and software that was new to me, thank you for your precious help. It is with a smile that I remember all the days passed in Berlenga island, where the wild beauty, distant from all the civilization’s noise make possible to any visitor to forget all the world preoccupations and absorb its peaceful beauty. But even better is to remind all the great persons that I met in that island, thank you Elisabete, Mourato, Paulo for receiving us so well and always with a smile. Also thank you Rahel for being my travel companion and for capturing on photo all the great moments of fieldwork, joy and laugh. A truly thank you to all my friends and family, especially to my mother for always providing everything she had, only to make her beloved children happy. (“Querida mãe, estarei eternamente grata por tudo o que fizeste e continuas a fazer por mim e por todo o imenso amor que só uma mãe pode dar!”). To my dear brother, I want to thank you for always being there for me and for being my “writing coach”, always giving me strength and motivation for keep going with the writing work. And last, but surely not least, I want to thank to my “Soulmate” for all the love and support, for always being there for me, believing in me and for the encouraging me in the difficult moments. I hope we can have a wonderful happy life together. Off course, I also want to thank his great family for always having me treated like a family member and for all the love and affection.

Table of contents Abstract ...... 5 Resumo ...... 7 List of Tables ...... 9 List of Figures ...... 10 Chapter 1 - Introduction ...... 11 Chapter 2 - Methods ...... 17 2.1. Study area ...... 18 2.1.1. The marine surroundings of Berlenga Island ...... 21 2.2. Study species ...... 23 2.3. GPS-loggers: deployment and specifications ...... 24 2.4. GPS tracking data analyses...... 25 2.4.1. Trip filtering ...... 25 2.4.2. Kernel analysis ...... 26 2.5. Environmental data ...... 27 2.6. Species distribution modelling (SDM) ...... 29 2.7. The Zonation algorithm ...... 31 2.8. Data analysis ...... 31 Chapter 3 – Results ...... 33 3.1. Seasonal and inter-annual differences in foraging patterns ...... 34 3.2. Species Distribution Modelling (SDM) of a pelagic marine predator...... 38 3.3. Overlap between usage of at-sea areas by birds and the Berlengas mIBa and MPA areas ...... 43 3.4. Conservation aspects ...... 43 Chapter 4 – Discussion ...... 45 4.1. Inter-annual and within season foraging patterns ...... 46 4.2. Habitat use and foraging segregation ...... 48 4.3. Management considerations ...... 50 References ...... 52

Abstract Marine ecosystems have been facing an increasing number of threats such as unsustainable exploitation of marine resources (e.g. overfishing), degradation and loss of marine habitats, pollution, invasive species and climate change. A possible way to diminish the impact of such threats is through the establishment of Marine Protected Areas (MPAs), which play an important role in biodiversity conservation and contribute to restock the entire marine environment. However the percentage of ocean that are already protected (2.8%) is alarmingly low considering one of the targets set by the Convention on Biological Diversity (CBD): to protect at least 10% of coastal and marine areas until 2020. Berlengas archipelago surrounding waters have great conservation value due to features that congregate a high marine productivity, allowing the survival and maintenance of several seabird species, some of them endangered (e.g. the Balearic shearwater Puffinus mauritanicus). Although this important area is currently under protection, the boundaries of the Berlengas Marine Protected Area (MPA) do not overlap fully with the Berlengas marine Important Bird Area (mIBA), which was designated based on the areas with great importance to the small range local biodiversity. The present boundaries of the Berlengas MPA were defined with the contribution of tracking information of the most important foraging areas of a wide range seabird top predator, the Cory’s shearwater Calonectris diomedea, mostly during the chick-rearing phases of 2005 – 2008. Yet, the boundaries of the Berlengas mIBA and MPA have not been assessed, using tracking information for the whole breeding cycle of this species during several years (in order to take into account inter-annual environmental stochasticity). The aim of this thesis was to use tracking data of a seabird top predator to define boundaries of key foraging grounds and assess their adequacy on the establishment of a network of Marine Protected Areas within Mainland Portugal. More specifically, this work intended to answer the following questions: (1) Which factors (environmental or Human-related) drive the foraging distribution of Cory’s shearwater along the breeding period and across years? (2) Are the current boundaries (from both Berlengas mIBA and Berlengas MPA areas) protecting all key foraging grounds of this species? To address these questions Global Positioning System (GPS) tracking devices were attached to the back feathers of Cory’s shearwater in order to identify the foraging areas used by this species. Species Distribution Models (SDM; MaxEnt) were performed using foraging tracking data combined with environmental

5 data in order to understand which environmental variables trigger the foraging distribution of the species. To better access the adequacy of the current Berlengas mIBA and MPA boundaries it was used the conservation planning software Zonation. This software can be used to design and evaluate Marine Protected Areas by producing different scenarios depending on specification of conservational priorities. Results showed very distinct foraging patterns in 2010 and 2012 comparing with 2011 and 2013 and also distinct foraging trip characteristics, with trips closer to the colony during chick-rearing periods. These annual differences may be related with climatic variation, as reflected by the North Atlantic Oscillation (NAO) index, with the highest positive index value in 2012 (considering our study period) and the lowest negative index value in 2010. During the chick-rearing period, birds showed a typical central-place foraging behaviour, as a consequence of the constant need to return to the colony to feed their chick. Therefore, areas in the vicinity of the colony are more likely to be used during chick-rearing than during pre-laying and incubation periods. Overall, birds mostly exploited productive areas above the continental shelf around the colony and along the Portuguese coast, as well as more distant productivity areas near known banks and seamounts of the north Atlantic region. Results also allowed to conclude that the actual Berlengas MPA is only protecting 25.9% of the most relevant foraging region for Cory’s shearwater. While the same foraging region overlapped 45.5% with the Berlengas mIBA. A considerably higher overlapped area (59.6%) was obtained when comparing with new proposed Marine Protected Areas, under evaluation by the Portuguese government. The implementation of these new areas would protect the most relevant foraging areas of Cory’s shearwater. This is because the implementation of these new areas would generate connectivity between the main areas north and south of the actual Berlengas MPA, thus providing an ecological corridor for Cory’s shearwaters and other marine taxa. Keywords: Marine Protected Area, Species Distribution Modelling, spatial ecology, foraging strategies, Cory’s shearwaters.

Resumo Os ecossistemas marinhos têm vindo a enfrentar um aumento no número de ameaças como a exploração insustentável dos recursos marinhos (por exemplo a pesca excessiva), degradação e perda de habitats marinhos, efeitos da poluição, da introdução de espécies invasoras e efeitos das alterações climáticas. Uma possível forma de diminuir tais impactos é através da implementação de Áreas Marinhas Protegidas (Marine Protected Areas, MPAs), que desempenham um papel importante na conservação da biodiversidade e contribuem para a recuperação do ambiente marinho. No entanto a percentagem do oceano que está actualmente sob protecção (2.8%) é bastante baixo considerando um dos objectivos estabelecidos pela CBD (Convention on Biological Diversity): proteger pelo menos 10% da costa e áreas marinhas até 2020. As águas circundantes ao arquipélago das Berlengas têm um grande valor de conservação devido às suas características que levam à agregação de grande produtividade marinha, permitindo a subsistência de várias aves marinhas, algumas delas em perigo de extinção (como a Pardela-Balear Puffinus mauritanicus). Embora esta área esteja actualmente sob protecção, os limites da Área Marinha Protegida (MPA) não se sobrepõem totalmente à área definida como importante para as aves marinhas (marine Important Bird Area, mIBA) que foi estabelecida com base apenas na biodiversidade local. Os limites actuais da MPA das Berlengas foram definidos com a contribuição de informação de rastreamento das áreas de procura de alimento mais importantes para uma ave marinha predadora de topo com uma ampla distribuição, a Cagarra Calonectris diomedea, sobretudo durante a fase de alimentação às crias entre 2005 e 2008. Contudo, os limites da mIBA e MPA das Berlengas não foram avaliados usando informação de rastreamento de todo o ciclo reprodutivo da espécie durante vários anos (para englobar a estocasticidade ambiental inter-anual). O objectivo desta tese foi usar dados de rastreamento de uma ave marinha predadora de topo para definir importantes áreas de procura de alimento e avaliar a sua adequabilidade no estabelecimento de uma rede de Áreas Marinhas Protegidas ao longo de Portugal Continental. Mais especificamente este trabalho pretende responder às seguintes questões: (1) Que factores (ambientais ou antropogénicos) influenciam a distribuição de procura de alimento da Cagarra ao longo do período de reprodução e através dos anos? (2) Estarão os actuais limites (da mIBA e MPA das Berlengas) a proteger todas as importantes áreas de procura de alimento desta espécie? Para responder a estas questões fixaram-se dispositivos de GPS (Global Positioning System) nas penas do dorso das

7 aves para identificar as áreas de procura de alimento usadas pela espécie. Modelos de distribuição de espécies (MaxEnt) foram realizados usando informação de rastreamento e dados ambientais de forma a entender que variáveis ambientais desencadeiam a distribuição de procura de alimento da espécie. Para avaliar a adequabilidade dos limites das actuais mIBA e MPA das Berlengas usou-se o software Zonation, que pode ser usado para o delineamento e avaliação de Áreas Marinhas Protegidas, produzindo diferentes cenários dependendo das especificações das prioridades de conservação. Os resultados demonstraram diferentes padrões de procura de alimento em 2010 e 2012 comparativamente com os anos de 2011 e 2013; e também diferenças nas características das viagens, com viagens mais perto da colónia nos períodos de alimentação das crias. Estas diferenças anuais podem estar relacionadas com a variação climática, como reflectido pelo índice NAO (North Atlantic Oscillation), que mostra valores de índice positivos mais altos em 2012 (considerando o período de tempo de estudo) e valores de índice negativos mais baixos em 2010. Durante o período de alimentação das crias, as aves mostraram um comportamento de procura de alimento designado por central-place foraging, como consequência da constante necessidade de voltar para a colónia para alimentar as crias. Portanto, áreas na periferia da colonia são mais propensas a ser usadas durante o período de alimentação das crias do que durante os períodos de incubação e antes da postura do ovo. De forma geral, as aves exploraram sobretudo áreas produtivas sobre a plataforma continental nos arredores da colónia e áreas ao longo da costa Portuguesa, assim como áreas produtivas mais distantes perto de bancos e montes submarinos na região do Norte Atlântico. Os resultados permitem-nos concluir que a actual MPA está apenas a proteger 25.9% das áreas de procura de alimento mais relevantes para a Cagarra. Enquanto essas áreas de procura de alimento sobrepõem-se 45.5% com a mIBA das Berlengas. Um valor de sobreposição mais elevado (59.6%) é obtido ao comparar com as novas Áreas Marinhas Protegidas propostas que estão actualmente sob a avaliação do governo Português. A implementação das novas Áreas Marinhas Protegidas iria permitir a protecção das áreas de procura de alimento mais relevantes para a Cagarra. Isto porque a sua implementação iria gerar uma conectividade entre a maioria das áreas a norte e sul da actual MPA das Berlengas, providenciando um corredor ecológico para a Cagarra e outros organismos marinhos. Palavras-chave: Áreas Marinhas Protegidas, modelação de distribuição de espécies, ecologia espacial, estratégias de procura de alimento, Cagarra.

