RISK ASSESSMENT FOR MARINE MAMMAL AND SEABIRD POPULATIONS IN SOUTH- WESTERN IRISH WATERS (R.A.M.S.S.I.)

Daphne Roycroft, Michelle Cronin, Mick Mackey, Simon N. Ingram Oliver O’Cadhla

Coastal and Marine Resources Centre, University College Cork

March 2007

HEA

Higher Education Authority An tÚdarás um Ard-Oideachas CONTENTS

i) Summary ii) Acknowledgements

General Introduction

Seabirds and marine mammals in southwest Ireland 2 Rationale for RAMSSI 6 Study sites 7 Inshore risks to seabirds and marine mammals 11 i. Surface pollution 11 ii. Ballast water 13 iii. Organochlorine pollution and antifoulants 14 iv. Disease 15 v. Acoustic pollution 15 vi. Disturbance from vessels 16 vii. Wind farming 17 viii. Mariculture 17 ix. Fisheries 19 Aims and Objective 22 References 23 Appendix 33

Chapter 1. Seabird distribution and habitat-use in Bantry Bay

1.1 Abstract 35 1.2 Introduction 35 1.3 Study site 37 1.4 Methods 37 1.4.1 Line transect techniques 37 1.4.2 Data preparation 40 1.4.3 Data analysis 45 1.5 Results 46 1.5.1 Modelling 46 1.5.2 Relative abundance 54 1.6 Discussion 59 1.7 References 66 1.8 Appendix 70

Chapter 2. Shore-based observations of seabirds in southwest Ireland

2.1 Abstract 72 2.2 Introduction 73 2.3 Methods 74 2.3.1 Shore-watch techniques 74 2.3.2 Analysis of relative abundance 75 2.3.3 Density calculation 77 2.3.4 Comparison of shore and boat-based densities 79 2.4 Results 80 2.4.1 Relative abundance 80 2.4.2 Density 87 2.5 Discussion 89 2.6 References 93

Chapter 3. Temporal variation in the use of haul-out sites by Harbour seals in Bantry Bay and the Kenmare River

3.1 Abstract 96 3.2 Introduction 97 3.3 Materials and Methods 99 3.3.1 Study site 99 3.3.2 Seal counts 102 3.3.3 Statistical modelling 102 3.4 Results 105 3.4.1 Seal counts 105 3.4.2 Model output and validation 106 3.5 Discussion 121 3.6 References 127

Chapter 4 Haul-out behaviour of Harbour Seals in the Kenmare River, Co. Kerry

4.1 Abstract 133 4.2 Introduction 134 4.3 Methods 136 4.3.1 Study site 136 4.3.2 Capturing and handling procedure and tag deployment 136 4.3.3 Tag operation 137 4.3.4 Information relay and interpretation 138 4.3.5 Statistical modelling 138 4.3.6 Bootstrap variance estimation 140 4.4 Results 141 4.4.1 Capture and tag deployment 141 4.4.2 Duration of transmission 141 4.4.3 Examination of haul-out data 142 4.4.4 Model outputs and validation 143 4.5 Discussion 163 4.5.1 Effects of time of day and tidal cycle on haul-out behaviour 163 4.5.2 Seasonal changes in haul-out behaviour 164 4.6 References 171

Chapter 5. Cetacean distribution and relative abundance in southwest Ireland

5.1 Introduction and methodology 180 5.2 Species accounts 180 5.2.1 Harbour porpoise 181 5.2.2 Common dolphin 182 5.2.3 Risso’s dolphin 183 5.2.4 Bottlenose dolphin 184 5.2.5 Minke whales 184 5.2.6 Fin whale 185 5.3 Conservation recommendations 189 5.4 References 190

Conclusions 192 Recommendations for future work 194 References 197

SUMMARY

The rugged coastline of southwest Ireland is home to the highest concentrations of breeding seabirds in the country, as well as high numbers of resident and migrating cetacean species, many of which are of conservation concern. Furthermore, the sheltered inlets of Bantry Bay and the Kenmare River provide a base for important populations of Harbour seals, an Annex II species under the EU Habitat’s Directive. This inshore environment contains a variety of potential risks to seabirds and marine mammals, most notably from surface & acoustic pollution, fisheries and the rapidly expanding mariculture industry.

The year-round distribution and habitat-use of seabirds and marine mammals in Ireland’s inshore marine environment is poorly studied. Such baseline data is critical for the assessment of the effects of anthropogenic disturbance on marine mammals and seabirds in the coastal environment. The main aim of this study was to establish the spatio-temporal distribution and habitat-use of seabirds and marine mammals in southwest Ireland.

A total of 21 seabird species were recorded in the survey area during the course of the three-year study. Seabird communities around selected headlands in southwest Ireland were dominated by Manx shearwaters (Puffinus puffinus) and Northern gannets (Morus basanus) while the seabird community in Bantry Bay was dominated by auks (mainly guillemots and razorbills, Uria aalge and Alca torda). Diversity was high in Bantry Bay however the relative abundance of many seabird species was higher at the outer headlands. Peak numbers of many species occurred in autumn (August - October) and the tidal cycle did not significantly influence the abundance of any of the species studied. Vulnerable concentrations of seabirds in areas of high risk from oil-pollution were identified.

Investigations of seabird habitat-use using generalized linear and generalized additive modelling in Bantry Bay revealed that seaward distance (the distance from the most inshore point of the study site) was a significant positive determinant of total seabird distribution in Bantry Bay. The density of many

1 seabird species was also positively related to distance from the nearest coast, while depth was a limiting factor in Phalacrocoracidae distribution. The possible conservation applications of this data for seabirds in similar habitats are discussed.

The year round changes in harbour seal abundance and haul-out site use in southwest Ireland were investigated by carrying out year round counts of seals at haul-out sites over a two and a half year period. There was a difference in the seasonal patterns of seal abundance between haul-out sites. The effect of the time of day on seal abundance at haul-out sites varied between the sites and was only significant at sites that also showed a seasonal pattern in abundance. Fewer seals were observed during strong winds and rain.

The haul-out behaviour and habitat use of individual seals was examined using telemetry. The haul-out behaviour of tagged seals varied over the tagging period with animals spending a higher proportion of time ashore post moult in October, decreasing over the winter months to a minimum in February. A strong tidal influence on haul-out behaviour was evident with tagged seals hauling-out more frequently at low tide. There was overall large variation in the patterns in behaviour over the tagging period (i) between individuals and (ii) between tidal periods for each individual.

A minimum of six cetacean species were recorded over the course of the three year study, with the most abundant and widespread species being the Harbour porpoise (Phocoena phocoena) and Common dolphin (Delphinus delphis). Minke whales (Balaenoptera acutorostrata) were also common across the study site and at least one Fin whale (Balaenoptera physalus) was sighted from the outer headlands. Peak numbers of many species occurred in Autumn and calves or immature adults were present throughout the year.

The findings of the study are discussed in relation to those conducted in different parts of the species geographical range and the significance of the information put into context of conservation management and monitoring requirements for the species.

2

ACKNOWLEDGEMENTS

This research was made possible through funding from the Higher Education Authority (HEA, Prtl-3) and was facilitated by Coastal and Marine Resources Centre and the Department of Zoology, Ecology and Plant Science, University College Cork.

This project was greatly enhanced by the expert supervision of Dr Tom Kelly and Dr Emer Rogan, of the Department of Zoology, Ecology and Plant Science, UCC.

Thanks to: Dr Derek Scott for his Dursey Island cable-car taxi service; Ann & Brendan Finch for accommodation on Dursey Island; Ann & Jerome Harrington for allowing access to Black Ball Head; Mr Stephan O'Sullivan (Irish Lights) for allowing free access to Mizen Head signal-tower; Vicki O’Donnell for help with GIS mapping; Lesley Lewis, Pete Jones, Mark Wilson, Alain Zuur and Lisa Borges for advice on statistical analysis; Claire Pollock for help with correction factors; Tom Hubbard for assistance on a number of boat surveys; Steve Newton and the Birdwatch Ireland/Seabird 2000 crew for kindly providing breakdowns of breeding seabird counts for southwest Ireland; Dave Millard and colleagues at BIM, as well as Gavin Burnell, Julie Maguire and Claire Lehane for information on mussel suspension culture. Thanks to Bernie McConnell, Alisa Hall (SMRU), Declan O’Donnell, Clare Heardman (NPWS) and Allen Whittaker for help with seal tagging.

All photographs in this report are the copyright of Mick Mackey and Michelle Cronin.

3

GENERAL INTRODUCTION

1 SEABIRDS & MARINE MAMMALS IN SOUTHWEST IRELAND

The coastal and offshore waters of southwest Ireland are essential feeding grounds for many seabirds and marine mammals, including non-breeders and passage migrants, throughout the year. A total of 52 species of seabirds and 21 cetacean species have been recorded both on and off the Irish shelf (Berrow & Rogan, 1997; Pollock et al., 1997; Mackey et al., 2004a; Mackey et al., 2004b).

The heavily indented, cliff-edged coastline of Ireland with its numerous offshore islands and accessible rich feeding grounds, offers an ideal base for the formation of seabird colonies. A total of 24 seabird species breed in the Republic of Ireland (Mitchell et al., 2004). Six of these are listed in Annex 1 of the E.U. Birds Directive (79/409/EEC) as species of conservation priority. A further nine species are listed as Birds of Conservation Concern in Ireland (BoCCI) either because their breeding populations are in moderate decline or of international importance (Table 1). A further eight regularly occurring winter or passage migrants are also listed as Annex 1 species, or as Birds of Conservation Concern in Ireland (Table 2). Member states of the European Union are required to protect important populations of Annex 1 species through the designation of Special Protection Areas (SPA’s).

Breeding colonies in the southwest of Ireland hold the largest concentration of breeding seabirds in the country (Figure 2 and Table 3). A total of 78% of the Irish Manx shearwater, Puffinus puffinus population and 77% of the Irish European storm-petrel, Hydrobates pelagicus population breed on the islands off southwest Ireland. The Island of Inishtooskert in the Blasket Island group holds the largest storm-petrel colony (27,297 pairs) in Britain and Ireland, and probably the world (Mitchell et al., 2004). The two largest gannet, Morus bassanus colonies in Ireland, holding 94% of the Irish breeding population, are also located off the southwest coast of Ireland on Skellig Michael and the Bull Rock (Appendix 1). Furthermore, over 45% of the Irish puffin, Fratercula arctica population breeds in southwest Ireland, with Puffin Island alone holding a quarter of the total (c. 5,000 individuals) (Mitchell et al., 2004).

2 Table 1. List of breeding seabird species in the Republic of Ireland, their breeding numbers (Seabird 2000 survey), percentage of the total biogeographical population, relevant biogeographical area and highest conservation status. Modified from Mitchell et al., (2004).

Species Total pairs Max % of Biogeog- Highest Rep. of biogeog- raphical Conservation Ireland raphical area status population Northern Fulmar 32,918 1.4 Atlantic Manx Shearwater 32,545 17.9 World BoCCI (B,E) European Storm-Petrel 99,065 42.7 NE Atlantic Annex 1 Leach’s Storm-Petrel 310 0.0 N Atlantic Annex 1 Northern Gannet 35,4571 9.01 World BoCCI (B,E) Great Cormorant 4,548 10.0 N Atlantic BoCCI (B) European Shag 3,426 5.6 NE Atlantic Great Skua 1 0.0 World Mediterranean Gull 3 0.0 Europe Black-headed Gull 3,876 0.7 World BoCCI (B) Mew (Common) Gull 1,060 0.4 NW & C Europe BoCCI (E) Lesser Black-backed Gull 2,876 2.7 NE Atlantic Herring Gull 5,521 0.9 NW Europe Great Black-backed Gull 2,243 2.2 Europe Black-legged Kittiwake 36,100 2.0 N Atlantic Sandwich Tern 1,762 5.4 Europe Annex 1 Roseate Tern 734 38.8 Europe Annex 1 Common Tern 2,485 1.9 Europe Annex 1 Arctic Tern 2,735 0.7 Europe/N Atlan. Annex 1 Little Tern 206 1.2 Europe Annex 1 Common Guillemot 138,108* 5.7 N Atlantic BoCCI (E) Razorbill 25,980* 6.6 NW Europe BoCCI (B) Black Guillemot 3,367* 1.7 Atlantic BoCCI (E) Atlantic Puffin 19,641 0.4 Atlantic BoCCI (B,E) *number of individuals (not pairs), 1 2004 Census (Newton pers.comm.) Annex 1 = Rare or vulnerable species listed in the E.U Birds Directive (79/409/EEC) BoCCI = Birds of Conservation Concern in Ireland (Newton et al., 1999) BoCCI (B) = Amber List. Breeding species with moderate decline, rare/sporadic breeding and/or internationally important or localized. BoCCI (E) = Amber List. European Conservation Concern

Table 2. List of regularly occurring winter/passage seabirds in the Republic of Ireland, their maximum wintering numbers and highest conservation status.