List of Tables

Table I. Biologically relevant explanatory variables used for habitat modelling (MaxEnt) and data manipulation and source...... 28

Table II. Characteristics of foraging trips and foraging areas of Cory’ shearwaters between 2010 and 2013, during different phases of their breeding cycle (i.e. pre-laying, incubation and chick-rearing) ...... 35

Table III. Generalized linear mixed model (GLMM) results analysing the effect of year (2010-2013) and breeding phase (pre-laying vs chick-rearing) on different parameters of Cory’s shearwaters foraging trips and foraging areas ...... 36

Table IV. Percentage of kernel overlap between the foraging distributions of birds during each of the nine fieldwork campaigns (2010-2013)...... 37

Table V. Estimates of model fit and relative contributions of the environmental variables to the MaxEnt model for Cory’s shearwaters between 2010-2013, during different phases of their breeding cycle (pre-laying, incubation and chick-rearing) ...... 42

List of Figures

Figure 1. (A) Sea Surface Temperature (SST) in the surroundings of the colony. (B) Representation of the several seamounts and banks near the colony. Also shown are the boundaries of the Berlengas Marine Protected Area (MPA) and marine Important Bird Area (mIBA) ...... 19

Figure 2. Cory’s shearwater resting in the sea surface...... 23

Figure 3. Attachment of mini-GPS loggers to bird’s back feathers ...... 25

Figure 4. Graphical representation of some environmental variables used in Species Distribution Modelling (SDM) ...... 29

Figure 5. Habitat values from MaxEnt models of the distribution of Cory’s shearwaters between 2010-2013, during different phases of their breeding cycle (pre-laying, incubation and chick-rearing) ...... 40

Figure 6. Zonation model results showing the top 5% most relevant foraging region for Cory’s shearwaters between 2010-2013, during different phases of their breeding cycle (pre-laying, incubation and chick-rearing). Also shown in the map, the Berlengas Marine Protected Area (MPA), marine Important Bird Area (mIBA) and the (new) proposed Marine Protected Areas for the entire coastal mainland Portugal region, under evaluation by the Portuguese government ...... 44

Chapter 1 - Introduction

Joana Faria

“The ocean stirs the heart, inspires the imagination and brings eternal joy to the soul.” Wyland

The oceans are very important to the equilibrium of marine and terrestrial ecosystems and provide many ecosystem services for humans. Not only they hold the major part of the planet’s water but they also play important roles in the storage of carbon dioxide, climate regulation and by constituting a great source of productivity and biodiversity, genuinely holding priceless sanctuaries of life (Angel, 1993; Toropova et al., 2010). However, the number and intensity of threats to marine habitats have been increasing, challenging the equilibrium of the marine ecosystems established by several oceanic processes. Major threats affecting the marine ecosystems have been attributed to the unsustainable exploitation of marine resources (e.g. overfishing; Boehlert, 1996; Hilborn et al., 2003), habitat loss and degradation of marine habitats (Dulvy et al., 2003), pollution, invasive species and climate change (Dulvy et al., 2003; Pitcher and Cheung, 2013). Overfishing has contributed to biomass loss from different trophic levels of marine ecosystems since 1950, with predatory species and coastal areas being the most affected by this threat (Tremblay-Boyer et al., 2011). One possible way to diminish the impact of such threat is through the establishment of Marine Protected Areas (MPAs). In such confined regions, fisheries could be limited and thus fish stocks and upper-trophic level species (both aquatic and aerial predators) would have the opportunity to recover from the Human-related stressors negatively affecting them outside those reserves. Thus, marine reserves can function as cradles of biodiversity contributing to restock the entire marine environment. The implementation of protected areas is more advanced in terrestrial than in marine environments (11.5% vs 2.8%, respectively; Toropova et al., 2010; IUCN and UNEP-WCMC, 2013). Although the protection of the oceans still too far away from the 10% protection target of the Convention on Biological Diversity (CBD), the numbers of MPAs has been increasing and progress has been made. Since 2003 to 2010 the number of MPAs has increased from 4116 to 5850, respectively, and the percentage of the oceans that are already protected increased from 0.5% to 1.17% (Chape et al., 2003; Toropova et al., 2010); and finally, the latest official statistics refer a percentage of 2.8% of the global ocean that is already covered by MPAs (IUCN and UNEP-WCMC, 2013). These numbers are alarmingly low considering one of the targets set by the Convention on Biological Diversity (CBD): to protect at least 10% of coastal and marine areas until 2020. The different advances in the implementation of protected areas in terrestrial and marine environments may be due to the additional difficulties of studying marine ecosystems, given the vastness of the sampling area. The lack of

12 resources and other challenges related with the complexity of marine ecosystems and with highly mobile marine species, whose ranges of their entire life history can cover long distances, were already subject of criticism against the implementation of ‘static’ and confined MPAs. Although it is a fact that these challenges constitute additional difficulties to the implementation of MPAs, some authors defend that MPAs can be an effective and feasible instrument to protect marine ecosystems (Game et al., 2009). For instance, instead of protecting the entire range of highly distributed species, it is more feasible to protect critical areas for the survival of marine species, such as the breeding and foraging regions, and migratory routes (Game et al., 2009). For seabirds, which tend to have high colony site fidelity but extended at-sea ranges throughout the year, relatively small MPAs located in critical areas for the species survival was shown to provide good protection (Kenchington, 1990). Because a large majority of seabird species tend to be central place foragers during the breeding season (Orians and Pearson, 1979; Weimerskirch, 2007), commuting between foraging grounds and the breeding colony, their critical feeding areas are usually located around the breeding site. Thus, the protection of rather small areas can be very effective to potentiate the breeding success and survival of seabird species (Pichegru et al., 2010a; 2012). Having this in mind, it is essential to define methods that enable the selection of areas that are fundamental to the survival and maintenance of marine species, especially potential umbrella or keystone species, which are able to represent other levels of the trophic chain in marine ecosystems. Although the resources for the study of marine environment tend to be scarce compared with terrestrial resources, the methods and equipment to study marine species has improved recently. For instance, to complement the data acquired by aerial and ship-based surveys, which are the most traditional methods to study seabirds (Garthe, 2006; Ramirez et al., 2008), the use of tracking systems has improved enormously since the advent of telemetry (Ropert-Coudert and Wilson, 2005) and a wide variety of devices are now available. These improvements give access to behavioural data such movement patterns, distances travelled by the species, foraging and home ranges, or habitat utilization, that were unavailable with the ‘classical’ at-sea surveys (Tremblay et al., 2009). The tracking data that these types of devices provide can be used to identify important feeding areas for marine species, which can thereafter be used for the design of MPAs. For example, this approach was used for the design of the Namibian Islands’ Marine Protected Area (NIMPA; Ludynia et al., 2012) that overlaps with the northern

Benguela Upwelling System off the southern coast of Namibia. The implementation and assessment of the NIMPA boundaries was based on several years (2005-2006, 2008-2010) of seabird tracking data during breeding seasons at several breeding sites along the Namibian coast. After the first NIMPA design, that used tracking data of African penguin (Spheniscus demersus) of the years 2005 and 2006, the adequacy of its boundaries was tested using additional data of African penguin (2008-2010) and Bank comorants (Phalacrocorax neglectus) tracking data of 2008. This additional data of African penguins foraging distribution did not differed much from the previous years, showing similar foraging patters to those of 2005 and 2006. The Bank cormorants foraged also mostly within the NIMPA limits and therefore the authors considered the first NIMPA boundaries (from 2006) as being adequate still in 2010 (Ludynia et al., 2012). This represents one of the few examples in the literature where several years of tracking data were used to design and test the adequacy of an MPA through time. When implementing an MPA, to achieve an effective protection in accordance with the MPA specifically protection aims, after the first main step of delineating the MPA boundaries, the next and equally important step is the assessment and monitoring of the MPA boundaries adequacy and effectiveness through time. As oceans productivity processes are extremely dynamic, the foraging distribution of marine species will vary from year to year. Therefore, acquiring foraging distribution data of the years that succeed the MPA implementation is crucial to evaluate its effectiveness and adequacy. In some cases the positive effects that result from the MPA implementation are rapidly visible. This was the case of a study conducted in South Africa that used tracking data of African penguin to evaluate the effects of fishing exclusion on their foraging behaviour (Pichegru et al., 2010a; 2012). A small area of 20 Km-radius around the birds’ colonies was closed to purse-seine fishing and just after three months of closure the birds decreased their foraging effort by 25-30%, reducing the energy expenditure, and feed more intensely inside the area that was closed to fisheries (i.e. the colony surroundings; Pichegru et al., 2010a; 2012). This is an example that even small protected areas, if criteriously implemented, can provide immediate benefits to the survival and reproduction on top predators. The implementation and assessment of the evolution of the Arrábida Marine Park (AMP) is another example of the positive effects that a marine reserve with no- take zones can provide. This reserve is located on the west coast of Portugal and due to its features, with a rocky substratum in shallow waters and other features resulted from

14 erosion of cliffs that contribute to habitat complexity, it represents an important hotspot of biodiversity. It is also an area that is commonly used for fishing (angling and spearfishing) and leisure activities (Gonçalves et al., 2002). The implementation of this reserve consisted on the designation of different areas with different levels of protection with more restrictions in a fully protected area, where fishing and human access is not allowed and other partially-protected and buffer areas with less restrictions. In order to understand whether the AMP constituted an effective tool in protecting the local biodiversity, an assessment was made based on abundance and biomass of different species, including targeted and non-targeted species by fishing, before and after the implementation of the reserve (Costa et al., 2013). In this evaluation most of the species groups showed higher density and biomass values than before the designation of the AMP, especially those species with commercial value which where intensively exploited before the fishing exclusion (Costa et al., 2013). The first step towards the implementation of Marine Protected Areas is the identification of top priority sites with high biodiversity levels and with ecological importance. However, the dimension and complexity of the oceans biological and physical processes brings additional difficulties, there are several approaches to identify MPA candidate sites. The use of “umbrella” species to assess priority sites and thus to protect several levels of the marine biota has been shown to be an efficient approach (Pichegru et al., 2010a). For instance, marine top predators have been considered good indicators of the marine ecosystems health (Sydeman et al., 2007; Ronconi et al., 2012) as they tend to have a wide distribution that usually overlap with the distribution of other aquatic and aerial top predators (Block et al., 2011). Also their trophic webs comprise several species of other marine taxa, which are commonly commercially valuable species. Therefore, marine top predators such as seabirds, are a good example of umbrella species, which can be used efficiently to identify candidate regions for the implementation of MPAs (Lascelles et al., 2012; Ronconi et al., 2012). Usually the designation of marine Important Bird Areas (mIBAs) precedes the implementation of MPAs. The mIBAs are marine areas considered of international importance for the conservation of bird species in a global level, but because they do not constitute a legal instrument, one of the aims of its designation is to identify areas that constitute excellent candidates for the implementation of Marine Protected Areas. Since 2012, the Berlengas archipelago mIBA was legislated as an MPA (Decree- Law nr. 105/2012, 17th of May). This archipelago and its surrounding waters have great

15 conservation value due to features that congregate a high marine productivity, allowing the survival and maintenance of several seabird species, some of them endangered (e.g. the Balearic shearwater Puffinus mauritanicus; Arcos, 2011). Some of the threats that may affect the avifauna of this area include foraging competition and egg predation by the Yellow-legged Gull (Larus michahellis) of Berlengas population that is ca. 13 000 individuals; disturbance of nesting areas; oil pollution; bycatch of several seabirds in nets and hooks of illegal fishing, including endangered species such as the Balearic shearwater, and even effects of climate change (Cabral et al., 2005; Ramirez et al., 2008; Arcos, 2011). Although this important area is currently under protection, the boundaries of the Berlengas MPA do not overlap fully with the Berlengas marine Important Bird Area, which was designated based on the areas with great importance to the small range local biodiversity. The present boundaries of the Berlengas MPA were defined with the contribution of tracking information of the most important foraging areas of a wide range seabird top predator, the Cory’s shearwater Calonectris diomedea, mostly during the chick-rearing phases of 2005 – 2008 (Ramirez et al., 2008). During the chick-rearing period birds are constrained to return often to the colony in order to feed their chicks, therefore areas in the vicinity of the colony are more likely to be used that areas further away. The boundaries of the Berlengas mIBA and MPA have not been assessed, using a similar method to that used in the NIMPA Marine Reserve of Namibia (Ludynia et al., 2012), nor using tracking information for the whole breeding cycle of this species during several years (in order to take into account annual environmental variation). Therefore our aim is to use tracking data of a seabird top predator to define boundaries of key foraging grounds and assess their adequacy on the establishment of a network of Marine Protected Areas within Mainland Portugal. To achieve this aim we intend to answer the follow questions: (1) Which factors (environmental or Human- related) drive the foraging distribution of Cory’s shearwater along the breeding period and across years? (2) Are the current boundaries (from both Berlengas mIBA and Berlengas MPA areas) protecting all key foraging grounds of this species? This assessment will allow to understand the adequacy of the former Berlengas mIBA and MPA from 2008 until 2013.

Chapter 2 - Methods

Rahel Borrmann

He was not bone and feather but a perfect idea of freedom and flight, limited by nothing at all” Richard Bach, Jonathan Livingston Seagull

2.1. Study area

This study was carried out on Berlenga Island (39°23'N, 9°36'W), the largest island of the Berlengas archipelago which also include Farilhões and Estelas islets. Berlengas is at approximately 10km apart from mainland Portugal, within the Portuguese continental shelf which has a 200km extension. The Berlenga island, that names the archipelago, has a total area of 78.8 ha. This archipelago is an important breeding ground to several seabird species such as Cory’s Shearwater, Calonectris diomedea, with 980-1070 breeding pairs (from those 310 pairs breed in Berlenga Island; Lecoq et al., 2011). Madeiran Storm-petrel, Oceanodroma castro, with 473 individuals breeding on Farilhão Grande (Mendes, 2013) and 82 breeding pairs of European Shag, Phalacrocorax aristotelis (Lecoq et al., 2012). Common Guillemots Uria aalge, had been declining, reaching a population of only 8 individuals in 2005, and presently no longer breeds in the archipelago (Amado, 2007). The most abundant seabird species breeding on the archipelago is the Yellow-legged Gull, Larus michahellis. In 1994 the Berlenga island population of this species reached ca. 44 698 individuals and since that year the population has been subject to population control measures, by culling adults on the first two years and destroying clutches since then (Amado, 2007). The number has decreased to ca. 25 000 individuals in 2008 (Ramirez et al., 2008) and to ca. 13 000 individuals in 2013 (unpublished data, ICNF 2014). At Berlenga island there is also a small population of ca. 30 individuals of the Lesser Black-backed Gull, Larus fuscus (Ramirez et al., 2008). Berlengas archipelago is not only an important breeding ground but its marine area is also important during spring (March) and autumn (September) migrations of the Northern gannet, Morus bassanus, which is commonly seen in flocks of several thousand individuals. There are also observations of a Critically Endangered species, Balearic shearwater, Puffinus mauretanicus, in large flocks passing near Cape Carvoeiro (Ramirez et al., 2008). Nearly 90% of the overall Balearic shearwater population should spent its non-breeding period a bit further north of this area, but is supposed to forage extensively over the Berlengas marine zone (Guilford et al., 2012). The very abundant Yellow-legged Gull on Berlenga island possibly has negative effects on the other seabird populations such as the Cory’s shearwater: Lecoq et al. (2011) registered predation of Cory’s shearwater eggs by L. michahellis at Farilhões island. Other limitation to the presence of seabirds on Berlenga island are the introduced

mammals: Black rats, Rattus rattus, and wild rabbits, Oryctolagus cuniculus (Amado, 2007). It is known that the presence of rats and other introduced mammals can lead to the decrease of population numbers or even to the extinction of other species including seabirds (Towns et al., 2006; Jones et al., 2008). All these seabird populations are supported by the surrounding waters that are rich in productivity hotspots as Ashton, Seine, Josephine, Gorringe seamounts and Galician banks, that are about 300 km (minimum distance) from Berlenga Island (Fig. 1B; Pitcher et al., 2007; Morato et al.,2008).

Figure 1. (A) Sea surface temperature (SST) in the surroundings of the colony which is marked with a star symbol. (B) Representation of the several seamounts and banks near the colony with bathymetry on the background of the figure. Also shown are the boundaries of the Berlengas Marine Protected Area (MPA; dashed area) and marine Important Bird Area (mIBA; thickest line).

In addition to the bathymetric features these surrounding waters also have other characteristics as ocean currents and wind circulation that are favourable to the occurrence of upwelling along the Portuguese coast. As upwelling is characterized by cold waters with lower salt concentration and rich in nutrients that came to surface, the occurrence of this phenomenon can be detected by the presence of filaments of cold water that penetrates into the open ocean, and can be observed by satellite imagery. These cold waters, originated from upwelling occurrence, can reach distances of about

30–50 km from the coast under calm wind conditions and distances of about 100–200 km during and after strong north winds (Fig.1A; Fiúza, 1983; Sousa et al., 2008). This phenomenon occurs most strongly between June and October and its effects are not homogenous along the Portuguese coast, the differences are related to the landings of different types of fish on different parts of the coast, south or north of Lisbon. Landings of bottom fish on the southern coast are almost twice as high as on the northern coast, and landings of pelagic fish on the northern coast are three times higher than those on the southern coast. This can be explained by the different characteristics of each part of the continental shelf. In the north of Lisbon, the shelf is larger and there are more freshwater input due to the presence of more rivers and higher occurrence of rainfall. In the south of Lisbon, the shelf is narrower and the coast is more irregular, with several headlands, and with less freshwater input (Mann and Lazier, 2006). This regular upwelling pattern can be influenced and altered by climate variability, which can be depicted by the North Atlantic Oscillation (NAO) index. Climate indices like NAO have been shown to be very useful in ecology studies, allowing a better visualization of the climatic fluctuations which influence ecological processes, and therefore, the abundance and distribution of species. The NAO refers to a north-south alternation in atmospheric mass between the subtropical Atlantic and the Artic, and thus involves out-of-phase behaviour between the climatological low- pressure center near Iceland and the high-pressure center near Azores (Stenseth et al., 2003). It is possible to know the state of the NAO index by calculating the average atmospheric pressure difference at sea level between the Azores and Iceland for the winter months (December, January, and February; Mann and Lazier, 2006). This alternation can affect upwelling patterns, sea surface temperature and the organisms of marine trophic levels (Santos et al., 2007). Years with positive NAO index are associated with increases in precipitation in northern Europe region, these results from alterations in the westerly winds that become stronger and moved northwards. Therefore, these years are related with the shift of the Atlantic storm activity to the north and drier than average anomalies to the south of Europe, Mediterranean region. With negative values of NAO index, the opposite is expected, a southward shift of the precipitation activity and dry anomalies are expected over northern Europe (Stenseth et al., 2003). The precise effects of the NAO index oscillations between positive and negative values are regionally dependent, therefore very low NAO values, usually associated

20 with extremely intense upwelling and low sea surface temperature, due to stronger winds, are expected to have negative effects in the abundance of plankton and consequently in the following marine trophic levels. The Portuguese coast is an example of a region where the occurrence of an excessively intense upwelling lead to negative effects on the abundance of plankton and pelagic fish. This seems to be caused by climatic and oceanographic conditions that drive the fish larvae and plankton offshore to unfavourable areas, leading to their mortality (Santos et al., 2007; Paiva et al., 2013a).

2.1.1. The marine surroundings of Berlenga Island

Because of its own underwater productivity features, the Berlengas archipelago and the surrounding waters are important for several seabird species, and therefore it was designated as mIBA (marine Important Bird Area). The IBA designation is attributed to areas with international significance to conservation of bird populations, which is based on the presence and abundance of species that occur there seasonally or during the whole year. The IBA Program of the BirdLife International intends to achieve an appropriate form of protection for these areas. Therefore the criteria used are compatible with the principles used to the designation of Marine Protected Areas in the European Community, from the EC Birds Directive. The criteria used to designate an IBA are divided in three groups: important areas at a global level, important areas at European continent level and at European Union level. In all these groups the criteria are based on the presence and abundance of threatened species; the presence and abundance of species with a restricted distribution; the presence of big concentrations of threatened, migratory or gregarious species and finally, the presence of species which are dependent on a certain type of biome (Ramirez et al., 2008). The designation of Berlengas mIBA was based on the following criteria: 1) supports regularly significant numbers of a species that is considered threatened at a global level, Puffinus mauretanicus; 2) supports concentrations of gregarious species and species that are considered threatened in the Union European; 3) supports concentrations of non-threatened migratory species such as Morus bassanus, and 4) it is one of the five most important areas in Europe for Calonectris diomedea considered threatened in the European Union (Ramirez et al., 2008).

The Berlengas mIBA led to the designation of the Berlengas Marine Protected Area, Decree-Law nr. 384 -B/99, 23rd of September, which boundaries were latter expanded by the Decree-Law nr. 105/2012, 17th of May. However the Marine Protected Area established to Berlengas archipelago does not fully overlap with the mIBA boundaries (Fig. 1B). This area is not only an important ground to seabirds, as mention before, but it also encompasses a rich biodiversity of species that belong to other trophic levels of marine ecosystem, such as several species of macro algae that find an rough rocky substratum when they can attach and develop, and also, species of marine lichens (Lichina pygmaea and Verrucaria Maura; Bengala et al., 1997b). On benthic fauna, we can also find diversity of marine invertebrates species, mostly mollusks and crustaceous, distributed along the littoral zones: the gastropod Littorina neritoides on supralittoral zone; the similar species Chthamalus montagui and Chthamalus stellatus on mid-littoral zone, when we can also find the mollusk Mytilus galloprovincialis, which is commonly more abundant in locals with strong water activity, and the Lasaea rubra which can be found from mid-littoral zone until infralittoral zone. Lastly, in the lower part of the mid-littoral zone we can find colonies of the crustaceous Pollicipes pollicipes (Bengala et al., 1997a; Amado 2007). In the open ocean, the waters surrounding this archipelago have great diversity of low and upper trophic level species, as fish and cetaceans diversity. The ichthyofauna of this area is composed by a large number of species including large shoals of sardine species that are a common prey of upper trophic level organisms. Concerning to cetaceans diversity, some observations were made including the following species: Delphinus delphis, Tursiops truncates, Phocoena phocoena, Stenella coeruleoalba, Balaenoptera acutorostrata and Ziphius cavirostris. In deeper waters were also registered observations of several whale species: Grampus griseus, Globicephala melaena, Physeter macrocephalus and Balaenoptera physalus. On more rare occasions were also recorded observations of pinnipeds on the surrounding waters of the archipelago (Amado, 2007).

Pedro Pires

Figure 2. Cory’s shearwater resting in the sea surface.

2.2. Study species

Cory’s shearwater, Calonectris diomedea (Fig. 2), is a colonial, long-distance migrant Procellariiform seabird with an extremely large range. The global population size estimation is between 600 thousands to 1.2 million mature individuals, with Europe holding 75-94% of the global population (BirdLife International, 2014). Although the population has been decreasing, the decline was not sufficiently rapid to be considered as a Vulnerable breeding species in mainland Portugal, according to the population trend criterion (>30% decline over ten years or three generations). Therefore this species was evaluated as Least Concern by the IUCN conservation status. This species has a wide breeding and wintering distribution and uses a trans- equatorial migration to move between non-breeding and breeding areas. During the non- breeding season (December-February) this species migrates to productive areas located in the South Atlantic which are associated with the Benguela, Agulhas, Brazil and Canary Currents (Gonzalez-Solis et al., 2007; Ramos et al., 2012). During the breeding season (March-October) Cory’s shearwater migrate and breed in the North Atlantic archipelagos such as Berlengas, Azores, Madeira and Canary archipelagos (Navarro et al., 2007; Paiva et al., 2010b). After arriving at their breeding grounds, they start to select and defend their nests. They present high levels of philopatry, and thus tend to maintain the same nest of the previous year. The couples are also usually faithful for life. After copulation females embark on a pre-laying exodus to build up body reserves

23 and form one single egg. This exodus typically lasts for about 20 days (Jouanin et al., 2001). After return, females lay their egg at the end of May, which represents the beginning of the incubation period that lasts approximately two months and is shared by both parents (Paiva et al., 2010b). The chick-rearing period (starting at the end of July) is also shared between the two parents and is a period of high energetic demands. To respond to these high energetic needs, individuals use a dual-foraging strategy, where they alternate between short foraging trips, to search food for chicks, and long foraging trips to fill up their own body reserves (Magalhães et al., 2008; Paiva et al., 2010a; 2010b). Cory’s shearwater can travel long distances to find productive areas and feed mainly on epi - and - mesopelagic fish and also on squid (Granadeiro et al., 1998; Xavier et al., 2011). As Cory’s shearwater are widely distributed throughout the North and South Atlantic Ocean along the year, and also because they are generalist surface feeders, their diet is expected to reflect the distribution of the more abundant prey (Granadeiro et al., 1998; Paiva et al., 2013a).