Species Maximum Rep. of Ireland Highest Conservation status wintering individuals* Red-throated Diver 136 Annex 1 Black-throated Diver 31 Annex 1 Great-northern Diver 318 Annex 1 Great-crested Grebe 1,157 BoCCI (B,W) Common Scoter 7,198 BoCCI (D,H) Cory’s Shearwater Annex 1 Great Shearwater BoCCI (W) Sooty Shearwater BoCCI (W) *I-WeBS counts for winter 2000/01 (Colhoun, 2002) BoCCI (W) = Amber List. Wintering/passage species which are internationally important BoCCI (D) = Red List. Breeding population has declined by >50% in the last 25yrs. BoCCI (H) = Red List. Breeding population Historically declining (since 1900).

3

Figure 2 The location of the major seabird breeding colonies in southwest Ireland.

Table 3 The important breeding populations, number of breeding seabird species and conservation designation of the major breeding colonies in southwest Ireland (after Larner & Douglas, 2002; Mitchell et al., 2004), Site Important Species Number of Designation (>1000 pairs) Breeding Species Blasket Islands SP, Mx, F 14 SPA, SAC, NHA Puffin Island Mx, SP, P 11 SPA, NHA Skellig Rocks G, SP, Mx, P 12 SPA, NHA, WHS Scarriff & Deenish SP, Mx 7 NHA Bull & Cow Rocks G, SP 9 SPA, NHA SP = European Storm Petrel, Mx = Manx Shearwater, F = Northern Fulmar, G = Northern Gannet, P = Atlantic Puffin. SPA = Special Protection Area, SAC = Special Area of Conservation, NHA = Natural Heritage Area and WHS = World Heritage Site.

4 Ireland’s continental slope – which is located less than 75km from the southwest coast - has been identified as an area of high importance for many cetacean species (Pollock et al., 1997; Mackey et al., 2004b). Several studies have also reported high densities of minke whales (Balaenoptera acutorostrata) and harbour porpoises (Phocoena phocoena) in the inshore waters of southwest Ireland in summer, indicating that this area provides an important habitat for these species (Pollock et al., 1997; Hammond et al., 2002; Mackey et al., 2004b). In addition there is evidence to suggest that Irelands Atlantic Margin forms part of the migratory pathway of a number of large baleen whales, including humpback, Megaptera novaeangliae, fin, Balaenoptera physalus, sei Balaenoptera borealis and even blue whales, Balaenoptera musculus, as they move from winter calving grounds in the south to summer feeding grounds at high latitudes (Charif et al., 2001; Harwood & Wilson, 2001; Mackey et al., 2004b and Irish Whale and Dolphin Group anecdotal sightings, www.IWDG.ie).

Two seal species, the harbour seal (Phoca vitulina) and grey seal (Halichoerus grypus), breed on the Irish coastline, with high numbers occurring on the southwest coast. Recent Harbour seal population estimates carried out during the 2003 moult season yielded a minimum population estimate of 2,905 for the Republic of Ireland with concentrations of this species in the southwest, west and northwest of the country (Cronin et al., 2004). In addition, approximately one third of the Irish grey seal population breed on the Blasket Islands, in southwest Ireland (Cronin & O'Cadhla, 2004) indicating the importance of this area for pinnipeds.

All cetaceans present in European waters are protected under EU law and listed in Annex IV of the EU Habitats Directive, as species of community interest in need of strict protection. In addition the bottlenose dolphin (Tursiops truncatus), harbour porpoise and the harbour and grey seal are listed in Annex II as requiring the designation of Special Areas of Conservation (SAC’s). In 1991 the Irish government declared Irish waters a whale and dolphin sanctuary including the State’s 200-mile exclusive fishery limit (Rogan & Berrow, 1995).

5 RATIONALE FOR RAMSSI

Despite the obvious importance of the inshore waters of southwest Ireland both for seabirds and marine mammals (many of which are of conservation concern), very little baseline data exist on their distribution and seasonal abundance in the area. The location and abundance of seabirds at breeding colonies in the region is well known (Mitchell et al., 2004) however the at-sea distribution of breeding populations as well as non-breeding migrants in inshore waters is poorly understood – apart from a few localized land-based records (e.g. Hutchinson, 1981). The biggest risks to seabirds and marine mammals from anthropogenic activities exist in inshore waters and so it is particularly important to assess the distribution of sensitive species in this region.

In general, recent studies of seabird and cetacean distribution in Irish waters have been carried out on relatively large scales with a focus on offshore areas and have indicated that the inshore (neritic) zone holds a higher diversity of seabird species with higher abundances than continental slope or oceanic waters (Pollock, 1994; Pollock et al., 1997; Mackey et al., 2004a). To date, there has been no attempt to determine relationships between the at-sea seabird and cetacean distribution described in these studies and physical habitat variables.

Harbour and Grey seals spend a significant proportion of their lives in the coastal zone, where they use terrestrial habitat to come ashore to haul-out to rest, breed and moult and inshore waters to forage in and to navigate through to more offshore areas. Despite recent efforts in addressing the shortfall in population data on a national scale information on the year round patterns in abundance and habitat use of harbour seals in Ireland was lacking. Such information is critical for the effective management and conservation of the species as required under national and international legislation.

Reproductive output is comparatively low in seabirds and marine mammals as both groups are k-strategists often producing only one offspring in each breeding attempt. This means that the rate at which their breeding populations can increase is relatively slow and populations may take several years to recover

6 from a discrete mass adult mortality event such as an oil spill. It is therefore essential that areas of high seabird and marine mammal use - particularly in inshore waters where risks from anthropogenic activities are high - be identified in order to protect these sensitive species.

Global climate change is likely to influence seabird and marine mammal distribution and abundance around Ireland in the coming decades - or even years. Slight changes in oceanographic conditions could have large-scale and pervasive effects on seabird distributions, feeding ecology, reproductive success and populations (Montevecchi & Myers, 1997). Marine mammals are adapted to specific temperature regimes and may be forced to leave otherwise suitable habitat if temperatures fall outside the ranges to which they are adapted (Hoelzel, 2002). A recent study has suggested that negative effects of climate change on sandeel availability may increasing the likelihood of starvation in harbour porpoise populations in the North Sea at certain times of year (MacLeod et al., 2007). Sandvik et al. (2005) documented correlations between the North Atlantic Oscillation (NAO) index and adult survival in North Atlantic seabirds (common guillemot, Brunnich’s guillemot, Uria lomvia, razorbill and puffin). Their evidence suggests that meteorological parameters affect seabird mortality indirectly, possibly through the food chain (see also Alheit & Hagen, 1997). However the effects of the NAO and global climate change may take many years to become apparent in long-lived marine top predators, e.g. northern fulmars, Fulmarus glacialis (Thompson & Ollason, 2001). It is vital that current baseline data on seabird and marine mammal abundance and distribution be gathered so that future distributional shifts resulting from climate change can be identified.

STUDY SITES

The RAMSSI study area comprises the inshore waters of southwest Ireland between Mizen Head (51o 27’N, 09 o 49’W) to the south and Lambs Head (51 o 45’N, 10 o08’W) on the northern shore of Kenmare River. Within this study area the major inlets of Bantry Bay and the Kenmare River were the focus of harbour

7 seal surveys, while Bantry Bay and its approaches (including Mizen Head) formed the seabird and cetacean survey area (Figure 2).

The coastline of the southwest coast of Ireland is characterized by steep cliffs and heavily indented rocky shores with numerous rocky islands. Several long, narrow inlets – typically drowned river valleys (ria’s), separated by exposed sandstone headlands, define this region. Due to their northeast-southwest orientations these inlets are open to the prevailing south-westerly winds and have a primarily wind-driven circulation (Raine et al., 1990; Irish Hydrodata, 1990). The tidal regime is semi-diurnal with two high waters and two low waters each day. Tidal currents are generally weak (<1 km.h-1) in the region however strong currents occur around the headlands (e.g. 6.5km.h-1 off Dursey Island; 5.6km.h-1 off the Mizen Head) and in areas of constricted water flow such as south of Whiddy Island (e.g. 2.8km.h-1) in Bantry Bay (Admiralty Charts, No. 2495, 2424 & 1838 respectively). Mean sea-surface temperatures range from 8oC in winter to 16oC in summer (Lee & Ramster, 1981).

The hydrography of the area is complex. Around the southwest coast of Ireland a frontal system known as the Irish Shelf Front comes very close to the coastline and in summer, on occasion, has been found at the mouth of Bantry Bay (Edwards et al., 1996; Raine & McMahon, 1998). When this inshore movement of the front occurs, the weak clockwise circulation of coastal water from the Celtic Sea is cut off and deep sub-thermocline Atlantic Shelf Water encroaches towards the coast. This results in an upwelling of highly productive water which promotes dense concentrations of phytoplankton. The overall consequence is that the coastal seas off southwest Ireland are highly productive (Raine et al., 1990; Edwards et al., 1996; Raine & Joyce, 1996) and provide rich feeding grounds for top predators.

8

Figure 2. The RAMSSI study sites showing locations of shore-based observation points, boat survey routes and Harbour seal survey areas in southwest Ireland. The locations of the main seabed depth contours (50 and 100m) are also shown (light blue).

Bantry Bay (51o 42′N, 9o 31′W) is the largest of the long marine inlets in southwest Ireland. The entrance to the bay at Sheep’s Head is approximately 10km wide, steadily narrowing to 3-4km at its head. Its length from Sheep’s Head to the inner-most point is 35km. The northern ‘arm’ of the bay extends a further 20km past Sheep’s Head to Dursey Island and the Bull Rock and thus provides some shelter from northerly winds. Depths are typically less than 30m in the inner bay but deepen to 70m in outer regions with a predominantly muddy substrate (Smith & McLaverty, 1997). The Bay is broad, gently sloping and steep-sided, particularly in the outer region, with only a few sections of muddy inter-tidal zone habitat. Freshwater input is relatively low, with surface salinities at the head of the bay typically in excess of 33 ‰ (Raine et al., 1993).

9 There are a number of sheltered harbours along the northern shore of the Bay, one of which (Glengarriff Harbour) has been designated a Special Area of Conservation (SAC) due in part to its large population of Harbour Seals. The bay is a known spawning ground for autumn-spawning herring Clupea harengus (see Smith & McLaverty, 1997).

Kenmare River (51o 43’N, 10o 05’W) is approximately 41km long and 8km wide at its mouth, between Lamb’s Head on the north side of the bay and Cod’s Head on the south (Figure 2). The inter-tidal areas of Kenmare River are dominated by rocky shores that run directly into the sea as cliffs at Dursey Island, Cod’s Head and Lamb’s Head. The inner parts of the smaller bays, for example Ardroom and Kilmackillogue harbours and the upper part of Kenmare River are sedimentary shores dominated by muddy sands (Smith & McLaverty, 1997). The bedrock is mainly old red sandstone which forms reefs along the middle of the bay throughout its length. Kenmare River can be described as a partially mixed estuary with stratification occurring during the summer months, and occasionally in winter due to high freshwater run-off. However the halocline can be dissipated within hours in strong winds and despite the freshwater influence, surface salinities approximately half way between the head and mouth of the river are in excess of 34‰ (McGovern et al., 2001). Depths range from 30m in the innermost parts to 75-80m in the central part of the outer bay (Admiralty Chart No. 2495). The bay is a designated Special Area of Conservation due partly to the presence of the Annex II Harbour seal species.

Bantry Bay is heavily utilized for the shipping and storage of oil and other hazardous substances - activities which potentially pose a high risk to seabirds and marine mammals at sea. In addition, the sheltered inner regions of Bantry Bay and Kenmare River are heavily utilized by the rapidly expanding mussel longline industry, as well as marine finfish farming – activities which may deprive seabirds and marine mammals of foraging habitat. Boat-based eco- tourism is intensive in Glengarriff Harbour, Bantry Bay and in Kenmare River during the summer, posing a potential source of disturbance to seals as well as feeding seabirds and cetaceans.

10 INSHORE RISKS TO SEABIRDS AND MARINE MAMMALS i) Surface pollution Many authors have identified the risks that water-borne pollution incidents pose to seabird populations, particularly those species that frequently associate with the ocean’s surface, e.g. auks, divers, sea-ducks, skuas, Manx shearwater, European shag and great cormorant (Cramp et al., 1974; Tasker et al., 1990; Williams et al., 1994; Webb et al., 1995; Pollock et al., 2000; Mackey et al., 2004a). Even slight exposure to oil can be fatal to seabirds due to fouling of feathers and pathological effects of oil ingestion (Briggs et al., 1997). Marine mammals rely on their blubber for insulation and so are less vulnerable to fouling by oil than seabirds, however they are at risk from evaporating hydrocarbon vapours which they may inhale when breathing at the surface (Geraci & St. Aubin, 1990). Symptoms from acute exposure to volatile hydrocarbons include irritation to the eyes and lungs, lethargy, poor coordination and difficulty breathing. These symptoms may occasionally result in drowning. Neonatal seal pups (particularly grey seals) are most at risk from oil spills as they lack blubber and rely on their fur for insulation. Grey seal pups in particular are restricted to the natal haul out site until they are weaned and so are incapable of leaving a contaminated area (Ekker et al., 1992).