2.3. GPS-loggers: deployment and specifications

Mini-GPS loggers were used in this study as they seem to provide the most accuracy data, with an error of meters (Steiner et al., 2000; Ramirez et al., 2008). The devices were programmed to collect one location every 5 minutes and were covered with long thermo-retractable rubber sleeve, which made them impermeable, preventing damages to the devices when birds dive preying their food. The devices were attached to the bird’s back feathers with Tesa® tape (Fig. 3; Wilson et al., 1997) and the birds were then returned to their nests, this process took less than 10 minutes in order to minimize the overall stress. The devices weigh were below 3% of the bird’s body weight, which is believed to have no adverse effects in the birds foraging behaviour (Phillips et al., 2003; Passos et al., 2010).

Rahel Borrmann Figure 3. Attachment of mini-GPS loggers to bird’s back feathers using Tesa® tape.

2.4. GPS tracking data analyses

2.4.1. Trip filtering

The data provided by GPS-loggers were divided into individual foraging trips using the distance between each GPS position and the location of the breeding colony, and considering that each foraging excursion starts and ends at the breeding location. Because this study aims to interpret the spatial distribution of Cory’s shearwaters in relation to the recently legislated Berlengas MPA, we only used the short foraging excursions (i.e. less than 4 days of duration; Magalhães et al., 2008; Paiva et al., 2010b). Locational points with distance to colony less than 2 Km were excluded to avoid the inclusion of non-foraging activities (e.g. resting at-sea in rafts nearby the breeding colony) related with the colony and to use only travelling related location points. Also in order to exclude location points where the birds were resting and drifting on the sea surface, a filter was applied removing all the locational points with speed < 9 km/h, a threshold selected by analysing the frequency distribution of speed records (Gremillet et al., 2004; Guilford et al., 2008; Paiva et al., 2010c). As described in previous studies, seabirds when acting as central-place foragers, tend to use a commuting type of movement, travelling in a direct flight until reaches an area of interest where it increases its turning rate and slow down its flight speed (Kareiva and Odell, 1987; Weimerskirch, 2007). Therefore the sinuosity index was calculated to detect foraging areas, which correspond to areas with higher sinuosity values. This index can be achieved by dividing the total distance travelled between two locational points by the absolute distance between the first and last point. Thus, if the

25 bird is foraging in circling motions, increasing its turning rate, the total distance travelled will be higher than the absolute distance between the first and last locational point of this trip, resulting in a higher sinuosity index than the observed when the bird is travelling in a direct flight. The foraging areas were selected when sinuosity index > 2.7 which is slightly lower than the used in previous studies (e.g. Gremillet et al., 2004).

2.4.2. Kernel analysis

In order to quantify the habitat utilization and to identify important foraging areas used by the species, a Kernel Utilization Distribution (Kernel UD) analysis was performed using the R adehabitatHR package (Calenge, 2006; R Core Team, 2014). The utilization distribution refers to the distribution of the positional points of an individual over time (Van Winkle, 1975). Whereas the Kernel method can be used for estimation of utilization distribution (Worton, 1989) since it allows the transformation of point patterns, location data, in to probability of distribution, density data. This transformation result in polygons that characterize areas with higher or lower probability of occurrence. Whereas the lower density kernel values will have greater importance as they depict the important foraging areas. Following previous studies (e.g. Paiva et al., 2010b) the 50% Kernel UD is assumed to represent foraging areas and the 95% Kernel UD to represent the home range of the species, the vital area. An important step in using this analysis is the selection of the smoothing factor (h), which will determine the type of detail of the kernel estimation. Small h values will highlight small details and larger values will evidence only prominent features (Kappes et al., 2011). Comparisons between kernel estimations of different years and breeding phases are only possible if the smoothing factor applied is the same between each foraging trip. We selected a random sample of N = 20 individual trips from each of the 9 field campaigns, totalizing 180 foraging excursions. On those trips we performed kernel estimations using the ad-hoc option of the kernelUD function, thus obtaining an optimal h-value for each of the individual trips. Then a mean smoothing factor of 0.11 was obtained from such exercise and used for all the calculated kernels. The overlap between kernels of each bird foraging excursion was also calculated to (1) determine the spatial segregation between all 9 campaigns (i.e. study years and breeding phases), and to evaluate the degree of overlap of each field campaign with (2) the designated marine Important Bird Area (mIBA) and (3) the legal protection area (the

Marine Protected Area; MPA). Overlaps were also estimated between each year within the groups to assess the consistency of habitat selection in each group. Both calculations were performed in R with the kerneloverlap function of the adehabitatHR package.

2.5. Environmental data

Environmental variables were selected based on its relevance in describing static and dynamic productivity features, and process that have been shown to be correlated with distribution and abundance of seabirds (Table I, Fig. 4; Louzao et al., 2009; Tremblay et al., 2009). These include: chlorophyll a concentration (CHL), sea surface temperature (SST; both are temporal mean from monthly climatologies) and the respective spatial gradients (CHLG, SSTG, respectively) were also calculated over a 3 X 3 grid cell window as [(maximum value – minimum value)*100]/maximum value, with maximum being the highest mean value and the minimum the lowest; peak of chlorophyll a concentration (CHLPK) in a period of 10 years, which correspond to a sum of higher productive cells with chlorophyll a concentration > 1 mg m-3. To account for annual anomalies, we included the SST and CHL anomalies (i.e. CHLA, SSTA, respectively) for each season, calculated as the difference between the average value for a given season and year and the average for that season over a 20-year period in that grid cell. The average distance to fronts of chlorophyll a concentration (DFCHL) and sea surface temperature (DFSST) were computed under the Marine Geospatial Ecology Tools (Roberts et al., 2010) inside ArcGIS 10.1. To account for productivity time lags all dynamic products were aggregated into 3 month composites prior to each fieldwork (e.g. February 2010 – April 2010 for the tracking data of April 2010 – pre-laying period; Louzao et al., 2009). Bathymetry (BAT) and its corresponding gradient (BATG; computed in a way similar to CHLG), which is a proxy of slope and thus the presence of seamounts and other topographic features, were used as static variables. Distance to colony (DCOL) was computed using tools of the Spatial Analysis toolbox inside ArcGIS 10.1. This last variable has great importance, given that seabirds become central place foragers during the breeding season (Orians and Pearson, 1979), becoming highly dependent on returning frequently to the colony. Monthly composites of some variables were obtained from http://coastwatch.pfeg.noaa.gov/ as ASCII files and then converted to raster format with Spatial Analyst toolbox.

Table I. Biologically relevant explanatory variables used for habitat modelling (MaxEnt) and data manipulation and source. Dynamic variables were downloaded on a monthly basis. Since they differed in spatial resolutions, they were aggregated to match the standard grid of 0.04° cell size. Static variables were extracted once and aggregated.

Derived Variable Units Manipulation Source Metric Sea Surface Temperature Temporal mean from monthly Aqua- Mean °C (SST) climatologies (2010-2013) MODIS Temporal deviation of sea surface Sea Surface Temperature Aqua- Mean °C temperature compared to a 20 year Anomaly (SSTA) MODIS average of sea surface temperature Sea Surface Temperature Temporal gradient from monthly Aqua- Mean % Gradient (SSTG) climatologies (2010-2013) MODIS Distance to Fronts of SST Minimum distance to the nearest Aqua- Minimum Km (DFSST) front of SST MODIS Chlorophyll a Concentration Temporal mean from monthly Aqua- Mean mg m-3 (CHL) climatologies (June-August) MODIS Temporal deviation of Chlorophyll Chlorophyll a Concentration a concentration compared to a 20 Aqua- Mean °C Anomaly (CHLA) year average of Chlorophyll a MODIS concentration Chlorophyll a Concentration Temporal gradient from monthly Aqua- Gradient % Gradient (CHLG) climatologies (2010-2013) MODIS Distance to Fronts of CHL Minimum distance to the nearest Minimum Km ArcGIS (DFCHL) front of CHL Sum of higher productive cells (> 1 Chlorophyll a Concentration Aqua- Sum - mg m-3) over a period of 10 years Peak (CHLPK) MODIS (2002-2013)

Bathymetry (BAT) Mean m Spatial mean ETOPO 1 Bathymetry Gradient Gradient % Spatial gradient ETOPO 1 (BATG) Minimum distance to the breeding Distance to Colony (DCOL) Minimum km ArcGIS colony

(A) BAT (B) SST (C) CHL

(D) BATG (E) SSTG (F) CHLG

Figure 4. Graphical representation of some environmental variables used in Species Distribution Modelling (SDM). Represented are bathymetry (A), sea surface temperature (B), chlorophyll a concentration (C) and the respective gradients (D-F).

2.6. Species distribution modelling (SDM)

Tracking positional data is an only occurrence type of data, not providing absent information, it is also spatial and temporally autocorrelated and it is not statistically independent, since several positions belong to one individual bird (Edrén et al., 2010). Thus, for species distribution modelling it was used the MaxEnt 3.3.3k software (Phillips et al., 2006), which is considered one of the best modelling techniques for

29 these type of data (Elith et al., 2006). MaxEnt requires presence-only type of data, can be used with small sample sizes, in the presence of spatial locational errors (Graham et al., 2008) and with the occurrence of spatial mismatch between positional and environmental data. The approach of this software consists in constructing the probability distribution of maximum entropy considering the presence distribution data and the corresponding environmental conditions (Edrén et al., 2010). Before building the SDM, the tracking positional data was rescaled using a spatial grid (mask) with cell size of 4x4 km in order to fit with the environmental data spatial resolution. Positional data was also coded as foraging, with value 1 when sinuosity ≥ 2.7 or as travelling with the value 0. To minimize the correlation between environmental variables, only were included variables with percentage of contribution ≥ 1 that were selected through the analyses of variable contributions provided by the software. MaxEnt models were finally performed with the following settings: logistic output format which classifies a range of values between 0 and 1 depending on the climatic conditions, with higher values representing more similar conditions; removing the duplicate presence records; 30% of the tracking data were randomly attributed as test data for models evaluation; 25 replicate runs of random (bootstrap) subsamples were performed to each breeding phase. To obtain the average prediction the mean of the 25 models was then calculated. The output for each MaxEnt model consist in a jackknife test result chart, which demonstrates the explanatory power of each variable when model is constructed using each variable in isolation, a predicted presence probability map and finally a plot resulted from an receiver operating characteristic (ROC) analysis. This last analysis characterizes the predictive performance of a model by using the Area Under the receiver operating characteristic Curve (AUC) that range from 0 to 1, where 0.5 indicate that model performance is equal to that of a random prediction. The remaining range can be interpreted as excellent if AUC > 0.90; good if 0.80

2.7. The Zonation algorithm

The conservation planning software Zonation (Moilanen et al., 2005) was used to assess the adequacy of the current Berlengas mIBA and MPA boundaries. This software can be used to design and evaluate Marine Protected Areas by producing different scenarios depending on specification of conservational priorities (Moilanen et al., 2005; 2007; Leathwick et al., 2008). This software starts with the assumption that the entire study area is protected and then proceeds to the removal of cells, one by one, starting with the lower conservation value ones and allowing to maintain the maximum connectivity. The criteria to decide which cells are removed first it is based on the removal that causes the smallest marginal loss in conservation, leaving the cells with highest conservation value for last. This remaining area will correspond to the area that is more relevant to protect, taking into the account the specific conservation priorities initially established. Usually, the top 5% of such priority distribution is considered to be the critical (or vital) area in terms of conservation of one or several species. When building the Zonation model we applied three scores to the nine SDM models, reflecting the biological need of the species to forage close to the breeding location (in areas where the MPA region lays): (1) 0.3 for pre-laying; (2) 0.6 for incubation; (3) 0.9 for chick-rearing. These scores will be used by Zonation to attribute different weights to the SDM models entering the marginal loss exercise.

2.8. Data analysis

For each at-sea excursion, the following foraging trip parameters were computed: (1) trip duration (h), (2) total distance covered (km), (3) maximum distance from colony (km), (4) maximum trip sinuosity, (5) maximum travelling speed (km h-1), and (6) loop factor (total distance covered / maximum distance from colony). For foraging areas, the calculated parameters comprised the (7) home range area (95 % Kernel UD; km2), (8) foraging zone area (50 % Kernel UD; km2), (9) 95% Kernel UD overlap with the marine Important Bird Area (mIBA) (%), (10) 50% Kernel UD overlap with the mIBA (%), (11) 95% Kernel UD overlap with the Berlengas Marine Protected Area (MPA) (%), (12) 50% Kernel UD overlap with the MPA (%). Inter-annual and within seasonal differences on the foraging trip and areas parameters (i.e. response variables) of adult Cory’s Shearwaters were tested with

Generalized Linear Mixed Models (GLMMs). GLMMs were fitted by the Laplace approximation and included trip identity nested within the individual as a random term to account for pseudo-replication issues (several trips belonged to the same bird; Zuur et al., 2009). These were performed using the packages “lme4” (Bates et al., 2014) and “lmerTest” (Kuznetsova et al., 2014) in R. All variables were visualized using quantile-quantile plots to see the normality and Cleveland dotplots to check homoscedasticity (Zuur et al., 2009) and, when necessary, log (distance and area metrics) and arcsin (percentages) transformations were adopted. All statistical analyses were performed using the software R. Computations were carried out using several functions within different R packages (e.g. MASS, maptools, adehabitat) and some custom-built functions. Presented are means ± SD at significance level of p < 0.05.

Chapter 3 – Results

Rahel Borrmann

“We ourselves feel that what we are doing is just a drop in the ocean, but the ocean would be less because of that missing drop.” Mother Teresa

3.1. Seasonal and inter-annual differences in foraging patterns

During the pre-laying and chick-rearing periods of 2012, the birds covered ~780 km and ~550 km less on average then in similar periods of 2010. They also kept their trips ~364 km closer to the colony during the pre-laying period of 2012 when compared to that of 2010 (Table II and III). However, in general the foraging behavior characteristics varied more among periods of the breeding cycle, with adults during the chick-rearing period making trips closer to the colony and covering a mean distance of 794 km less than during the pre-laying period (Table II and III). On average, during the chick-rearing period birds showed a mean of 3.9 and 5.2 for maximum sinuosity and loop factor, respectively, whereas during the pre-laying period they showed a mean of 2.0 and 3.1, respectively. Such strong differences in foraging patterns among years may be related with climatic variation, as reflected by the NAO index values, which also showed great differences among years, with a higher positive index value in 2012 and a lower negative index value in 2010 (Table II).

Table III. Generalized linear mixed model (GLMM) results analysing the effect of year (2010- 2013) and breeding phase (pre-laying vs chick-rearing) on different parameters of Cory’s shearwaters foraging trips and foraging areas. All GLMMs included trip identity nested within the individual as a random term.

In respect to the consistency on the use of habitats between periods of the breeding cycle and study years (Table IV), 2012 was the year with less percentage of overlap with any other sampling periods. In general, there was a high overlap for the same periods of the breeding cycle between different years, than between different periods of the same year. The highest overlap occurred between the pre-laying periods of 2011 and 2013 with a difference of 66.6%, when compared with the lowest overlap between the pre-laying period of 2013 and the chick-rearing period of 2012. However, during the years of 2010 and 2013 relatively high overlap values occurred between the three periods of the breeding cycle (Table IV). In relation to foraging areas the same trend can be observed although with lower values and 2012 was the year with less overlap with the other sampling periods. High values of overlap between the same periods of the breeding cycle were also registered. The year of 2013 was the year with higher overlap between the three periods of the breeding cycle, showing the highest overlap values of all comparisons (Table IV).

Table IV. Percentage of kernel overlap between the foraging distributions of birds during each of the nine fieldwork campaigns (2010-2013).

3.2. Species Distribution Modelling (SDM) of a pelagic marine predator

All models were fitted and obtained a score of excellent for the Area Under the ROC Curve (AUC), except the model for the pre-laying period of 2010, which obtained a score of good (Table V). In all models distance to colony (DCOL) was the parameter which contributed more in explaining the habitats selected by the birds and obtained high importance in the permutation test. This parameter showed higher importance during the pre-laying period of 2012, when compared with other years especially with the pre-laying period of 2013, with a difference of 13.2%. Bathymetry (BAT) also showed great importance during the chick-rearing period of 2012, as well as the mean sea surface temperature (SST) during the pre-laying period of 2010 and 2013, and also during the chick-rearing period of 2012. Other important parameters included sea surface chlorophyll a concentration anomaly (CHLA) only for the chick-rearing period of 2012; bathymetry gradient (BATG) for the pre-laying periods in all years except 2010; chlorophyll a concentration (CHL) only for the chick-rearing period of 2011, and finally, distance to front of chlorophyll a concentration (DFCHL) for the pre-laying periods of 2010 and 2012 (Table V). Other variables with lower contribution in explaining the occurrence of the individuals in the final models, but with great importance during the construction of each model (i.e. high parameter permutation value) were bathymetry (BAT), with values over 10% in all breeding phases except during 2010, and chick-rearing periods of 2012 and 2013; and sea surface temperature (SST), with values over 10% in all periods except during chick-rearing of 2013 and pre-laying of 2011 and 2012. Distance to fronts of chlorophyll a concentration (DFCHL) was also relevant during both breeding periods of 2012 and during the chick-rearing periods of 2010 and 2011. The bathymetry gradient (BATG) was important during the incubation period of 2012 and during the pre-laying periods of 2012 and 2013 (Table V). Overall, the most suitable habitat for the species during our study period was the surroundings of Berlenga (breeding colony), showing a core-area with high probability occurrence, almost always above 70% (Fig. 5). However, great differences were found on lower probability occurrence ranges (< 50%), especially during the different breeding periods of 2010, with larger areas at distance from the colony. The reverse situation was detected during the breeding periods of 2012, especially during pre-laying, where the obtained areas were almost exclusively around the colony. Some of the areas

38 at distance from the breeding colony appear to be related with lower depth foraging grounds, i.e. seamounts such as Ashton, Josephine and Gorringe (during the pre-laying periods for all years except 2011), and areas along the Portuguese coast above the continental shelf (chick-rearing period for all years). Plus the continental slope, an area with a large variation in depth, that only appeared during the breeding periods of 2010. Other areas that were predicted by MaxEnt models include habitats around cape S. Vicente in south Portugal (chick-rearing of 2010) and areas near north and south of the Portuguese coast (all chick-rearing periods, except in 2013). During incubation of 2013 the areas predicted were also above the continental shelf in areas around the colony and near the Portuguese coast (Fig. 5).

Figure 5. (continuation). Habitat values from MaxEnt models of the distribution of Cory’s shearwaters between 2010-2013, during different phases of their breeding cycle (pre-laying, incubation and chick-rearing). In the background a representation of the relief, using contour lines.

Table V. Estimates of model fit and relative contributions of the environmental variables to the MaxEnt model, normalized to percentages (values over 10% are marked bold), for Cory’s shearwaters between 2010-2013, during different phases of their breeding cycle (pre-laying, incubation and chick-rearing). To determine the first estimate (parameter contribution), in each iteration of the training algorithm, the increase in regularized gain is added to the contribution of the corresponding variable, or subtracted from it if the change to the absolute value of lambda is negative. For the second estimate (permutation importance), for each environmental variable in turn, the values of that variable on training presence and background data are randomly permuted. The model is re- evaluated on the permuted data, and the resulting drop in test AUC is shown in the table. Values shown are averages over 25 replicate runs.

3.3. Overlap between usage of at-sea areas by birds and the Berlengas mIBa and MPA areas

Foraging areas varied considerably more among years than among periods of the breeding cycle. In 2012 the overlap between the areas used by the birds (home range and foraging areas) and both the Berlengas mIBA and MPA was much higher (range of overlaps: 49 – 71%) than during any of the other study periods years (range of overlaps for 2010, 2011 and 2013: 23 – 51%; Table II). Although not so significant, there were also some differences between periods of the breeding cycle (Table I), with the chick- rearing period always overlapping more with the mIBA and MPA areas, when compared with the pre-laying or incubation periods (Table II). During 2013, the overlap between the foraging area and the mIBA areas was 18.8% higher during chick-rearing than during the pre-laying phase; and the foraging area overlapping with the MPA area was 17.2% higher during chick-rearing than during the incubation period (Table II).

3.4. Conservation aspects

When comparing the top 5% of most relevant foraging region for Cory’s shearwater (as modeled by the Zonation algorithms) with the Berlengas marine Important Bird Area (mIBA) and Marine Protected Area (MPA) areas it is possible to realize that all these areas do not fully overlap (Fig. 6). Comparing the most relevant foraging region with the Berlengas MPA the overlapped area between both polygons represented 25.9% of the overall area, while the same foraging region overlapped 45.5% with the Berlengas mIBA. A new proposed Marine Protected Area for Berlengas (Fig. 6), based on the distribution of several marine taxa is currently under consideration by the Portuguese government, as part of new MPAs proposals for the entire coastal area of mainland Portugal. Comparing these new areas with the most relevant foraging region for Cory’s shearwater a considerably higher overlap of 59.6% is obtained.

Figure 6. Zonation model results (filled grey polygon) showing the top 5% most relevant foraging region for Cory’s shearwaters between 2010-2013, during different phases of their breeding cycle (pre-laying, incubation and chick-rearing). Also shown in the map, the Berlengas Marine Protected Area (MPA; dashed area) and marine Important Bird Area (mIBA; thickest line) and the (new) proposed Marine Protected Areas for the entire coastal mainland Portugal region, under evaluation by the Portuguese government (slimmest line). Bathymetry represented on the background.

Chapter 4 – Discussion

Rahel Borrmann

“At the end of the day, no amount of investing (…) will save the natural world (…) if we are not paying attention to all the things that nature give us for free: (…) mountains for skiing, rivers for fishing, oceans for sailing, sunsets for poets, and landscapes for painters. What good is it to have wind-powered lights to brighten the night if you can't see anything green during the day?” Thomas L. Friedman

4.1. Inter-annual and within season foraging patterns

In the present study great differences were found in the foraging range of Berlengas’ Cory’s shearwater population during 2012 compared with the other years, especially when comparing with the foraging strategies and habitats used during 2010. Differences related with the species foraging behaviour were also found between periods of the breeding cycle. In 2012 the birds foraged closer to the colony, contrasting with 2010 where the opposite occurred. Consequently, 2012 was also the year with higher overlap between the birds foraging area and the mIBA and MPA zones, because birds concentrated their foraging effort in the surroundings of their breeding colony. This seems to be related with the climatic conditions as depicted by the extended winter North Atlantic Oscillation (NAO) index, with the higher NAO index value in 2012 and the lower value during the winter of 2010, in fact it was the lowest NAO index value since 1648 (Osborn, 2011; Paiva et al., 2013a; Hurrell, 2014). Climatic fluctuations reflected by extreme variations in NAO index values are known to influence several processes in marine ecosystems through variations in wind conditions, Sea Surface Temperature (SST) and upwelling patterns (Stenseth et al., 2002; Santos et al., 2007). In Southern Europe (including the mainland Portuguese coast) negative values are related with increases in storm activity, stronger winds, lower SST and alterations in intensity of coastal upwelling. These alterations could led to the dispersion of eggs and fish larvae to unfavourable areas during the spawning season, consequently leading to their mortality, by spatial mismatching with plankton, and leading to a lower recruitment of fish during these years (Santos et al., 2007). The lower availability of fish will probably influence the foraging behaviour of birds, which increase their foraging effort by travelling longer distances and further from the colony. Similar differences in foraging behaviour of Cory’s shearwater during 2010 were also reported by Paiva et al. (2013a), where birds also covered longer distances during their chick provisioning trips and foraged farther from the colony when compared to the breeding seasons of 2005 – 2007. In accordance with this behaviour, in the same study, it was also reported a decrease in abundance of sardine species around the colony, which is usually one of the most important prey for the population of Cory’s shearwater breeding in Berlenga (Paiva et al., 2010d; Paiva et al., 2013a). Other studies using African penguins foraging tracking data also found differences in foraging behaviour during 2010. In a study that intended