During the past 30 years several oil spills have affected the Irish coast, with many of these on the southwest coast. One of the first significant kills was associated with the 1979 BETELGEUSE spill at the Whiddy Island oil terminal. This resulted in the oiling of approximately 1,000 seabirds in the vicinity of Bantry Bay (Cross et al., 1979). Subsequently, a spill of unknown origin in the southeast resulted in the beaching of 545 birds in the winter of 1982/83. In 1986, the loss of between 1,000 and 1,500 seabirds, primarily auks as well as a small number of grey seals, was attributed to the sinking of the KOWLOON BRIDGE on the Stag rocks near Baltimore, Co. Cork (Hutchinson, 1989; Smiddy, 1992). A further 1,500 seabirds, primarily guillemots, as well as two grey seals were killed in an oil spill in Cork Harbour (south coast) in November 1997 (Smiddy, 1998).

11

Incidents involving very large crude carriers such as the grounding of the SEA EMPRESS (72,000t of crude oil) at the entrance of Milford Haven in South Wales in 1996, The PRESTIGE (25,000t of heavy fuel oil) off the Spanish Coast in 2002 and the total loss of the BREAR (85,000t of crude oil) in the Shetland Islands in 1993 also resulted in the spillage of crude oil on adjacent coastlines in the northeast Atlantic and caused high seabird mortality (Fletcher, 2003).

Williams et al. (1994) developed an oil vulnerability index (OVI), based on four easily scored factors, to assess the relative risks to different seabird species from surface pollution. Briefly, these included: a) the proportion of time spent on the sea surface by that species; b) the size and bio-geographical population of the species; c) the potential rate of recovery following a reduction in numbers for each species and d) the reliance on the marine environment by each species. The maximum OVI score is 30. Using this technique, Webb et al. (1995) calculated that divers (OVI=29), grebes (23-26), great skuas (25), shags (24) auks (21-29) and Manx shearwaters (23) are the most vulnerable groups of marine birds occurring south and west of Britain. These values can be applied to seabird populations in southwest Ireland, thus enhancing the quality of conservation advice that can be given, either in consultations or in reactive situations following a pollution incident.

Chemicals are also transported at sea in bulk by specialist tankers. These generally carry thousands of tonnes of one or a small number of chemicals. If these vessels sink, as did the ‘PERINITIS’ in 1989 and the ‘IEVOLI SUN’ in 2000 in the English Channel, these chemicals can be released into the sea. In the first case, 6 tonnes of insecticide were lost in a container carried on deck, and in the second, 1000 tonnes each of two solvents were released from the wreck of the vessel on the seabed (Fletcher, 2003). The impact of these spills on seabirds and marine mammals are unknown, but likely result in mortalities.

The largest single threat to seabirds and marine mammals in Bantry Bay is the oil terminal located on Whiddy Island, which was the site of a major oil spill in 1979 (Cross et al., 1979). This terminal provides storage facilities (1 million

12 tonnes capacity) for the Irish strategic oil reserve and for output from offshore oil fields in Irish coastal waters. Both crude oil and refined products, including kerosene, gas oil and jet fuel are pumped ashore from tankers that moor onto the Single Point Mooring (SPM) - an offshore jetty linked to the terminal via pipeline along the sea bed. The SPM is capable of handling tankers of up to 320,000 tonnes. Cargo ships, naval vessels and large trawlers also use the terminal for re-fuelling (http://bantrybaycharter.ucc.ie).

A total of 52 trading vessels entered Bantry Bay in 2004, with almost half of these carrying liquid bulk, amounting to a total of 535,000 tonnes of oil (Anonymous, 2005). Additional shipping traffic (with associated fuel oil) is generated by a small number of privately owned quarrying operations within the coastal zone of Bantry Bay. On the north side of the bay at Leahill point, one of the largest quarries in Ireland has its own jetty, which accommodates ships of up to 75,000 tonnes. Quartzitic chippings from the quarry are exported to the UK and continent on a regular basis from this jetty in the inner bay (http://bantrybaycharter.ucc.ie).

ii) Ballast water The use of seawater as ballast in ships has resulted in harmful introductions of alien species. Translocations of marine life have become much more common in recent decades, as vessels have become larger and faster. This can result in the alteration of food-webs and consequently the food source of top-predators such as seabirds and marine mammals. The introduction of the American comb jellyfish (Mnemiopsis leidyi) via ballast water into the Black Sea and Sea of Azov in 1982 resulted in the near collapse of the anchovy fishery which had become its main prey species (Fletcher, 2003). Species of phytoplankton such as Chatonella verruculosa, originally from Japan but introduced to the Skagerrak and northern Kattegat in 1998 can form toxic blooms and cause deaths of wild and farmed fish (i.e. 350 tonnes of farmed salmon in Scandinavia) (Fletcher, 2003). These effects of ballast water also constitute a potential ecological disturbance which could adversely affect seabird and marine mammal populations.

13 iii) Organochlorine pollution and antifoulants Polychlorinated Biphenyls (PCB’s) can cause infertility in Harbour seals (Rejinders, 1986) and cause hormonal effects which reduce the ability of seabirds to reproduce successfully (e.g. Dirksen et al., 1995; Bustnes et al., 2003). Relatively high levels of PCBs (mean 2.38mg/kg² extractable fat, similar to that found in Cork city) were found in otter spraints from the Bantry Bay coastal zone in the early 1990’s, indicating possible contamination from oil-refining activities here (O'Sullivan et al., 1993). A number of studies have recorded the presence of PCB’s in tissue samples taken from live bottlenose dolphins, Tursiops truncatus and bycaught common dolphins, Delphinus delphis and Harbour porpoises in Irish waters (Smyth et al., 2000; Berrow et al., 2002). However levels of PCB in the environment have been declining steadily since the 1970’s and levels of these compounds in Europe are below the levels required to cause symptoms (Becker, 1991; Berrow et al., 2002; Mitchell et al., 2004). Organochlorine pesticides such as DDT (in the form of DDE) caused eggshell thinning of many birds i.e white-tailed sea eagles, Haliaeetus albicilla in the Baltic Sea in the 1960’s (Helander et al., 2002) however the effects are now abating as concentrations fall.

Tributyltin (TBT) was used widely as an antifoulant until 1993 and caused sexual abnormalities (i.e. imposex) in gastropods such as dogwhelks, Nucella lipillus (Gibbs et al., 1987). It is also known to bio-magnify in the tissues of top predators such as seabirds, particularly sea-duck species which feed on molluscs and causes immunosuppression in marine mammals (Guruge et al., 1996; Kannan et al., 1998; Berge et al., 2004). TBT-based anti-foulants are still being used on large ships (>25m length) but will be phased out between 2003-2008 (Fletcher, 2003). Bantry Bay is subject to moderate levels of shipping traffic and so may be contaminated to some degree with TBT. The effects of TBT on seabirds is little known, but some species (i.e. the cormorant, Phalacrocorax carbo) appear to be able to metabolise butyltins and to shed up to one fourth of their body burden during a complete moulting cycle (Guruge et al., 1996).

14 iv) Disease An outbreak of Phocine Distemper Virus (PDV) in 2002 caused over 22,000 seal mortalities (mostly Harbour seals) in western Europe and affected animals in Northern Ireland and parts of the Republic (see Cronin et al., 2004). This was the second outbreak of PDV in western Europe, with the first occurring in the summer of 1988 causing 17,000 seal mortalities (Van der Toorn, 1990). The disease is perceived as a natural phenomenon, however future outbreaks could pose a threat to the population. v) Acoustic pollution The noise associated with marine exploration and resource extraction represents a source of acoustic degradation in the marine environment. Boat sonar, airguns, drilling operations and shipping traffic all produce sounds with combined low and high frequency components (Goold, 1996), which may potentially affect both the low frequency-sensitive baleen whales and high frequency-sensitive toothed cetaceans (Harwood & Wilson, 2001). These effects are heightened in shallow waters due to the presence of subsidiary pulses resulting from reflection off the seafloor.

Damage to marine mammals from high-level sound may be direct (lethal, sub- lethal or non-lethal) or indirect (changes in behavioural or distribution patterns). Direct injuries include physical damage (blast trauma) to the ear, lungs and gastro-intestinal tracts (often causing mortalities) while indirect effects include avoidance reactions and reduction of calling rates by various baleen whale species (Ketten, 1993; Richardson et al., 1995). Other possible indirect effects include disease, reduced foraging opportunities, reduced mating success, decompression sickness, live mass stranding events and the exclusion of cetaceans from important habitats. Detailed information on the distribution of cetaceans is vital in order to mitigate for these indirect acoustic effects.

The southwest coast of Ireland is a ‘priority area’ for seabed mapping through the INtegrated mapping FOr the sustainable development of Ireland’s MArine Resource (INFOMAR) program (phase 2 – the successor to the Irish National Seabed Survey INSS, which covered offshore areas of Ireland’s territorial

15 seabed). Bantry Bay in particular has been identified as one of 26 priority bays which are currently being mapped (2006 onwards). The outer Kenmare River and Dursey Island area were surveyed in 2004 by the Geological Survey of Ireland (GSI) and the Marine Institute (MI) (www.gsiseabed.ie). In addition to these surveys there has been an increase in oil and gas exploration in the waters around Ireland, particularly along the west coast. All of these investigations require the use of active acoustic survey techniques, the impacts of which are poorly understood for marine mammals.

vi) Disturbance from vessels. Many toothed whales appear to be tolerant of vessel noise and are regularly observed in areas of heavy traffic (e.g. bottlenose dolphins) (Rogan et al., 2000). However this is not true of many larger baleen whales. Sperm whales have been reported to react to vessels with powerful outboard engines at distances of up to 2 km. Humpback whales and right whales are also reported to avoid large vessels in some areas, with vessel collisions forming a serious cause of right whale (Eubalaena glacialis) mortality in the North Atlantic (Katona & Kraus, 1999). Fin whales are reputed to ignore large vessels, but they respond to close approaches by whale-watching vessels by spending less time at the surface and by making shorter dives (Richardson et al., 1995).

Boat-based ecotourism, mainly focused on seals, is popular in Glengarriff Harbour in Bantry Bay and parts of Kenmare River. During the peak summer months several tour vessels operate in close proximity to seal haul-out sites in these areas and may disturb seals or deter them from hauling out. Although a certain degree of habituation is likely, it is possible that critical behaviours are disrupted, possibly resulting in high stress levels, energy expenditure and reduced fitness. It is also possible that other Annex II marine mammal species such as the Harbour porpoise are excluded from these areas of high vessel activity due to avoidance responses.

16

Vessel collisions are also a possible source of mortality for some seabirds, particularly shearwaters and petrels which are attracted to lights at night (Wiese et al., 2001; Le Corre et al., 2002)

vii) Wind farming At least 13,000 turbines are currently planned for northeast Atlantic marine waters, although only ten marine wind farms with a total of 163 turbines are operating worldwide. In Irish waters one large marine wind farm is operational on the Arklow Bank, off County Wicklow in the Irish Sea and another is proposed for Clogher Head, off County Louth. The impacts of these wind farms on seabirds are poorly studied but are likely to include: death by collision with turbines, as well as other effects resulting from flight avoidance, habitat modification, feeding displacement and disturbance (Anonymous, 2003a). At present the RAMSSI study site is free from any wind farm developments. viii) Mariculture Global seafood demand is projected to increase by 70% by the year 2025. Aquaculture will have to increase production by 700% to a total of 77million metric tonnes annually to meet that projected demand (Hayden, 2000). In the northeast Atlantic and Baltic Sea, total mariculture production has increased 2½ times over the past two decades to 1.2 million tonnes, divided almost equally between fish and molluscs (Fletcher, 2003).

It is widely acknowledged that seals and seabirds, particularly cormorants, shags, herons, Ardea cinerea and eiders, Somateria mollissima directly benefit from the enhanced food supply provided by marine fin-fish farming as well as from lost fish pellet food (Leukona, 2002; Davenport et al., 2003; Quick et al., 2003; Nash et al., 2000; Boylan et al., 2003). Gulls readily scavenge on remains of fish and invertebrates, steal food pellets and take the growing product from accessible areas of farms (Davenport et al., 2003). However, marine fin-fish farming can also have an indirect negative impact on seabird and marine mammal feeding success through the loss of wild fish stocks. In 2001 an

17 estimated 11 million tonnes of fish were taken from the wild to provide fish-meal for carnivorous farmed fish such as salmon. As aquaculture continues to boom, it will exact a growing toll on species such as sardines, Sardina pilchardus and herring, Clupea harengus (Powell, 2003) and will likely affect the seabirds that prey upon them.