46 to design and test the adequacy of the Namibian Islands’ Marine Protected Area (NIMPA), African penguins breeding at Mercury Island travelled greater distances and used an extended area during 2010 comparing with other years (2005-2009; Ludynia et al., 2012). Another study that aimed was to test the potential benefits of the purse-seine fishing exclusions on African penguins breeding in St. Croix Island, also found an increasing foraging effort during 2010 comparing with 2009, where birds decreased their foraging effort after the fishing ban (Pichegru et al., 2012). Cory’s shearwaters exhibited different foraging behaviour throughout the breeding season. Notoriously, birds foraged closer to their colony during the chick- rearing period, when compared with the incubation or pre-laying periods, because of the different energetic requirements of each breeding period and by the central-place foraging behaviour (Orians and Pearson, 1979). During the pre-laying period birds tend to have specific energetic requirements related with their reproductive role; females tend to require highly energetic prey with higher calcium content (Carey, 1996; Mallory et al., 2008) in order to form the egg. After laying the egg, females leave the nest to restore their body condition, while males take the first incubation shift which is usually longer then subsequent incubation shifts. Therefore, during the pre-laying period the energetic requirements are highly demanding for the egg formation by females, and for males preparing for a long starvation period during the first incubation shift. For this reason trips during the pre-laying period tend to be farther from the colony, including the exploitation of pelagic productivity areas, in search of specific highly energetic prey. Péron et al. (2010) found differences between pre-laying and chick-rearing trips of white-chinned petrels (Procellaria aequinoctialis), with trips during the pre-laying period more widely dispersed than trips during the chick-rearing period. Barau’s petrels (Pterodroma baraui) in the Indian Ocean exhibited intersexual differences in their foraging strategy during pre-laying, with males showing higher foraging effort in highly productivity areas (Pinet et al., 2012). In fact this period is so crucial, that if males do not succeed in building up their body condition, they may abandon the egg during the first incubation shift, leading to breeding failure (Warham, 1996; Tveraa et al., 1997). Similarly, Paiva et al. (2013b) reported a decrease in the hatching success rate for Cory’s shearwater breeding in Berlenga, when body condition was lower, presumably because females were forced to exploit remoter areas due to the decrease in productivity around the colony (during years of unfavourable climatic conditions). During chick- rearing, Cory’s shearwater act as central-place forager, and must travel between

47 foraging areas and the colony to provision food to their chick. Adults need to cope with the higher energetic demands of finding food not only for themselves but also to rear the growing chick (Orians and Pearson, 1979). Thereby, during chick-rearing Cory’s shearwater tend to feed in areas closer to the colony, resulting in a higher overlap between foraging areas and Berlengas mIBA and MPA areas, and also between chick- rearing periods of other years. This is in line to what Paiva et al. (2013a) reported for the same colony, where during the chick-rearing periods of 2005-2007 birds were consistent on their habitat use but travelled ~3000 km longer and foraged ~1200 km farther from their colony during 2010. This was assumed to be related with harsh climatic conditions, which should have caused a decrease in productivity patterns and fish availability on the colony surroundings.

4.2. Habitat use and foraging segregation

Although the Cory’s shearwater foraging areas differed between years and breeding phases, they are not randomly distributed across the open ocean, maintaining certain patterns. Overall the habitat that showed to be the most suitable for this species during the nine different study periods included areas surrounding the colony, especially during chick-rearing. Therefore, after the performance of the Species Distribution Models (SDM; MaxEnt) the variable that showed a high importance in all models was “distance to colony”. This variable is commonly used in describing the distribution of seabirds during the breeding season because of their central-place foraging behaviour during this season (e.g. Louzao et al., 2012; Afán et al., 2014). Bathymetry also showed a high importance, especially in describing foraging areas during the chick-rearing period of 2012, as well as the bathymetry gradient which was important in the pre- laying period models of all years except 2010. Bathymetry and bathymetry gradient usually describe shallower waters above the continental shelf and seamounts, which are areas that usually aggregate marine biodiversity and consequently aggregate marine top predators (Morato et al., 2008). Therefore they are two of the most used variables in describing foraging distribution of seabirds (e.g. Louzao et al., 2010; Paiva et al., 2010b; Louzao et al., 2012; Afán et al., 2014). Both these features were highly used by Cory’s shearwater during all our study years. Seamounts located in more distant areas were mainly visited during pre-laying periods. Bathymetry gradient can also describe the large variation in depth along the continental slope, a habitat heavily used during the

48 chick-rearing period of 2010. Sea Surface Temperature (SST) was also very important as a trigger for the foraging distribution of Cory’s shearwater during the pre-laying period of 2010, which could be related with the climatic variations that may have occurred associated with the extremely negative extended winter NAO index of that year. Similar results using feeding habitat use models in Cory’s shearwater were found by Paiva et al. (2013a) with SST as an important variable during the chick-rearing period of 2012 showing lower values comparing with 2005-2007. Sea surface temperature along with chlorophyll a concentration (a marine productivity proxy) are descriptors of upwelling areas that can occur at the edge of the continental shelf along the Portuguese coast (Fiúza, 1983; Santos et al., 2007; Sousa et al., 2008) and near seamounts, which are usually associated with the occurrence of a more restricted upwelling phenomena (Pitcher et al., 2007). Quillfeldt et al. (2012) also found mean chlorophyll a concentration and sea surface temperature to be the most important parameters in describing the year-round distribution of Antarctic and thin billed prions (Pachyptila desolata, Pachyptila belcheri, respectively). Regarding the importance of chlorophyll a anomaly (CHLA), this only appeared important in the model for the chick-rearing period of 2011, which also had chlorophyll a concentration (CHL) as other important variable. This could be related with the decrease of productivity production observed by Paiva et al. (2013b) during this period in the surroundings of the colony, which might have ‘forced’ the species to search and use more productive pelagic environments, i.e. water regimes with higher chlorophyll a concentration. Finally distance to fronts of chlorophyll a concentration (DFCHL) also appeared as an important variable in the models for the pre-laying periods of 2010 and 2012. Although proxies of frontal regions (e.g. distance to fronts of chlorophyll a concentration) are not one of the most commonly used variables in studying the habitats used by seabirds (Tremblay et al., 2009), frontal zones are known to be areas of enhanced productivity and therefore areas where marine taxa, as fish and seabird species, congregate for foraging reasons (e.g. Olson and Backus, 1985; Decker and Hunt, 1996; Spear et al., 2001). In a previous study with Cory’s shearwater in several north Atlantic islands shallower areas with high levels of chlorophyll a concentration and low sea surface temperature were the most important features to explain the at-sea distribution of this species (Paiva et al., 2010b). The at-sea distribution models for the critically endangered Balearic shearwater had similar variables to those of our study: bathymetry, distance to colony, chlorophyll a gradient and chlorophyll a concentration (Louzao et al., 2012). A

49 study using Cory’s shearwater individuals from different populations, also had similar variables explaining species distribution, as sea surface temperature, bathymetry, bathymetry gradient and distance to colony (Louzao et al., 2009). Interestingly Gremillet et al. (2008) observed Cape gannets (Morus capensis) foraging in highly productivity areas of the Benguela upwelling area, with low sea surface temperature and high chlorophyll a concentration, while mismatching the areas with higher abundance of prey-fish. However this specific case may be a consequence of overfishing of the fish stocks in that area and direct competition for marine resources between seabirds and Humans, which might even be leading the populations of Cape gannets from that area to the extinction (Pichegru et al., 2010b). In fact, this case shows the great impact that human overexploitation may have on the ‘normal’ functioning of marine ecosystems; and emphasizes the need to apply Ecosystems Based Fisheries Management (EBFM) strategies (Pikitch et al., 2004) in highly productivity areas that constitute important areas for the survival and reproduction of several marine key-species.

4.3. Management considerations

Regarding the adequacy of Berlengas Marine Protected Area (MPA) in protecting the top 5% of the Cory’s shearwater most important foraging areas, we can conclude that it is far from fully cover the most important foraging areas of this species. The Berlengas MPA is presently only protecting approximately a quarter (25.9%) from the top 5% most important foraging areas of Cory’s shearwater. This is a small value considering that the foraging areas used in this comparison were already restricted to the only 5% most relevant foraging areas. Also, if the mIBA area were converted into a marine protected area, it would protect 45.5% of these top 5% most relevant areas for Cory’s shearwater, which would be larger but still might be insufficient. However, the adequacy would be greater, when comparing with the addition of new proposed areas to the existent Berlengas MPA. These new areas under the Portuguese government evaluation along with the Berlengas MPA would protect 59.6% of the top 5% most important Cory’s shearwater foraging areas. These areas were proposed in the framework of the LIFE+MarPro project, which meets the Birds and Habitats directive aims regarding the marine environment. To perform the new proposed areas several species data were used, including seabird species (Calonectris diomedea, Puffinus mauretanicus, Morus bassanus and Alca torda) and also taking into account the

50 distribution of some cetacean species such as Phocoena phocoena, Tursiops truncatus and Delphinus delphis. Thereby, these areas based on multiple-species distribution provide a greater protection, more than half percent of the top 5% most relevant Cory’s shearwater foraging areas, compared with the actual Berlenga MPA. They also include two mIBAs in the surrounds of cape Raso and Figueira da Foz and the surrounding areas of Sagres, in the south-west of Portugal, an important migratory location with bottleneck effect in the migration of several species (Ramirez et al., 2008). Nevertheless, a better protection would be achieved if such areas would be connected, providing an ecological corridor, which would potentiate the conservation of Cory’s shearwaters and other seabird and cetacean species, among them the endangered Puffinus mauretanicus and Tursiops truncatus. The implementation of larger protected areas is expected to increase the costs associated with the maintenance, enforcement, oversight and motorization. Also, in a short time period it could have a socio-economic impact, especially regarding the implementation of no-fishing areas, which could lead to the reduction of fish supplies. However, considering a medium to long term period of time, protected areas would lead to increases in fish availability, by functioning as reservoirs/ sources of biodiversity that would replenish fish stocks in non-protected areas, thus benefiting fisheries (Sumaila et al., 2007). Nevertheless, in some situations even the implementation of relatively small marine no-take zones positively impacts the populations of both upper-trophic level predators and low-trophic level preys (Pichegru et al., 2010a; 2012). Several other benefits came with the implementation of protected areas, not only related with the conservation of a specific target species’ habitat, but also the habitats of many other marine organisms and other benefits such as eco-tourism with the creation of several new jobs, among others (Sumaila et al., 2007; Toropova et al., 2010). Regarding the costs associated with the implementation of large marine protected areas, in several cases it was found a decrease in the associated costs when incorporating single protected areas into networks (Balmford et al., 2004; Toropova et al., 2010). From the conservation perspective, as marine ecosystems processes are mobile and frequently dependent on ocean currents and other mobile features, the implementation of a single larger protected area instead of protecting several separated protected areas would constitute a more effective way to protect the targeted species and associated ecosystems (e.g. Ludynia et al., 2012).

References

Rahel Borrmann

“One school is finished, and the time has come for another to begin.”