The impact of intertidal mollusc farming on shorebirds (as well as gulls) is relatively well known (Hilgerloh, 1997; Hilgerloh et al., 1997; Ferns et al., 2000; Hilgerloh et al., 2001). Negative impacts have been found for some shorebirds at oyster farms due to loss of open estuarine foraging habitat (Hilgerloh et al., 2001) and over-harvesting of natural seed stocks (Anonymous, 2000). However some species benefit from the readily available prey provided by intertidal mussel farms (Goss-Custard et al., 1993).

Mussel suspension aquaculture occupies large sections of inshore seabird and seal foraging habitat without providing a direct source of food for these primarily-piscivorous species. The influence of this type of mariculture on the seabird and seal community in Bantry Bay has generally been found to be positive or neutral (Roycroft et al., 2004; Roycroft et al., 2006) at its current intensity. However, the mariculture industry is undergoing a rapid expansion in Ireland (Hayden, 2000) and the widespread alteration of inshore foraging habitat by this industry could potentially impact upon the foraging success of seabirds, even if this is not apparent at small scales. According to Gill et al. (2001) species with little suitable habitat elsewhere cannot show marked avoidance of disturbance even if the fitness costs are high. For species that feed on mobile or highly aggregated prey, the costs of moving to alternative sites may be great, especially if they are territorial or experience high levels of competition. Such species could then be forced to tolerate disturbance which may or may not affect survival or fecundity and hence population size (Gill et al., 2001). Therefore it is essential to consistently monitor the impacts of mussel suspension culture on seabirds in order to adequately determine future trends.

Potential impacts of mariculture on cetaceans include death or injury through entanglement in gear, displacement, alteration of the food chain and human

18 persecution. Unlike pinnipeds, cetaceans have not been reported to consume fish or shellfish from farms, but have been known to get entangled in equipment, resulting in the damage of gear, release of fish, and self injury (Kemper & Gibbs, 2001; Hall & Donovan, 2002). Displacement of cetaceans by aquaculture may also occur because they frequently share the same coastal habitat. Studies of bottlenose dolphin movements around oyster farms in Shark Bay, Western Australia showed that dolphins were less likely to go into areas where farming was occurring compared to an ecologically similar area nearby (Watson-Capps & Mann, 2005). Similar avoidance behaviour was observed in dusky dolphins at mussel farming areas of the Marlborough Sounds, New Zealand (Markowitz et al., 2004). For this reason, baseline information on marine mammal distribution is needed in order to assess future impacts of this rapidly expanding industry.

Bantry Bay is the largest mussel producing bay in Ireland, with four fish farms, 13 oyster farms and over 54 mussel farms mainly concentrated in the inner bay (BIM, 2001). Aquaculture is increasing in importance in Kenmare River with extensive areas devoted to mussel suspension culture, especially sheltered areas such as Kilmakilloge Harbour, Ardgroom Harbour, Coongar Harbour and outer Sneem Harbour. There is also a new fin-fish farm development on the northern shore. ix) Fisheries The effects of fishing on birds may be direct or indirect. Most direct effects involve mortalities from fishing gear e.g. bycatch of albatross and petrel species in long-lines in the north Pacific and southern ocean (Brothers et al., 1999; Furness, 2003). Longlining occurs along the shelf edges of Ireland, possibly causing fulmar mortalities (Brothers et al., 1999) however seabird bycatch impacts in these areas tend to be of a localised nature, diluting any possible population effect (Tasker et al., 2000). Drift nets (banned since 2000 in the Atlantic) and other gillnets have had a considerable impact on seabirds in the northern Pacific and northwest Atlantic (King, 1984). Inshore fixed gillnets may be a source of considerable mortality for pursuit-diving seabirds, especially if set close to large breeding colonies. Bycatch of seabirds in salmon nets also occurs in Ireland and has been associated with population declines at some auk colonies

19 during the 1970’s and 1980’s (Lloyd et al., 1991). Some seabird mortality is also caused by lost nets and lines, however these impacts are generally low (Montevecchi, 2002).

Gannets feeding on discarded blue whiting (Photo. M. Mackey).

Indirect effects of fisheries on seabirds mostly work through the alteration of food supplies. Fishing activities have led to depletion of some fish species (e.g. lesser sandeels, Ammodytes marinus) fed upon by seabirds but may also lead to an increase in small fish prey available to seabirds by reducing numbers of large predatory fish (Furness & Tasker, 2000; Tasker et al., 2000; Furness, 2003) or by provision of offal and discards (Hillis, 1971; Hudson & Furness, 1988; Tasker et al., 2000). Increased populations of some scavenging seabirds (e.g. large gulls and fulmars) have been attributed to the availability of discards (Furness, 1999; Garthe et al., 1999), with an estimated 5.9 million scavenging individuals potentially supported by fishery waste in the North Sea (Garthe et al., 1996). Thus, it is not surprising that the recent declines in discard rates (Alverson et al., 1994; Kelleher, 2004). have resulted in an apparent decline in breeding success, population size and body condition of some scavenging species (see Furness, 2003). This reduction in discard availability can also result in an increase in predation pressure on smaller birds by large scavenging species such as great skuas, Stercorarius skua, particularly in the North Sea (Furness, 2003; Votier et

20 al., 2004). Therefore, declines in fisheries discards may not only affect scavenging seabird populations but also a wide range of smaller seabird species.

There has been widespread concern for many years on the impact of fisheries on marine mammals (Northridge, 1984; Northridge, 1991). Incidental capture (by-catch) of cetaceans in fishing gear causes high mortalities of some species, particularly Harbour porpoises and dolphins in areas of high fishing activity such as the Celtic Sea and continental shelf. Different species of cetacean are at risk from different fisheries, depending on the fishing gear and techniques used. Passive gear, especially gillnets generally kill more marine mammals than actively fished gear. Marine debris, especially derelict fishing gear, is also responsible for substantial incidental mortality of marine mammals (Laist et al., 1999).

Harbour porpoises are particularly susceptible to entanglement in bottom set gillnets with an estimated annual mortality of 2200 porpoises (95% C.I. 900– 3500) in the Celtic Sea gillnet fishery off southern Ireland (Tregenza et al., 1997). This is 6.2% of the estimated number of porpoises in the Celtic Sea and there is serious cause for concern about the ability of the population to sustain this level of by-catch. Bottlenose dolphins can be assumed to be at risk from the same fishing gear while common dolphins appear particularly susceptible to being caught in pelagic (mid-water) trawl gear (DEFRA, 2004).

A study of pelagic trawl fisheries by-catch in the northeast Atlantic recorded mortalities of white-sided dolphin, Lagenorhynchus acutus, common dolphin and grey seal as well as a probable record of bottlenose dolphin. This amounted to a mortality rate of 1 marine mammal every 17 tows (or 1 per 80.6 h of towing) with all dolphin by-catches occurring at night (Morizur et al., 1999).

The UK Department of the Environment, Food and Rural Affairs has recommended the use of pingers (acoustic deterrents) in selected fisheries including vessels of the Celtic Sea gillnet fishery operating 6 nautical miles or more from the coast and using bottom-set gill nets (DEFRA, 2003; DEFRA, 2004). However the effectiveness of these measures are poorly understood.

21

Seals are generally perceived to benefit from inshore fisheries with reports of fish theft, particularly from salmon drift nets by seals commonplace. For this reason seals are regarded as pests by many fishermen and localized culls are occasionally carried out illegally to reduce damage to fish stocks from seals, e.g c. 50 seals culled on Beginish Island in November 2004 (www.marinetimes.ie).

Bantry Bay is the site of the country’s second largest fishing harbour, Castletownbere. There are over 100 fishing vessels based in Castletownbere, of which over 50 are more than 40 foot in length. In 1998 the principle demersal species landed in Castletownbere were whiting, Merlangius merlangus, haddock, Melangrammus aeglefinus, monkfish, Squatina squatina, megrim, Lepidorhombus whiffiagonis, hake, Merluccius spp. and cod Gadus spp. and the pelagic catch was dominated by mackerel, Scomber scombrus (http://bantrybaycharter.ucc.ie). Bantry and Castletownbere ports together landed 5,927 tonnes live weight of sea fish in 2002 (Anonymous, 2003b). The bulk of the fishing effort from this fleet takes place in offshore regions, outside of Bantry Bay however.

1.5 AIMS AND OBJECTIVES

The aims and objectives of this study were to:

i) identify significant determinants of seabird distribution in Bantry Bay using physical habitat characteristics (Chapter 1), ii) establish the spatio-temporal distribution of seabirds in southwest Ireland using shore-based observation points (Chapter 2), iii) examine seasonal changes in harbour seal abundance and haul-out site use in southwest Ireland (Chapter 3), iv) assess the haul-out behaviour and habitat use of individual seals using telemetry (Chapter 4), v) investigate seasonal variations in cetacean abundance and distribution in southwest Ireland (Chapter 5).

22 REFERENCES

Alheit, J. & Hagen, E. 1997. Long-term climate forcing of European herring and sardine populations. Fisheries Oceanography 6, 130-139. Alverson, D.L., Freeberg, M.H., Murawski, S.A. & Pope, J.G. 1994. A global assessment of fisheries bycatch and discards. FAO Fisheries Technical Paper, No. 339. pp 233. Anonymous 2000. Report of the Working Group on Seabird Ecology. International Council for the Exploration of the Sea, Copenhagen. pp 72. Anonymous 2003a. Report of the Working Group on Seabird Ecology. International Council for the Exploration of the Sea, Oceanography Committee, Copenhagen. pp 92. Anonymous 2003b. Department of the Marine and Natural Resources, Fishery Statistics 2002. Central Statistics Office (CSO), www.cso.ie. Ref 188/2003., Skehard Road, Cork, Ireland. Anonymous 2005. Statistics of Port Traffic 2004. Central Statistics Office (CSO), www.cso.ie. Ref 110/2005., Skehard Road, Cork, Ireland. Becker, P.H. 1991. Population and contamination studies in coastal birds: the common tern Sterna hirundo. In: C.M. Perrins, J.-D. Lebreton & G.J.M. Hirons (eds), Bird Population Studies. Oxford University Press Berge, J.A., Brevik, E.M., Bjorge, A., Folsvik, N., Gabrielsen, G.W. & Wolkers, H. 2004. Organotins in marine mammals and seabirds from Norwegian territory. Journal of Environmental Monitoring 6, 108-112. Berrow, S. & Rogan, E. 1997. Review of cetaceans stranded on the Irish Coast, 1901-95. Mammal Review 27, 51-76. Berrow, S., Mchugh, B., Glynn, D., Mcgovern, E., Parsons, K.M., Baird, R.W. & Hooker, S.K. 2002. Organochlorine concentrations in resident bottlenose dolphins (Tursiops truncauts) in the Shannon estuary, Ireland. Marine Pollution Bulletin 44, 1296-1313. Boylan, P., Crozier, W.W., McGinnity, P. & O'Maoileidigh 2003. Seals/Atlantic Salmon Interaction Workshop. A recent Irish Review of the evidence. The Loughs Agency of the Foyle, Carlingford and Irish Lights Commission, TSO Ireland. pp 80.

23 Briggs, K.T., Gershwin, M.E. & Anderson, D.W. 1997. Consequences of petrochemical ingestion and stress on the immune system of seabirds. ICES Journal of Marine Science 54, 718-25. Brothers, N.P., Cooper, J.P. & Lokkeborg, S. 1999. The incidental catch of seabirds by longline fisheries: worldwide review and technical guidelines for mitigation. FAO Fisheries Circular 937, Rome. Bustnes, J.O., Erikstad, K.E., Skaare, J.U., Bakken, V. & Mehlum, F. 2003. Ecological effects of organochlorine pollutants in the Arctic: A study of the Glaucous Gull. Ecological Applications 13, 504-515. Charif, R., Clapham, P.J. & Clarke, C. 2001. Acoustic detections of singing humpback whales in the deep waters off the British Isles. Marine Mammal Science 17, 751-769. Colhoun, K. 2002. Waterbird Monitoring in Ireland 2000/1: results of the seventh year of the Irish Wetland Bird Survey (I-WeBS). Irish Birds 7, 43-52. Cramp, S., Bourne, W.R.P. & Saunders, D. 1974. The seabirds of Britain and Ireland. Collins, London Cronin, M., Duck, C., O'Cadhla, O., Nairn, R., Strong, D. & O'Keeffe, C. 2004. Harbour seal population assessment in the Republic of Ireland: August 2003. Irish Wildlife Manuals, No. 11. National Parks and Wildlife Service, Department of the Environment, Heritage and Local Government, Dublin, Ireland. pp 39. Cronin, M. & O'Cadhla, O. 2004. Aerial surveying of grey seal breeding colonies on the Blasket Islands, Co. Kerry, the Inishkea Group, Co. Mayo and the Donegal coast, during the 2003 breeding season. National Parks and Wildlife Service, Dept. of Environment, Heritage and Local Government, Dublin. Cross, T., Southgate, T. & Myers, A.A. 1979. The initial pollution of shores in Bantry Bay, Ireland, by oil from the Tanker Betelgeuse. Marine Pollution Bulletin 10, 104-107. Davenport, J., Black, K., Burnell, G., Cross, T., Culloty, S., Ekaratne, S., Furness, R.W., Mulcahy, M. & Thetmeyer, H. 2003. Aquaculture: the ecological issues. British Ecological Society. Blackwell Science Ltd. pp 89.