Richard Bach, Jonathan Livingston Seagull

Afán, I., Navarro, J., Cardador, L., Ramírez, F., Kato, A., Rodríguez, B., Ropert- Coudert, Y., Forero, M., 2014. Foraging movements and habitat niche of two closely related seabirds breeding in sympatry. Marine Biology 161, 657-668. Amado, A.(Coord), 2007. Plano de Ordenamento da Reserva Natural das Berlengas - Relatório para Discussão Pública. Instituto da Conservação da Natureza e da Biodiversidade. Angel, M.V., 1993. Biodiversity of the Pelagic Ocean. Conservation Biology 7, 760- 772. Araújo, M.B., Pearson, R.G., Thuiller, W., Erhard, M., 2005. Validation of species – Climate impact models under climate change. Global Change Biology 11, 1504-1513. Arcos, J.M. (compiler), 2011. International species action plan for the Balearic shearwater, Puffinus mauretanicus. SEO/BirdLife & BirdLife International. Balmford, A., Gravestock, P., Hockley, N., McClean, C.J., Roberts, C.M., 2004. The worldwide costs of marine protected areas. Proceedings of the National Academy of Sciences of the United States of America 101, 9694-9697. Bates, D., Maechler, M., Bolker, B., Walker, S., 2014. Linear mixed-effects models using Eigen and S4. R package version 3.0. http://CRAN.R- project.org/package=lme4. Bengala, N.L., Metelo, I.N., Neto, J.M., Pardal, M.A., 1997a. Dinâmica anual de espécies-chave da Ilha da Berlenga. IMAR-Coimbra. Bengala, N.L., Metelo, I.N., Neto, J.M., Pardal, M.A., 1997b. Variação sazonal das principais espécies de macroalgas da Reserva Naturalda Berlenga. IMAR- Coimbra. BirdLife International, 2014. Species factsheet: Calonectris diomedea. Retrieved from http://www.birdlife.org on 09/06/2014. Block, B.A., Jonsen, I.D., Jorgensen, S.J., Winship, A.J., Shaffer, S.A., Bograd, S.J., Hazen, E.L., Foley, D.G., Breed, G.A., Harrison, A.L., Ganong, J.E., Swithenbank, A., Castleton, M., Dewar, H., Mate, B.R., Shillinger, G.L., Schaefer, K.M., Benson, S.R., Weise, M.J., Henry, R.W., Costa, D.P., 2011. Tracking apex marine predator movements in a dynamic ocean. Nature 475, 86-90. Boehlert, G.W., 1996. Biodiversity and the sustainability of marine fisheries. Oceanography 9, 28-35.

Cabral, M.J.(Coord), Almeida, J., Almeida, P.R., Dellinger, T., Ferrand de Almeida, N., Oliveira, M.E., Palmeirim, J.M., Queiroz, A.L., Rogado, L., Santos-Reis, M.(eds), 2005. Livro Vermelho dos Vertebrados de Portugal. Instituto da Conservação da Natureza e das Florestas. Calenge, C., 2006. The package “adehabitat” for the R software: A tool for the analysis of space and habitat use by animals. Ecological Modelling 197, 516-519. Carey, C., 1996. Female Reproductive Energetics. Chapman & Hall, New York. Chape, S., Blyth, S., Fish, L., Fox, P., Spalding, M., 2003. United Nations List of Protected Areas. IUCN - World Conservation Union and UNEP World Conservation Monitoring Centre, Gland, Switzerland and Cambridge, UK. Costa, B.H.E., Erzini, K., Caselle, J.E., Folhas, H., Goncalves, E.J., 2013. 'Reserve effect' within a temperate marine protected area in the north-eastern Atlantic (Arrabida Marine Park, Portugal). Marine Ecology Progress Series 481, 11-24. Decker, M.B., Hunt, G.L.J., 1996. Foraging by murres (Uria spp.) at tidal fronts surrounding the Pribilof Islands, Alaska, USA. Marine Ecology Progress Series 139, 1-10. Dulvy, N.K., Sadovy, Y., Reynolds, J.D., 2003. Extinction vulnerability in marine populations. Fish and Fisheries 4, 25-64. Edrén, S.M.C., Wisz, M.S., Teilmann, J., Dietz, R., Söderkvist, J., 2010. Modelling spatial patterns in harbour porpoise satellite telemetry data using maximum entropy. Ecography 33, 698-708. Elith, J., H. Graham, C., P. Anderson, R., Dudík, M., Ferrier, S., Guisan, A., J. Hijmans, R., Huettmann, F., R. Leathwick, J., Lehmann, A., Li, J., G. Lohmann, L., A. Loiselle, B., Manion, G., Moritz, C., Nakamura, M., Nakazawa, Y., McC. M. Overton, J., Townsend Peterson, A., J. Phillips, S., Richardson, K., Scachetti- Pereira, R., E. Schapire, R., Soberón, J., Williams, S., S. Wisz, M., E. Zimmermann, N., 2006. Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29, 129-151. Fiúza, A.F.G., 1983. Upwelling Patterns off Portugal, in: Suess, E., Thiede, J. (Eds.), Coastal Upwelling Its Sediment Record, Springer US, pp. 85-98. Game, E.T., Grantham, H.S., Hobday, A.J., Pressey, R.L., Lombard, A.T., Beckley, L.E., Gjerde, K., Bustamante, R., Possingham, H.P., Richardson, A.J., 2009. Pelagic protected areas: the missing dimension in ocean conservation. Trends in Ecology & Evolution 24, 360-369.

Garthe, S., 2006. Identification of areas of seabird concentrations in the German North Sea and Baltic Sea using aerial and ship-based surveys. Progress in Marine Conservation in Europe, 225-238. Gonçalves, E.J., Henriques, M., Almada, V.C., 2002. Use of a temperate reef-fish community to identify priorities in the establishment of a Marine Protected Area. In: Beumer JP, Grant A, Smith DC (eds) Aquatic protected areas: What works best and how do we know? Proceedings of the World Congress on Aquatic Protected Areas. Australian Society for Fish Biology, 261−272. Gonzalez-Solis, J., Croxall, J.P., Oro, D., Ruiz, X., 2007. Trans-equatorial migration and mixing in the wintering areas of a pelagic seabird. Frontiers in Ecology and the Environment 5, 297-301. Graham, C.H., Elith, J., Hijmans, R.J., Guisan, A., Townsend Peterson, A., Loiselle, B.A., The Nceas Predicting Species Distributions Working Group, 2008. The influence of spatial errors in species occurrence data used in distribution models. Journal of Applied Ecology 45, 239-247. Granadeiro, J.P., Monteiro, L.R., Furness, R.W., 1998. Diet and feeding ecology of Cory's shearwater Calonectris diomedea in the Azores, north-east Atlantic. Marine Ecology Progress Series 166, 267-276. Gremillet, D., Dell'Omo, G., Ryan, P.G., Peters, G., Ropert-Coudert, Y., Weeks, S.J., 2004. Offshore diplomacy, or how seabirds mitigate intra-specific competition: a case study based on GPS tracking of Cape gannets from neighbouring colonies. Marine Ecology Progress Series 268, 265-279. Gremillet, D., Lewis, S., Drapeau, L., van Der Lingen, C.D., Huggett, J.A., Coetzee, J.C., Verheye, H.M., Daunt, F., Wanless, S., Ryan, P.G., 2008. Spatial match- mismatch in the Benguela upwelling zone: should we expect chlorophyll and sea-surface temperature to predict marine predator distributions? Journal of Applied Ecology 45, 610-621. Guilford, T., Wynn, R., McMinn, M., Rodríguez, A., Fayet, A., Maurice, L., Jones, A., Meier, R., 2012. Geolocators Reveal Migration and Pre-Breeding Behaviour of the Critically Endangered Balearic Shearwater Puffinus mauretanicus. PLoS ONE 7, e33753. Guilford, T.C., Meade, J., Freeman, R., Biro, D., Evans, T., Bonadonna, F., Boyle, D., Roberts, S., Perrins, C.M., 2008. GPS tracking of the foraging movements of

Manx Shearwaters Puffinus puffinus breeding on Skomer Island, Wales. Ibis 150, 462-473. Hilborn, R., Branch, T.A., Ernst, B., Magnusson, A., Minte-Vera, C.V., Scheuerell, M.D., Valero, J.L., 2003. State of the world's fisheries. Annual Review of Environment and Resources 28, 359-399. Hurrell, James and National Center for Atmospheric Research Staff (Eds). Last modified 27 May 2014. The Climate Data Guide: Hurrell North Atlantic Oscillation (NAO) Index (station-based). Retrieved from https://climatedataguide.ucar.edu/climate-data/hurrell-north-atlantic- oscillation-nao-index-station-based. IUCN, UNEP-WCMC, 2013. The World Database on Protected Areas (WDPA) Cambridge, UK: UNEP- WCMC. Jones, H.P., Tershy, B.R., Zavaleta, E.S., Croll, D.A., Keitt, B.S., Finkelstein, M.E., Howald, G.R., 2008. Severity of the effects of invasive rats on seabirds: A global review. Conservation Biology 22, 16-26. Jouanin, C., Roux, F., Mougin, J.L., Stahl, J.C., 2001. Prelaying exodus of Cory's shearwaters (Calonectris diomedea borealis) on Selvagem Grande. Journal Fur Ornithologie 142, 212-217. Kappes, M.A., Weimerskirch, H., Pinaud, D., Le Corre, M., 2011. Variability of resource partitioning in sympatric tropical boobies. Marine Ecology Progress Series 441, 281-294. Kareiva, P., Odell, G., 1987. Swarms of predators exhibit preytaxis if individual predators use area-restricted search. American Naturalist 130, 233-270. Kenchington, R.A., 1990. Managing Marine Environments. Taylor and Francis. Kuznetsova, A., Brockhoff, P.B., Christensen, R.H.B., 2014. Tests for random and fixed effects for linear mixed effect models (lmer objects of lme4 package). R package version 3.0. http://CRAN.R-project.org/package=lmerTest. Lascelles, B.G., Langham, G.M., Ronconi, R.A., Reid, J.B., 2012. From hotspots to site protection: Identifying Marine Protected Areas for seabirds around the globe. Biological Conservation 156, 5-14. Leathwick, J., Moilanen, A., Francis, M., Elith, J., Taylor, P., Julian, K., Hastie, T., Duffy, C., 2008. Novel methods for the design and evaluation of marine protected areas in offshore waters. Conservation Letters 1, 91-102.

Lecoq, M., Crisostomo, P., Mourato , E., Morais, L., Andrade, J., 2012. Censo da População Reprodutora do Corvo-marinho-de-crista no Arquipélago das Berlengas em 2012. Project Report. Sociedade Portuguesa para o Estudo das Aves. (Unpublished). Lecoq, M., Ramírez, I., Geraldes, P., Andrade, J., 2011. First complete census of Cory’s Shearwaters Calonectris diomedea borealis breeding at Berlengas Islands (Portugal), including the small islets of the archipelago. Airo 22, 31-34. Louzao, M., Bécares, J., Rodríguez, B., Hyrenbach, K.D., Ruiz, A., Arcos, J.M., 2009. Combining vessel-based surveys and tracking data to identify key marine areas for seabirds. Marine Ecology Progress Series 391, 183-197. Louzao, M., Delord, K., García, D., Boué, A., Weimerskirch, H., 2012. Protecting persistent dynamic oceanographic features: transboundary conservation efforts are needed for the critically cndangered Balearic Shearwater. PLoS ONE 7, e35728. Louzao, M., Pinaud, D., Péron, C., Delord, K., Wiegand, T., Weimerskirch, H., 2010. Conserving pelagic habitats: seascape modelling of an oceanic top predator. Journal of Applied Ecology 48, 121-132. Ludynia, K., Kemper, J., Roux, J.-P., 2012. The Namibian Islands' Marine Protected Area: Using seabird tracking data to define boundaries and assess their adequacy. Biological Conservation 156, 136-145. Magalhães, M.C., Santos, R.S., Hamer, K.C., 2008. Dual-foraging of Cory's shearwaters in the Azores: feeding locations, behaviour at sea and implications for food provisioning of chicks. Marine Ecology Progress Series 359, 283-293. Mallory, M.L., Forbes, M.R., Ankney, C.D., Alisauskas, R.T., 2008. Nutrient dynamics and constraints on the pre-laying exodus of High Arctic northern fulmars. Aquatic Biology 4, 211-223. Mann, K.H., Lazier, J.R.N., 2006. Dynamics of Marine Ecosystem. Blackwell Publishing, Oxford. Mendes, A.R.N., 2013. Status and conservation of Madeiran Storm-petrel Oceanodroma castro in Farilhão Grande, Berlengas, Portugal: relevance to the management plan of this protected area, Universidade de Lisboa. Moilanen, A., 2007. Landscape Zonation, benefit functions and target-based planning: Unifying reserve selection strategies. Biological Conservation 134, 571-579.

Moilanen, A., Franco, A.M.A., Early, R.I., Fox, R., Wintle, B., Thomas, C.D., 2005. Prioritizing multiple-use landscapes for conservation: methods for large multi- species planning problems. Proceedings of the Royal Society B-Biological Sciences 272, 1885-1891. Morato, T., Varkey, D.A., Damaso, C., Machete, M., Santos, M., Prieto, R., Santos, R.S., Pitcher, T.J., 2008. Evidence of a seamount effect on aggregating visitors. Marine Ecology Progress Series 357, 23-32. Navarro, J., González-Solís, J., G, V., 2007. Nutritional and feeding ecology in Cory's shearwater Calonectris diomedea during breeding. Marine Ecology Progress Series 351, 261-271. Olson, D.B., Backus, R.H., 1985. The concentrating of organisms at fronts: A cold- water fish and a warm-core Gulf Stream ring. Journal of Marine Research 43, 113-137. Orians, G.H., Pearson, N.E., 1979. On the theory of central place foraging. In: Horn D.J., Mitchell R.D., Stairs G.R. (eds) Analysis of ecological systems. Ohio State University Press, Columbus, 154 - 177. Osborn, T.J., 2011. Winter 2009/2010 temperatures and a record-breaking North Atlantic Oscillation index. Weather 66, 19-21. Paiva, V.H., Geraldes, P., Marques, V., Rodriguez, R., Garthe, S., Ramos, J.A., 2013a. Effects of environmental variability on different trophic levels of the North Atlantic food web. Marine Ecology Progress Series 477, 15-28. Paiva, V.H., Geraldes, P., Ramirez, I., Garthe, S., Ramos, J.A., 2010a. How area restricted search of a pelagic seabird changes while performing a dual foraging strategy. Oikos 119, 1423-1434. Paiva, V.H., Geraldes, P., Ramirez, I., Meirinho, A., Garthe, S., Ramos, J.A., 2010b. Oceanographic characteristics of areas used by Cory's shearwaters during short and long foraging trips in the North Atlantic. Marine Biology 157, 1385-1399. Paiva, V.H., Geraldes, P., Ramirez, I., Werner, A.C., Garthe, S., Ramos, J.A., 2013b. Overcoming difficult times: the behavioural resilience of a marine predator when facing environmental stochasticity. Marine Ecology Progress Series 486, 277-288. Paiva, V.H., Guilford, T., Meade, J., Geraldes, P., Ramos, J.A., Garthe, S., 2010c. Flight dynamics of Cory's shearwater foraging in a coastal environment. Zoology 113, 47-56.

Paiva, V.H., Xavier, J., Geraldes, P., Ramirez, I., Garthe, S., Ramos, J.A., 2010d. Foraging ecology of Cory's shearwaters in different oceanic environments of the North Atlantic. Marine Ecology Progress Series 410, 257-268. Passos, C., Navarro, J., Giudici, A., González-Solís, J., 2010. Effects of Extra Mass on the Pelagic Behavior of a Seabird. The Auk 127, 100-107. Péron, C., Delord, K., Phillips, R.A., Charbonnier, Y., Marteau, C., Louzao, M., Weimerskirch, H., 2010. Seasonal variation in oceanographic habitat and behaviour of white-chinned petrels Procellaria aequinoctialis from Kerguelen Island. Marine Ecology Progress Series 416, 267-U288. Phillips, R.A., Xavier, J.C., Croxall, J.P., 2003. Effects of satellite transmitters on albatrosses and petrels. Auk 120, 1082-1090. Phillips, S.J., Anderson, R.P., Schapire, R.E., 2006. Maximum entropy modeling of species geographic distributions. Ecological Modelling 190, 231–259. Pichegru, L., Gremillet, D., Crawford, R.J.M., Ryan, P.G., 2010a. Marine no-take zone rapidly benefits endangered penguin. Biology Letters 6, 498-501. Pichegru, L., Ryan, P.G., Crawford, R.J.M., van der Lingen, C.D., Grémillet, D., 2010b. Behavioural inertia places a top marine predator at risk from environmental change in the Benguela upwelling system. Marine Biology 157, 537-544. Pichegru, L., Ryan, P.G., van Eeden, R., Reid, T., Gremillet, D., Wanless, R., 2012. Industrial fishing, no-take zones and endangered penguins. Biological Conservation 156, 117-125. Pikitch, E.K., Santora, C., Babcock, E.A., Bakun, A., Bonfil, R., Conover, D.O., Dayton, P., Doukakis, P., Fluharty, D., Heneman, B., Houde, E.D., Link, J., Livingston, P.A., Mangel, M., McAllister, M.K., Pope, J., Sainsbury, K.J., 2004. Ecosystem-Based Fishery Management. Science 305, 346-347. Pinet, P., Jaquemet, S., Phillips, R.A., Le Corre, M., 2012. Sex-specific foraging strategies throughout the breeding season in a tropical, sexually monomorphic small petrel. Animal Behaviour 83, 979-989. Pitcher, T.J., Cheung, W.W.L., 2013. Fisheries: Hope or despair? Marine Pollution Bulletin 74, 506-516. Pitcher, T.J.E., Morato, T.E., Hart, P.J.B.E., Clark, M.R.E., Haggan, N.E., Santos, R.S.E., 2007. Seamounts: Ecology, fisheries and conservation. Blackwell Publishing.

Quillfeldt, P., Masello, J.F., Navarro, J., Phillips, R.A., 2012. Year-round distribution suggests spatial segregation of two small petrel species in the South Atlantic. Journal of Biogeography 40, 430-441. R Core Team, 2014. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R- project.org/. Ramirez, I., Geraldes, P., Meirinho, A., Amorim, P., Paiva, V.H., 2008. Important Areas for Seabirds in Portugal. Projecto LIFE04NAT/PT/000213 - Sociedade Portuguesa Para o Estudo das Aves. Lisboa Ramos, R., Granadeiro, J.P., Nevoux, M., Mougin, J.-L., Dias, M.P., Catry, P., 2012. Combined Spatio-Temporal Impacts of Climate and Longline Fisheries on the Survival of a Trans-Equatorial Marine Migrant. Plos One 7. Roberts, J.J., Best, B.D., Dunn, D.C., Treml, E.A., Halpin, P.N., 2010. Marine Geospatial Ecology Tools: An integrated framework for ecological geoprocessing with ArcGIS, Python, R, MATLAB, and C++. Environmental Modelling & Software 25, 1197-1207. Ronconi, R.A., Lascelles, B.G., Langham, G.M., Reid, J.B., Oro, D., 2012. The role of seabirds in Marine Protected Area identification, delineation, and monitoring: Introduction and synthesis. Biological Conservation 156, 1-4. Ropert-Coudert, Y., Wilson, R.P., 2005. Trends and perspectives in animal-attached remote sensing. Frontiers in Ecology and the Environment 3, 437-444. Santos, A.M.P., Chícharo, A., Dos Santos, A., Moita, T., Oliveira, P.B., Peliz, Á., Ré, P., 2007. Physical–biological interactions in the life history of small pelagic fish in the Western Iberia Upwelling Ecosystem. Progress in Oceanography 74, 192-209. Sousa, F.M., Nascimento, S., Casimiro, H., Boutov, D., 2008. Identification of upwelling areas on sea surface temperature images using fuzzy clustering. Remote Sensing of Environment 112, 2817-2823. Spear, L.B., Ballance, L.T., Ainley, D.G., 2001. Response of seabirds to thermal boundaries in the tropical Pacific: the thermocline versus the Equatorial Front. Marine Ecology Progress Series 219, 275-289. Steiner, I., Burgi, C., Werffeli, S., Dell'Omo, G., Valenti, P., Troster, G., Wolfer, D.P., Lipp, H.P., 2000. A GPS logger and software for analysis of homing in pigeons and small mammals. Physiology & Behavior 71, 589-596.

Stenseth, N.C., Mysterud, A., Ottersen, G., Hurrell, J.W., Chan, K.-S., Lima, M., 2002. Ecological Effects of Climate Fluctuations. Science 297, 1292-1296. Stenseth, N.C., Ottersen, G., Hurrell, J.W., Mysterud, A., Lima, M., Chan, K.S., Yoccoz, N.G., Ådlandsvik, B., 2003. Review article. Studying climate effects on ecology through the use of climate indices: the North Atlantic Oscillation, El Niño Southern Oscillation and beyond. Proceedings of the Royal Society of London. Series B: Biological Sciences 270, 2087-2096. Sumaila, U.R., Zeller, D., Watson, R., Alder, J., Pauly, D., 2007. Potential costs and benefits of marine reserves in the high seas. Marine Ecology Progress Series 345, 305-310. Swets, J., 1988. Measuring the accuracy of diagnostic systems. Science 240, 1285-1293. Sydeman, W.J., Piatt, J.F., Browman, H.I., 2007. Seabirds as indicators of marine ecosystems. Marina Ecology Progress Series 352, 199-204. Toropova, C., Meliane, I., Laffoley, D., Matthews, E., Spalding, M.E., 2010. Global ocean protection: present status and future possibilities. Brest, France. Agence des aires marines protégées, Gland, Switzerland, Washington, DC and New York, USA: IUCN WCPA, Cambridge, UK: UNEP-WCMC, Arlington, USA: TNC, Tokyo, Japan: UNU, New York, USA: WCS. Towns, D., Atkinson, I.E., Daugherty, C., 2006. Have the Harmful Effects of Introduced Rats on Islands been Exaggerated? Biol Invasions 8, 863-891. Tremblay-Boyer, L., Gascuel, D., Watson, R., Christensen, V., Pauly, D., 2011. Modelling the effects of fishing on the biomass of the world's oceans from 1950 to 2006. Marine Ecology Progress Series 442, 169-185. Tremblay, Y., Bertrand, S., Henry, R.W., Kappes, M.A., Costa, D.P., Shaffer, S.A., 2009. Analytical approaches to investigating seabird - environment interactions: a review. Marine Ecology Progress Series 391, 153-163. Tveraa, T., Lorensten, S.-H., Sæther, B.-E., 1997. Regulation of foraging trips and costs of incubation shifts in the Antarctic petrel (Thalassoica antarctica). Behavioral Ecology 8, 465-469. Van Winkle, B., 1975. Comparison of Several Probabilistic Home-Range Models. The Journal of Wildlife Management 39, 118-123. Warham, J., 1996. The Behaviour, Population Biology and Physiology of the Petrels. Elsevier Science.

Weimerskirch, H., 2007. Are seabirds foraging for unpredictable resources? Deep-Sea Research Part Ii-Topical Studies in Oceanography 54, 211-223. Wilson, R.P., Putz, K., Peters, G., Culik, B., Scolaro, J.A., Charrassin, J.B., Ropert- Coudert, Y., 1997. Long-term attachment of transmitting and recording devices to penguins and other seabirds. Wildlife Society Bulletin 25, 101-106. Worton, B.J., 1989. Kernel Methods for Estimating the Utilization Distribution in Home-Range Studies. Ecology 10, 164-168. Xavier, J.C., Magalhaes, M.C., Mendonça, A.S., Antunes, M., Carvalho, N., Machete, M., Santos, R.S., Paiva, V., Hamer, K.C., 2011. Changes in diet of Cory's Shearwaters Calonectris diomedea breeding in the Azores. Marine Ornithology 39, 129-134. Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A., Smith, G.M., 2009. Mixed Effects Models and Extensions in Ecology with R. Springer Science + Business Media, LLC, New York, USA.