24 DEFRA 2003. UK Small cetacean by-catch response strategy. Department of the Environment, Food and Rural Affairs; The Scottish Executive; The Welsh Assembly Government and the Department of Agriculture and Rural Development in Northern Ireland, Crown Copyright, London. pp 33. DEFRA 2004. Caught in the net: by-catch of dolphins and porpoises off the UK coast. Third Report of Session 2003-04. House of Commons, Environment, Food and Rural Affairs Committee, London. pp 44. Dirksen, S., Boudewijn, T.J., Slager, L.K., Mes, R.G., Van Schaick, M.J.M. & de Voogt, P. 1995. Reduced breeding success of Cormorants (Phalacrocorax carbo sinensis) in relation to persistent organochlorine pollution of aquatic habitats in The Netherlands. Environmental Pollution 88, 119-132. Edwards, A., Jones, K., Graham, J.M., Griffiths, C.R., MacDougall, N., Patching, J., Richard, J.M. & Raine, R. 1996. Transient Coastal upwelling and water circulation in Bantry Bay, a Ria on the Southwest Coast of Ireland. Estuarine, Coastal and Shelf Science 42, 213-230. Ekker, M., Lorentsen, S.H. & Rov, N. 1992. Chronic oil-fouling of grey seal pups at the Froan breeding ground, Norway. Marine Pollution Bulletin 24, 92-93. Ferns, P.N., Rostron, D.M. & Siman, H.Y. 2000. Effects of mechanical cockle harvesting on intertidal communities. Journal of Applied Ecology 37, 464-474. Fletcher, N.E. 2003. ICES 2003. Environmental status of the European Seas. German Federal Ministry for the Environment, Nature Conservation and Nuclear Safety. pp 75. Furness, R.W., Ensor, K. & Hudson, A.V. 1992. The use of fishery waste by gull populations around the British Isles. Ardea 80, 105-113. Furness, R.W. 1999. Will reduced discarding help or harm seabird populations? In, Ecosystem Approaches for Fisheries Management. Alaska Sea Grant College Program AK-SG-99-01, Fairbanks. pp 481-488. Furness, R.W. & Tasker, M. 2000. Seabird-fishery interactions: quantifying the sensitivity of seabirds to reductions in sandeel abundance, and

25 identification of key areas for sensitive seabirds in the North Sea. Marine Ecology Progress Series 202, 253-264. Furness, R.W. 2003. Impacts of fisheries on seabird communities. Scientia Marina 67, 33-45. Garthe, S., Camphuysen, C.J. & Furness, R.W. 1996. Amounts of discards by commercial fisheries and their significance as food for seabirds in the North Sea. Marine Ecology Progress Series 136, 1-11. Garthe, S., Walter, U., Tasker, M.L., Becker, P.H., Chapdelaine, G. & Furness, R.W. 1999. Evaluation of the role of discards in supporting bird populations and their effects on the species composition of seabirds in the North Sea. In, Diets of Seabirds and Consequences of Changes in Food Supply. ICES Cooperative Research Report No. 232, Copenhagen Geraci, J.R. & St. Aubin, D.J. 1990. Sea Mammals and Oil: Confronting the Risks. Academic Press, San Diego Gibbs, P.E., Bryan, G.W., Pascoe, P.L. & Burt, G.R. 1987. The use of the dog-whelk, Nucella lapillus, as an indicator of tributyltin (TBT) contamination. Journal of the Marine Biological Association of the United Kingdom 67, 507-523. Gill, J.A., Norris, K. & Sutherland, W.J. 2001. Why behavioural responses may not reflect the population consequences of human disturbance. Biological Conservation 97, 265-268. Goss-Custard, J.D., West, A.D. & Dit Durell, S.E.A.L. 1993. The availability and quality of the mussel prey (Mytilus edulis) of oystercatchers (Haematopus ostralegus). Netherlands Journal of Sea Research 31, 419- 439. Guruge, K.S., Tanabe, S., Iwata, H., Taksukawa, R. & Yamagishi, S. 1996. Distribution, biomagnification, and elimination of butyltin compound residues in common cormorants (Phalacrocorax carbo) from Lake Biwa, Japan. Archives of Environmental Contamination and Toxicology 31, 210-217. Hall, M.A. & Donovan, G.P. 2002. Environmentalists, fisherman, cetaceans, and fish: is there a balance and can science help to find it? In: P.G.H. Evans & J.A. Raga (eds), Marine Mammals: Biology and Conservation. Kluwer Academic/Plenum Publishers, New York. pp 491–521.

26 Hammond, P.S., Heimlich, S., Benke, H., Berggren, P., Borchers, D.L., Buckland, S.T., Collet, A., Heide-Jorgensen, M.P., Hiby, A.R., Leopold, M.F. & Oien, N. 2002. Distribution and abundance of the Harbour porpoise and other small cetaceans in the North Sea and adjacent waters. Journal of Applied Ecology 29, 361 - 376. Harwood, J. & Wilson, B. 2001. The implications of developments on the Atlantic Frontier for marine mammals. Continental Shelf Research 21, 1073–1093. Hayden, J. 2000. Aquaculture in the next millennium. The journal of the Irish Aquaculture Association, Aquaculture Ireland 88. Helander, B., Olsson, A., Bignert, A., Asplund, L. & Litzen, K. 2002. The role of DDE, PCB, coplanar PCB and eggshell parameters for reproduction in the white-tailed sea eagle (Haliaeetus albicilla) in Sweden. Ambio 31, 386-403. Hilgerloh, G. 1997. Predation by birds on blue mussel Mytilus edulis beds of the tidal flats of Spiekeroog (southern North Sea). Marine Ecology Progress Series 146, 61-72. Hilgerloh, G., Herlyn, M. & Michaelis, H. 1997. The influence of predation by herring gulls Larus argentatus and oystercatchers Haematopus ostralegus on a newly established mussel Mytilus edulis bed in autumn and winter. Helgoländer Meeresuntersuchungen 51, 173-189. Hilgerloh, G., O`Halloran, J., Kelly, T.C. & Burnell, G.M. 2001. A preliminary study on the effects of oyster culturing structures on birds in a sheltered Irish Estuary. Hydrobiologia 465, 175-180. Hillis, J.P. 1971. Seabirds scavenging at trawlers in Irish waters. Irish Naturalists Journal 17, 129-132. Hudson, A.V. & Furness, R.W. 1988. Utilization of discarded fish by scavenging seabirds behind whitefish trawlers in Shetland. Journal of Zoology (London) 215, 151-166. Hutchinson, C.D. 1989. Birds in Ireland. T & A D Poyser, Calton. pp 215. Kannan, K., Senthilkumar, K., Elliot, J.E., Feyk, L.A. & Giesy, J.P. 1998. Occurrence of Butyltin Compounds in Tissues of Water Birds and Seaducks from the United States and Canada. Archives of Environmental Contamination and Toxicology 35, 64-69.

27 Katona, S.K. & Kraus, S.D. 1999. Efforts to conserve the North Atlantic right whale. In: J.R. Twiss Jr & R.R. Reeves (eds), Conservation and Management of Marine Mammals. Smithsonian Istitution Press, Washington DC. pp 342-366. Kelleher, K. 2004. Discards in the world's marine fisheries: an update. FAO Fisheries Technical Paper (Draft), 470. pp 134. Kemper, C.M. & Gibbs, S.E. 2001. Dolphin interactions with tuna feedlots at Port Lincoln, South Australia and recommendations for minimising entanglements. Journal of Cetacean Resource Management 3, 283–292. Ketten, D.R. 1993. Blast injury in humpback whale ears: Evidence and implications (A). The Journal of the Acoustical Society of America 94, 1849-1850. King, W.B. 1984. Incidental mortality of seabirds in gillnets in the North Pacific. In: J.P. Croxall, P.G.H. Evans & R.W. Schreiber (eds), Status and Conservation of the World's Seabirds. ICBP Tech. Publication No. 2, Cambridge Laist, D.W., Coe, J.M. & O'Hara, K.J. 1999. Marine debris pollution. In: J.R. Twiss Jr & R.R. Reeves (eds), Conservation and Management of Marine Mammals. Smithsonian Istitution Press, Washington DC. pp 342-366. Larner, J. & Douglas, J., (eds). 2002. Special Protection Areas for Birds in Ireland. Duchas, The Heritage Council, Dublin, Ireland. pp 165. Le Corre, M., Ollivier, A., Ribes, S. & Jouventin, P. 2002. Light-induced mortality of petrels: a 4-year study from Réunion Island (Indian Ocean). Biological Conservation 105, 93-102. Lee, A.J. & Ramster, J.W. 1981. Atlas of the Seas around the British Isles. Lowestoft: Ministry of Agriculture, Fisheries and Food. Leukona, J.M. 2002. Food intake, feeding behaviour and stock losses of cormorants, Phalacrocorax carbo, and Grey herons, Ardea cinerea, at a fish farm in Arcachon Bay (Southwest France) during breeding and non- breeding season. Folia Zoologica 51, 23-34. Lloyd, C., Tasker, M.L. & Partridge, K. 1991. The status of Seabirds in Britain and Ireland. T & A D Poyser Ltd., London. pp 355. Mackey, M., O'Cadhla, O., Kelly, T.C., Aguiler de Soto, N. & Connolly, N. 2004a. Cetaceans and Seabirds of Ireland's Atlantic Margin. Volume 1 -

28 Seabird distribution, density and abundance. Report on research carried out under the Irish Infrastructure Programme (PIP): Rockall Studies Group (RSG) projects 98/6 and 00/13, Porcupine Studies Group project P00/15 and Offshore Support Group (OSG) project 99/38, Cork. pp 95. Mackey, M., O'Cadhla, O., Kelly, T.C., Aguiler de Soto, N. & Connolly, N. 2004b. Cetaceans and Seabirds of Ireland's Atlantic Margin. Volume 2 - Cetacean distribution and abundance. Report on research carried out under the Irish Infrastructure Programme (PIP): Rockall Studies Group (RSG) projects 98/6 and 00/13, Porcupine Studies Group project P00/15 and Offshore Support Group (OSG) project 99/38, Cork. pp 89. MacLeod, C.D., Begona Santos, M., Reid, R.J., Scott, B.E. & Pierce, G.J. 2007. Linking sandeel consumption and likelihood of starvation in harbour porpoises in the Scottish North Sea: could climate change mean more starving porpoises? Biology Letters doi:10.1098/rsbl.2006.0588. Markowitz, T.M., Harlin, A.D., Wursig, B. & Mcfadden, C.J. 2004. Dusky dolphin foraging habitat: overlap with aquaculture in New Zealand. Aquatic Conservation - Marine and Freshwater Ecosystems 14, 133-149. Mitchell, I.M., Newton, S.F., Ratcliffe, N. & Dunn, T.E. 2004. Seabird Populations of Britain and Ireland. Results of the Seabird 2000 Census (1998-2002). T & AD Poyser, London. pp 511. Montevecchi, W.A. & Myers, R.A. 1997. Centurial and decadal oceanographic influences on changes in northern gannet populations and diets in the north-west Atlantic: implications for climate change. ICES Journal of Marine Science 54, 608-614. Montevecchi, W.A. 2002. Interactions between fisheries and seabirds. In: E.A. Schreiber & J. Burger (eds), Biology of Marine Birds. CRC Press, Boca Raton Morizur, Y., Berrow, S.D., Tregenza, N.J.C., Couperus, A.S. & Pouvreau, S. 1999. Incidental catches of marine-mammals in pelagic trawl fisheries of the northeast Atlantic. Fisheries Research 41, 297-307. Nash, C.E., Iwamoto, R.N. & Mahnken, C.V.W. 2000. Aquaculture risk management and marine mammal interactions in the Pacific Northwest. Aquaculture 183, 307-323.

29 Newton, S., Donaghy, A., Allen, D. & Gibbons, D. 1999. Birds of Conservation Concern in Ireland. Irish Birds 6. Northridge, S.P. 1984. World review of interactions between marine mammals and fisheries. FAO Fisheries Technical Paper, 251 Northridge, S.P. 1991. World review of interactions between marine mammals and fisheries. FAO Fisheries Technical Paper, 251, Supplement 1. O'Sullivan, W.M., Macdonald, S.M. & Mason, C.F. 1993. Organochlorine pesticide residues and PCBs in otter spraints from southern Ireland. Biology and Environment: Proceedings of the Royal Irish Academy 93B, 55-57. Pollock, C., Reid, J., Webb, A. & Tasker, M. 1997. The distribution of seabirds and cetaceans in the waters around Ireland. JNCC Report, No. 267, Peterborough. pp 167. Pollock, C.M. 1994. Observations on the distribution of seabirds off south-west Ireland. Irish Birds 5, 173-182. Pollock, C.M., Mavor, R., Weir, C.R., Reid, A., White, R.W., Tasker, M.L., Webb, A. & Reid, J.B. 2000. The distribution of Seabirds and Marine Mammals in the Atlantic Frontier, North and West of Scotland. Joint Nature Conservation Committee Powell, K. 2003. Fish farming: eat your veg. Nature 426, 378-379. Quick, N.J., Middlemas, S.J. & Armstrong, J.D. 2003. A survey of antipredator controls at marine salmon farms in Scotland. Aquaculture 230, 169-180. Raine, R., O`Mahony, J., McMahon, T. & Roden, C. 1990. Hydrography and Phytoplankton of waters off Southwest Ireland. Estuarine, Coastal and Shelf Science 30, 579-592. Raine, R., Joyce, B., Richard, J., Pazos, Y., Moloney, M., Jones, K.J. & Patching, J.W. 1993. The development of a bloom of the dinoflagellate [Gyrodinium aureolum (Hulbert)] on the Southwest Irish Coast. ICES Journal of Marine Science 50, 461-469. Raine, R. & Joyce, B. 1996. The Summer Phytoplankton Ecology of Waters off Southwestern Ireland. In: B.F. Keegan & R. O'Connor (eds), Irish Marine Science 1995. Galway University Press, Galway. pp 131-142.

30 Raine, R. & McMahon, T. 1998. Physical dynamics on the continental shelf off southwestern Ireland and their influence on coastal phytoplankton blooms. Journal of Continental Shelf Research 18, 883-914. Rejinders, P.J.H. 1986. Reproductive failure in common seals feeding on fish from polluted coastal waters. Nature 324, 456-457. Richardson, W.J., Greene Jr., C.R., Malme, C.I. & Thomson, D.H. 1995. Marine Mammals and Noise. Academic Press, San Diego Rogan, E. & Berrow, S. 1995. The management of Irish waters as a whale and dolphin sanctuary. In: A.S. Blix, L. Walloe & O. Ulltang (eds), Developments in Marine Biology, 4. Whales, seals, fish and man. Elsevier, Amsterdam. pp 671-683. Rogan, E., Ingram, S., Holmes, B. & O`Flanagan, C. 2000. A survey of bottlenose dolphins (tursiops truncatus) in the Shannon Estuary. Marine Institute of Ireland, Dublin Roycroft, D., Kelly, T.C. & Lewis, L.J. 2004. Birds, seals and the suspension culture of mussels in Bantry Bay, a non-seaduck area in Southwest Ireland. Estuarine Coastal and Shelf Science 61, 703-712. Roycroft, D., Kelly, T.C. & Lewis, L.J. 2006. Behavioural interactions of seabirds with suspended mussel longlines. Aquaculture International (In Press, now available online at www.springer.com). Sandvik, H., Erikstad, E., Barrett, R.T. & Yoccoz, G. 2005. The effect of climate on adult survival in five species of North Atlantic seabirds. Journal of Animal Ecology 74, 817-831. Smiddy, P. 1992. The effect of the Kowloon Bridge oil spill in east Cork. Irish Birds 4, 559-570. Smiddy, P. 1998. Cormorant Phalacrocorax carbo breeding numbers in Waterford, east Cork and Mid Cork. Irish Birds 6, 213-216. Smith, J. & McLaverty, A. 1997. The South West coast of Ireland. An Environmental Appraisal. BHP, Chevron, Marathon, Occidental, Statoil and Total, Ireland. pp 64. Smyth, M., Berrow, S.D., Nixon, E. & Rogan, E. 2000. Polychlorinated biphenyls and organochlorines in by-caught harbour porpoises Phoecena phocoena and common dolphins Delphinus delphis from Irish coastal waters. Biology and Environment 100B, 85-96.

31 Tasker, M., Webb, A., Harrison, N.M. & Pienkowski, M.W. 1990. Vulnerable concentrations of marine birds west of Britain. Nature Conservancy Council, Peterborough Tasker, M.L., Camphuysen, C.J., Cooper, J., Garthe, S., Montevecchi, W.A. & Blaber, S.J.M. 2000. The impacts of fishing on marine birds. ICES Journal of Marine Science 57, 531-547. Thompson, P.M. & Ollason, J.G. 2001. Lagged effects of ocean climate change on fulmar population dynamics. Nature 413, 417-420. Tregenza, N.J.C., Berrow, S.D., Hammond, P.S. & Leaper, R. 1997. Harbour Porpoise (Phocoena phocoena L.) by-catch in set gillnets in the Celtic Sea. ICES Journal of Marine Science 54, 896-904. Van der Toorn, J.D. 1990. The seal epidemic in Europe and its consequences. Soundings 15, 1-5. Votier, S.C., Furness, R.W., Bearhop, S., Crane, J.E., Caldow, R.W.G., Catry, P., Ensor, K. & Hamer, K.C. 2004. Changes in fisheries discard rates and seabird communities. Nature 427, 727-730. Watson-Capps, J.J. & Mann, J. 2005. The effects of aquaculture on bottlenose dolphin (Tursiops sp.) ranging in Shark Bay, Western Australia. Biological Conservation 124, 519-526. Webb, A., Stronach, A., Tasker, M.L. & Stone, C.J. 1995. Vulnerable Concentrations of Seabirds south and West of Britain. Joint nature Conservation Committee Wiese, F.K., Montevecchi, W.A., Davoren, G.K., Huettmann, F., Diamond, A.W. & Linkee, J. 2001. Seabirds at Risk around Offshore Oil Platforms in the North-west Atlantic. Marine Pollution Bulletin 42, 1285-1290. Williams, J.M., Tasker, M.L., Carter, I.C. & Webb, A. 1994. A method of assessing seabird vulnerability to surface pollutants. Ibis 137, 147-152.

32 APPENDIX

Appendix 1 Breeding seabird numbers at major seabird colonies in southwest Ireland. Seabird 2000 results (Mitchell et al., 2004 and Newton pers. comm.). Important Blasket Puffin Skellig Scarriff & Bull & Species/pop Islands Island Rocks Deenish Cow Northern Fulmar 1305 447 761 385 1683 E. Storm Petrel 51,6911 5,177 9,994 6,2001 3,5002 Manx Shearwater 19,534 6,329 738 2,311 Northern Gannet 29,683** 3,694** LBBG 439 139 97 GBBG 123 B-L. Kittiwake 336 25 694 3563 Guillemot * 471 92 1422 14303 Razorbill * 511 35 386 1323 Atlantic Puffin 389 5125 4000 2584 Other (<100 AT, BG, SH, HG, SH, HG, SH, HG, SH, MG pairs) SH, MG, GBBG LBBG, GBBG HG GBBG MG = Mew Gull, SH = European Shag, LBBG= Lesser black-backed gull, HG = Herring Gull, GBBG = Great black-backed gull, AT = Arctic Tern, BG = Black Guillemot * Number of individuals, **2004 census, (Newton pers. comm.) 1counts for the islands of Inishnabro and Inishtearaght were estimated from previous surveys (not counted in seabird 2000) 2mid-points of estimates from previous surveys, (not counted in seabird 2000) 31993 counts (Newton pers. comm.) 41994 count from Bull and 1970 count from Cow (Newton pers. comm.)

33 CHAPTER 1.

SEABIRD DISTRIBUTION AND HABITAT USE IN

BANTRY BAY, SOUTHWEST IRELAND.

34 1.1 ABSTRACT

The inshore waters of southwest Ireland provide foraging grounds for the highest concentrations of breeding seabirds in the country, augmented by annual movements of passage migrants. However, the fine-scale habitat use of these seabirds, many of which are of conservation concern, is poorly understood in this region. Reliable predictors of seabird distribution in inshore areas would provide a valuable tool for the designation of marine protection areas for their populations and for their conservation in the event of a major pollution event. Habitat use of seabirds at sea was investigated using generalized linear and generalized additive modelling of densities in relation to a number of physical habitat variables in Bantry Bay, southwest Ireland. Seaward distance (the distance from the most inshore point of the bay) was the most important determinant of seabird distribution in the bay with shearwater (Puffinus sp.), auk (Alca torda & Uria aalge) and kittiwake (Rissa tridactyla) density all increasing with increasing distance from the inner bay. Distance from the nearest coast was also an important variable for many species, while shags (Phalacrocorax aristotelis) and cormorants (Phalacrocorax carbo) were influenced only by depth. There was no significant relationship between large gulls (Larus sp.) and any of the physical habitat variables. Determinants of seabird distribution (all species combined) differed between winter and summer, with the inner bay of greater relative importance in winter than in summer. The outer bay was identified as a ‘hot-spot’ of seabird distribution in summer, with many offshore species utilizing the rich feeding grounds provided by the adjacent Irish Shelf Front.

1.2 INTRODUCTION

Seabird distribution at sea is largely determined by prey distribution (Veit et al., 1993; Wright & Begg, 1997; Skov et al., 2000; Yen et al., 2004; Ainley et al., 2005). Many studies have documented the relationships between seabirds and large scale oceanographic features such as fronts (e.g. Haney & McGillivary,

35 1985; Begg & Reid, 1997; Durazo et al., 1998; Hoefer, 2000; Spear et al., 2001), offshore eddies (e.g. Ribic et al., 1997) and areas of upwelling (Skov & Durinck, 2000; Ainley et al., 2005). These areas of high prey density accordingly (and predictably) attract high numbers of seabirds. However, at small scales (i.e. 1- 100km), these oceanographic processes are highly variable, both spatially and temporally (Edwards et al., 1996), and are difficult and time-consuming to measure. This means that oceanographic variables may be unsuitable for use in studies of seabird distribution (particularly those repeated over time) in small inshore environments such as Bantry Bay.

Static habitat features such as depth, seabed slope and distance from the coast provide a useful and readily measured representation of inshore seabird habitat for longer-term studies of seabird distribution. These variables, if proved reliable determinants of seabird distribution, could also aid the identification of Marine Protected Areas (MPA’s) for seabirds (see Hyrenbach et al., 2000).

According to Hunt & Schneider (1987) the abundance of individual seabird species at small scales (1-100km) often reflects opportunities to forage at local concentrations of prey. These opportunities are likely to differ between species because of differing foraging strategies and may be influenced by the physical habitat. Many species such as gannets, shearwaters, terns and gulls take fish or crustaceans from high up in the water column, while others such as guillemots and razorbills dive deep in pursuit of prey (Shealer, 2002). Some species such as cormorants, shags and divers take a proportion of benthic prey (e.g. flat-fish, Pleuronectes sp. and crabs, Carcinus sp.) and so are likely to be influenced by sea depth more-so than surface-feeding species. Some marine bird species such as gulls are also partially dependent on terrestrial sources of food and so are likely to be limited in distribution by distance from the coast.

Aims and Objectives The aims and objectives of this study were to: vi) map the distribution and identify ‘hotspots’ of seabird density in Bantry Bay.

36 vii) identify significant determinants of seabird distribution within the study site using available habitat characteristics.

1.3 STUDY SITE

For a full description of Bantry Bay refer the general introduction. The survey area extended from the inner bay, east of Whiddy Island to Sheep’s Head. A minimum of nine seabird species breed in Bantry Bay (Appendix 1.1). Over 100 pairs of Arctic terns (Sterna paradisaea), an Annex 1 species, breed near Whiddy Island and over 50 pairs each of cormorants (Phalacrocorax carbo), shags (Phalacrocorax aristotelis) and black guillemots (Cepphus grylle) also breed in the inner bay. Herring gulls (Larus argentatus) and great black-backed gulls (Larus marinus) breed in small numbers in both inner and outer regions of the Bay. Dursey Island, in the outer bay, holds relatively large numbers (>500 pairs) of northern fulmars (Fulmarus glacialis), black guillemots (79 individuals) and small numbers (<10) of lesser black-backed gulls (Larus fuscus) and razorbills (Alca torda). The Annex 1 species, common tern (Sterna hirundo) also breeds in small numbers (<50 pairs) in the neighbouring Special Area of Conservation (SAC) in Kenmare bay (Mitchell et al., 2004). The bay is also likely to be utilized by foraging seabirds from adjacent colonies, particularly gannets (Morus bassanus), from the Bull and Cow Rocks as well as Manx shearwaters (Puffinus puffinus) and storm petrels (Hydrobates pelagicus) from Scarriff and Deenish or the Skellig Rocks (see Chapter 1, Figure 2).

1. 4 METHODS

1.4.1 Line-transect techniques.

Regular, standardised boat-based surveys of Bantry Bay were conducted throughout the study using a modified version of techniques outlined by Tasker et al. (1984) and Webb & Durinck (1992) for surveying seabirds at sea from ships. The survey platform used for this study was a 5.8m Rigid Inflatable Boat (RIB). To adjust for the reduced eye-height of a RIB-based observer (< 2m)

37 compared to a ship based observer (> 6m) for whom the technique was developed, the transect width was reduced from 300m to 200m.

All birds associated with the water (including flying birds which made contact with the water) within 200m abeam of one side of the boat and forward to the horizon, were recorded as ‘on transect’ (Figure 1). Birds within this 90o arc were further categorised into one of four different divisions depending on their perpendicular distance from the observer (Band A: 0-50m; Band B: 51-100m; Band C: 101-200m; Band D: 200+m). These bands allowed correction factors to be calculated to account for the drop-off in detectability of birds with increasing distance form the observer (see section 1.4.2). Birds in band D were not included in density calculations or data analysis as they were recorded as far as the eye could see. Distances were approximated by eye following a training exercise using the boats’ Global Positioning System (GPS) and marker buoys of known position (this was repeated regularly throughout the study). Flying birds were not recorded.

Figure 1. The survey platform showing the trackline, observer position and transect band widths. Only bird sightings within this 200m transect were included in the density analysis (birds in band D were excluded).

38 55000 Glengarriff Harbour

50000 Whiddy Island

ngs 45000 hi t

r Bere Is. o N 40000

N 35000

Sheeps Head 30000 63000 68000 73000 78000 83000 88000 93000 98000 Eastings

Figure 2. The transect route used in Bantry Bay. The waypoints used to navigate boat surveys are shown in red with boat transect lines in black.

Birds associated with fishing vessels were excluded from all analyses.

Surveys followed an 80km predefined survey route at approximately 20km.h-1. The survey route followed transects between GPS waypoints and consisted of an outward and return leg spaced a minimum of 1km apart (Figure 2). The track of the boat during each survey was recorded automatically every minute using the onboard GPS. Data relating to seabird sightings were recorded using a Dictaphone.

Surveys were initiated in July 2001 and were attempted on a monthly basis (or bi-monthly in Summer) until September 2002. Surveys recommenced in June 2003 and were attempted at least twice a month until September 2003. Boat surveys were conducted on days with Beaufort sea-states of three or less and were not conducted on consecutive days in order to reduce autocorrelation of sightings data.

39 1.4.2 Data preparation.

Habitat Mapping To investigate relationships between bird density and a number of physical habitat variables in Bantry Bay, the survey area was divided into a grid of 1km2 squares. The following five habitat variables were chosen as possible determinants of seabird distribution in the study sites (Figure 3). Depths were taken from UK Hydrographic Office Admiralty Charts (chart no.’s 1840 & 1838) and were reduced to chart datum (the Lowest Astronomical Tide): i) Seabed slope; the difference between maximum and minimum depth (m) in each 1km2 section of the survey area. ii) Minimum Depth; the minimum recorded depth (in metres) of each grid square, iii) Maximum Depth; the maximum recorded depth (in metres) of each grid square, iv) Seaward Distance; the distance in kilometres of each grid square (midpoint) from the most inshore point of the study site (i.e. the head of the bay), v) Distance to Nearest Coast; the distance in metres of each grid square (midpoint) from the nearest coast (including islands).

Density calculation The total number of birds of each species was then calculated for each grid square. To correct for effort (Figure 3f), the number of birds recorded in each grid square (or 1km section of transect) was divided by the number of times the square was surveyed. All grid squares surveyed less than five times were deleted from the database. Mean densities (birds km-2) were then calculated by dividing the mean number of birds observed in each 1km section of transect by the transect area (1km x 0.2km). The resulting densities were then multiplied by the relevant correction factor for that species (see below) to obtain the corrected mean density km-2. These densities were then mapped using ArcView (version 8.1).

40 a) Slope b) Minimum Depth

c) Maximum Depth d) Seaward

e) Nearest Coast f) Effort

N Figure 3 (a-e). Bantry Bay habitat maps showing levels of the five explanatory variables in each 1km grid square surveyed. Figure 3 (f) shows the number of times each grid square was sampled. Squares visited less than five times were excluded.

41

The following formula was used to calculate the correction factors:

X = 4A___ nA+nB+nC

Where: nA= number of birds on the water in transect band A (the first 50m) nB= number of birds on the water in transect band B (50-100m) nC= number of birds on the water in transect band C (100-200m) (adapted from Pollock et al., 2000)

Related species were pooled into taxonomic family groups to increase sample size and data from an identical study in the Shannon Estuary (Roycroft, 2005) were included for calculation of correction factors (again to improve sample size). Table 1 shows the correction factors calculated for this study compared with those from two ship-based (as opposed to RIB-based) studies.

Table 1. Correction Factors calculated for a 200m wide strip transect in sea states of <= 3 on Beaufort Scale and with an observer eye-height of < 2m above sea level. Also shown are correction factors from two ship-based surveys; Stone et al. (1995) and Pollock et al. (2000) for a 200m wide strip transect. This study Stone et. Pollock et. Species group al. (1995) al. (2000) Shearwaters 2.9 1.1 1 Gannets 1 1 1 Cormorants & Shags 1.7 1.1 1 Gulls & Kittiwakes 2.3 1.2 1.1 Auks 2.4 1.2 1-*1.3 *in sea states >=3 Beaufort scale.

Preparation for modelling Related species showing similar distributions (when mapped using ArcView 8.1) were grouped together to improve sample size for the purpose of data analysis. Densities were calculated for the following groups: 1) Phalacrocoracidae: great cormorant, Phalacrocorax carbo and northern shag, Phalacrocorax aristotelis,

42 2) Large Laridae (lesser & great black-backed gull, Larus fuscus and Larus marinus and herring gull, Larus argentatus), 3) Kittiwakes, Rissa tridactyla only, 4) Alcidae: razorbill, Alca torda and common guillemot, Uria aalge, 5) Shearwaters (manx & sooty shearwater, Puffinus puffinus and Puffinus griseus) and 6) Total seabirds (all birds except Larus spp, recorded in the study). Gulls were excluded as they do not depend directly or entirely on the marine environment for food, but utilise a wide range of terrestrial and/or man-made food sources. Thus, members of the Laridae family are likely to be relatively independent of the habitat variables measured in this study. To investigate seasonal variations in total seabird distribution, the data were divided into two groups; Winter (October to March inclusive) and Summer/breeding (April to September inclusive). Effort per grid square was also calculated separately for each season for density calculation. Only the ‘total seabird (excluding gulls)’ dataset was analysed by season as effort was low and bird observations insufficient in Winter for seasonal analysis of individual species groups. (Note shearwaters were only present in Summer).

Maximum species richness per square was also calculated. This comprised the maximum number of species observed per grid square in any one survey. Maximum species richness per grid square was not corrected for effort (no. of times the square was surveyed). Instead, grid squares with low effort were omitted until there was no significant relationship between effort and maximum species richness. Figure 4 shows that an asymptote between mean maximum species richness and effort was reached at an effort of 13 surveys. However, there was no significant relationship between effort and maximum species richness when grid squares surveyed less than nine times were removed (GLM, P>0.05). For this reason, only grid squares with an effort of nine or more were included in the habitat analysis.

43 4 3.5 ss e

n 3 ch

i 2.5 R

. 2 p 1.5 S

m 1 u

m 0.5 i x

a 0 ) ) ) ) ) ) ) ) ) ) ) ) ) M 9) 9) 2) 2) 7) 3) 6 7 6 4 5 6 5 3 5 2 3 6 8 3 1 1 1 1 1 ======an (n (n (n (n (n (n (n (n (n (n (n (n (n e (n= (n= (n= (n= (n= (n= M 1 2 3 4 5 6 7 8 9 10111213141516171819 Effort

Figure 4. The relationship between effort (number of surveys per grid square) and mean maximum species richness in Bantry Bay. n = the number of grid squares within each effort category. Standard error bars are displayed. Only grid squares with an effort of nine or over were included in the species richness habitat analysis.

Similar studies of seabird distribution have shown that correlation between counts from survey segments that are close in space and time can be a problem (Elphick & Hunt, 1993; Begg & Reid, 1997; Clarke et al., 2003; Ainley et al., 2005). In order to carry out accurate statistical analysis, survey segments should be independent of each other. Since bird densities in each grid square of this study area were averaged over a 19-month period, the problem of autocorrelation between neighbouring squares has been minimised.

Linear relationships existed between many of the explanatory variables in Bantry Bay (Pearson product-moment correlation, p<0.01). Any two variables with an r-value of more than 0.8 were considered to be highly co-linear (Alan Zuur pers. comm.) and were not included together in the same model. For this reason only maximum depth or seaward distance was included in Bantry Bay models (r=0.95).

44 Table 2. Pearson’s product-moment correlation coefficients (r) among the physical explanatory variables. Relationships shown are significantly correlated (P<0.01). NS= not significantly correlated. Explanatory Slope Minimum Maximum Seaward Variable Depth Depth Distance Minimum Depth -0.75 Maximum Depth NS 0.68 Seaward Distance NS 0.62 0.95 Nearest Coast -0.57 0.76 0.5 0.47

1.4.3 Data analysis

Data analysis was carried out using R (free software) and the Brodgar statistical package version 2.4.3 (Highland Statistics Ltd.). Generalized linear modelling, GLM (McCullagh & Nelder, 1989) with a Poisson error structure (variance = mean) and log link function was used to relate seabird density to physical habitat variables.

For species with an over-dispersed distribution (i.e. variance > mean) a Quasi- Poisson error structure with log link function was used. That is - a dispersion parameter was introduced, which allowed for more spread than the standard Poisson mean-variance relationship. For each species group, the Minimal Adequate Model was selected using backwards and forwards stepwise selection and the Akaike Information Criterion, AIC statistic (Akaike, 1973).

Generalized additive modelling (Hastie & Tibshirani, 1990) was applied to species that had poor model fits using GLM. This technique investigates non- linear relationships between one or more explanatory variable and the response variable. Poisson or Quasi-Poisson families with log link functions were applied and stepwise model selection using the AIC statistic was carried out in the same way as for the GLM. Cross-validation was used to select the optimal amount of smoothing for each variable. A cubic spline smoothing function was used in this study. The concept of parsimony - that the simplest explanation is best - is inherent in such modelling efforts (Guisan et al., 2002) therefore GLMs were chosen over GAMs when similar results were obtained.

45

1.5 RESULTS

A total of 21 boat surveys of Bantry bay were carried out during the course of the study. Sightings from 19 of these surveys (i.e. 67 survey hours or 1273 kms of transect line) were successfully linked with the GPS track data and included in the data analysis. Of the 178 grid squares surveyed in Bantry Bay, 107 were considered to have sufficient survey repetition (effort >4). Only sightings from these squares were included in the data analysis (apart from the winter analysis as only five surveys were carried out in this season, thus only cells with effort < 2 were excluded). The mean effort per grid square was 13.9 (±0.4 standard error, n = 107 grid squares) and the mode was 18.

1.5.1 Modelling

Significant linear relationships between densities of three seabird groups (Phalacrocoracidae, Alcidae and shearwaters) and a number of habitat variables were found using generalized linear modelling (Table 3). There was no significant relationship (linear or otherwise) between large gull density and any of the habitat variables. This was also the case for maximum species richness. Kittiwakes showed a linear response to seaward distance and curvilinear responses to slope and nearest coast (generalized additive modelling). Total seabird density was positively related to seaward distance and nearest coast. Total seabird densities in Summer showed a positive response to seaward distance and nearest coast, however in Winter total seabird densities showed a negative response to maximum depth and a positive response to nearest coast (both excluding gull species). The percentage deviance explained by the models varied between 30.7% and 64% of the total deviance.

46

Table 3. Summary results of generalized linear and generalized additive models showing determinants of seabird density and species richness n Bantry Bay. Gull species were omitted from ‘total’ bird densities but not sp. richness.

Phal- Large Kitti- Alcidae Shear- Max. Sp. Total Total Total Variable acro. Laridae wakes waters Richness Birds Summer Winter Slope NS S-2 -5 NS Minimum Depth NS NS Maximum Depth - NS NS -2 Seaward Distance NS +1 +1 +1 NS +1 +1 Nearest Coast NS S-3 +2 +2 NS +2 +2 +1 Interaction Terms *3&4 Model GLM GAM GLM GLM GLM GLM GLM Deviance explained 57.5% 0% 55.8% 42% 64% 0% 50.1% 44% 30.7% Dispersion Parameter 0.6 3.98 9.4 10.35 9.63 13.12 7.35

Shown are statistically significant relationships. (+) = positive response of seabird density/biomass/richness to increase in a variable, (-) = negative response (GLM’s). (S+) = significant smoother term with overall positive response, (S-) = overall negative response (GAM’s). Superscripts (1-5) denote the order of importance of the variables in explaining the variation in bird density/richness (1 = most important). NS = model not significant. * indicates that 2 or more of the significant terms have an interaction effect on bird density.

Phalacrocoracidae Table 4. Results of the generalized linear model for cormorants and shags in Bantry Bay. Poisson distribution with a Log-link function. Significant Terms Estimate Std.Error z-value P-value Max. Depth -0.10449 0.01229 -8.502 <0.0001

Maximum Depth was the only significant variable influencing cormorant and shag distribution in Bantry Bay (Table 4). The relationship was negative, indicating that mean cormorant and shag density decreased linearly with increasing depth. Figure 5(a) shows that this species group occurred in relatively high mean densities (4-8/km2) in the shallow, inner portion of the bay (south and east of Whiddy Island) but were not present at all in the deeper regions near the mouth of the bay.

Large Laridae Although appearing in higher mean densities in the outer half of Bantry Bay (11-33/ km2) than in the inner portion (max 6-10/ km2) large gulls were not significantly influenced by seaward distance or any other habitat variable included in the model. Figure 5(b) shows that large gulls were distributed ubiquitously throughout the bay in relatively high mean densities.

Kittiwakes Table 5. Results of the generalized additive model for kittiwakes in Bantry Bay. Quasi-Poisson distribution with a Log-link function. Smoother Terms edf Chi-sq Model r2 P-value Slope 3.6 18.9 0.612 <0.01 Nearest Coast 4.7 16.84 <0.01 Parametric terms Estimate std. Err t-ratio Seaward Distance 0.1971 0.027 7.297 <0.0001

There was a highly significant positive linear relationship between mean kittiwake density and seaward distance in Bantry Bay (Table 5). Slope and nearest coast had significant negative curvilinear effects on mean kittiwake distribution however. Figure 6 (a & b) shows the smoothing curves produced from generalized additive modelling. Mean kittiwake density decreased rapidly with increasing slope in areas where slope was high but remained constant in areas of lower slope (<25m). Kittiwake density decreased with increasing distance from the coast up to a distance of 3500 metres, but then levelled off.

48 a) Phalacrocoracidae b) Large Laridae

c) Kittiwakes d) Alcidae

e) Shearwaters f) Max. Species Richness

Figure 5 (a-f). Distribution of the five seabird groups for which analysis was carried out as well as for maximum species richness in Bantry Bay. Flying birds N are not included. Average densities per transect for the entire study period are shown (includes all seasons).

49 g) Total seabirds (excluding gulls)

h) Summer Totals (excluding gulls) i) Winter Totals (excluding

Figure 5 (g-i). Distribution of total seabird density g) and total seabird density in N Summer h) and Winter i) in Bantry Bay. Flying birds are not included. Average biomass or densities per transect are shown. Gulls are not included.

50 a b

Figure 6. Smoothing curves produced from generalized additive modelling of mean kittiwake density and slope (a) and mean kittiwake density and nearest coast (b). Dotted lines represent confidence intervals. Degrees of freedom are shown in parenthesis on the y-axis label. The vertical lines above the x-axis show positions of the measured data points.

Alcidae Table 6. Results of the generalized linear model for Guillemots and Razorbills in Bantry Bay. Quasi-Poisson distribution with a Log-link function. * interaction terms Significant Terms Estimate Std.Error t-value P-value Slope -3.356e+02 1.510e-02 -2.222 <0.05 Seaward Distance 1.618e-01 3.898e-02 4.152 <0.0001 Nearest Coast 1.919e-03 5.396e-04 3.556 <0.0001 Slope*NearCoast 4.101e-05 1.186e-05 3.458 <0.0001 SeaDist*NearCoast -6.616e-05 2.040e-05 -3.244 <0.01

Mean Alcidae density in Bantry Bay was highly positively correlated with seaward distance and nearest coast but was negatively correlated with slope (Table 6). These relationships were complex however with two interaction terms being significant in the model. Coplots revealed that bird densities decreased slightly with increasing slope at close proximity to the coast (<1km), but increased markedly with slope at higher distances from the coast (>1.5km). From the habitat map (Figure 3a), it can be seen that all areas of high slope (>30m) in Bantry Bay occurred within 1.5km of the coast. Since mean bird density increased significantly with distance from the coast (Table 6), this may have resulted in the appearance of a negative response to slope. Coplots revealed

51 that bird density increased with increasing distance from the coast at all seaward distances, but this relationship was much less marked in the inner bay. Figure 5(d) shows the distribution of Guillemots and Razorbills in Bantry Bay. Highest mean densities (10-60/ km2) occurred in the outer half of the bay in areas with relatively high distance from the coast.

Shearwaters Table 7. Results of the generalized linear model for Shearwaters in Bantry Bay. Quasi-Poisson distribution with a Log-link function. Significant Terms Estimate Std.Error t-value P-value Seaward Distance 0.2741 0.094 2.915 <0.01 Nearest Coast 0.0005 0.0002 2.271 <0.05

Mean shearwater density showed a positive linear relationship with seaward distance and nearest coast (Table 7). Figure 5(e) shows the distribution of shearwaters in Bantry Bay. High mean densities (55-75/km2) occurred only in the outer-most regions of Bantry Bay, with no sightings recorded in the inner bay. Shearwaters typically did not occur, or occurred only in low densities (<15/ km2) at close proximity to the coast.

Maximum Species Richness There was no significant relationship (linear or otherwise) between maximum species richness and any of the explanatory variables. Figure 5(f) shows that maximum species richness was randomly distributed throughout Bantry Bay with a clustering of higher values along the northern side of the bay and near the bay mouth.

52

a) b)

Figure 7. Coplots showing: a) relationships between Alcidae density and slope at increasing distance from the coast and (b) relationships between density and nearest coast at increasing seaward distance. Panels are ordered from lower left to upper right, i.e. the lower left-hand panel of a) shows density (y) versus slope (x) at distances of 0-1000m from the coast and the upper right-hand panel shows density versus slope at distances >2000m. See section 2.2.4. for description of coplots.

Total seabird density (excluding gulls) Table 8(a-c). Results of the generalized linear models for total seabirds (excluding gulls) in Bantry Bay in all seasons (a) in summer only (b) and in winter only (c). Quasi-Poisson distribution with a Log-link function. a) All seasons Significant Terms Estimate Std.Error t-value P-value Seaward Distance 8.237e-02 1.510e-02 5.456 <0.0001 Nearest Coast 3.560e-04 8.172e-05 4.356 <0.0001 b) Summer Significant Terms Estimate Std.Error t-value P-value Seaward Distance 9.853e-02 1.694e-02 5.818 <0.0001 Nearest Coast 2.479e-04 9.635e-05 2.573 <0.05 c) Winter Significant Terms Estimate Std.Error t-value P-value Max depth -0.066 0.019 -3.460 <0.001 Nearest Coast 0.001 1 0.0002 5.207 <0.0001

Total seabird density (excluding gulls) in Simmer and Winter When all seabird species (excluding gulls) in Bantry Bay were pooled across the summer season, there was a positive linear relationship between mean density and seaward distance as well as between mean density and nearest coast (Table 8a). Mean seabird densities in winter (October to March) however showed a negative response to maximum depth (Table 8b). Seabird densities also showed a positive linear relationship to nearest coast in winter, just as in summer. Figure 5(h & i) shows the distribution of all seabird species recorded in Bantry Bay in summer and winter. Highest mean seabird densities (highest 150-230/km2, Figure 5h) occurred in the outer half of Bantry Bay in the summer months. However in winter seabird density was distributed in a more random fashion with a small clustering of higher densities in the inner bay (Figure 5i).

1.5.2 Relative Abundance

Modelling was not carried out on species with low sighting rates. For these species mean density was calculated if sufficient sightings were present (i.e. gannets and black guillemots) but no statistical analysis was carried out.

54 For all other species with low sighting rates, or for species that were only recorded in flight, density was not calculated, and the grid square system was not used. The relative abundance of these species was represented using dot-plots of the original sighting data (Figure 8 a-g). These sightings were not corrected for effort and represent the total number of birds recorded ‘on transect’ over the entire study period.

Great Northern Divers A total of eight great northern divers (Gavia immer) were recorded ‘on transect’ during the course of the study. Figure 8(a) shows the distribution of these sightings, all of which were associated with the water. All sightings occurred within the shallow inner portion of the bay and consisted largely of solitary individuals. Sightings took place between the months of November and April only.

Fulmars All fulmars recorded ‘on transect’ during the study were observed in flight. Figure 8(b) shows that these birds were distributed mainly in the outer regions of Bantry Bay and along its northern side. All sightings were of single flying individuals.

Northern Gannets Figure 8(c) shows the mean density of gannets associated with the water in Bantry Bay. This species was distributed ubiquitously throughout the bay in relatively low mean densities (max 3/km2), with the highest mean density occurring at the mouth of the bay. The overall mean density of gannets in Bantry Bay was 0.16/km² (±0.04se).

Great Skuas A total of seven great skuas (Stercorarius skua) were sighted during the study period. Figure 8(d) shows the distribution of these sightings, all of which were flying individuals. Sightings were restricted to the outer half of Bantry Bay with the majority occurring at the mouth of the bay.

55

Small Laridae Small gulls were distributed mainly within the inner half of Bantry Bay in small groups of less than 10 birds on the water (Figure 8e). These sightings consisted mainly of black-headed gulls (Larus ridibundis) however some birds were unidentified (possibly mew gulls, Larus canus). A large raft (150 individuals) of black-headed gulls was sighted on the southern side of the bay. Flying birds were not included in the sighting map.

Terns Tern sightings were concentrated in the inner portion of Bantry Bay near the arctic tern breeding colony east of Whiddy Island (Figure 8f). All terns recorded were in flight and so were difficult to identify to species level. Most sightings were likely to be arctic terns however common terns were also present as they breed in the neighbouring Kenmare Bay (Mitchell, 2004). All sightings of these Annex 1 species occurred between the months of June and September.

Black Guillemots Mean black guillemot density was highest in the inner portion of Bantry bay in the shallow region south of Whiddy Island (Figure 8g). Mean densities were low with a maximum of 3 per km2. The overall mean density of black guillemots in Bantry Bay was 0.09/km² (±0.04se)

56

a) Great Northern Divers b) Fulmars (flying)

c) Northern Gannets d) Skuas (flying)

e) Small Laridae f) Terns (flying)

Figure 8 (a-f). Total numbers of great northern divers (g), fulmars (h), gannets (i), skuas (j), small gulls (l) and terns (m) recorded during all surveys. Dot-plots N include flying birds.

57 g) Black Guillemots

N Figure 8 (g). Distribution of black guillemots (average density/km2) sighted during the survey period in Bantry Bay.

58 1.6 DISCUSSION

Seaward distance and distance from the nearest coast were the most important determinants of seabird distribution in Bantry Bay, followed by maximum depth and slope.

Of the five seabird groups analysed, three were positively related to seaward distance (i.e. kittiwakes, shearwaters, razorbills and guillemots). Shearwaters, razorbills and guillemots were also positively related to distance from the coast, however kittiwakes showed a negative response to this variable.

Phalacrocoracidae The mean density of cormorants and shags in Bantry Bay was negatively related to maximum depth. Cormorants and shags were recorded in highest densities in waters of less than 30m depth and were never recorded in areas deeper than 45m depth. Pollock et al. (1997) and Mackey et al. (2004) found a complete absence of cormorants and shags in deepwater habitats around Ireland. The rock cormorant (Phalacrocorax magellanicus) was also found to be negatively related to sea depth in the Beagle Channel, Argentina (Raya Rey & Schiavini, 2000). According to Cramp & Simmons (1977) cormorants prefer sheltered seas and avoid deep water even close to land. Both species are foot-propelled pursuit divers, feeding predominantly on fish obtained on or near the seabed (Cramp & Simmons, 1977; Wanless et al., 1991). It is likely therefore that these species minimise energy expenditure by foraging for benthic prey in shallow water. Small numbers of shags (82 pairs) and cormorants (52 pairs) also breed within the inner bay (Appendix 1.1), and so may be restricted by foraging range during the summer months. Mean cormorant & shag density in Bantry Bay was 0.38 ±0.1/km², which was similar to densities found in Irish coastal waters by Pollock et al. (1997). The inner bay held mean densities of 2-8/ km² however, indicating that this high-risk area is locally important for foraging cormorants and shags.

59 Large Laridae There was no significant relationship between large gull density and any of the habitat variables. This group consisted of great black-backed gulls, herring gulls and lesser black-backed gulls in equal proportions (although the majority of sightings were not identified to species level). All of these species are scavengers and food pirates, taking a wide variety of terrestrial and marine prey. Great black-backed gulls are also voracious predators, taking chicks and eggs during the breeding season (Cramp & Simmons, 1983). The distribution of many gull species at sea can be influenced by the location of fishing vessels (Garthe, 1997; Pollock et al., 1997) and therefore is largely independent of physical habitat variables. Pollock et al. (1997) and Mackey et al. (2004) recorded low to moderate densities (typically <2/km2) of these species in Irish coastal waters, with lesser black-backed gulls exhibiting a more offshore distribution than the others. The mean density of large gulls in Bantry Bay was 2.4 ±0.5/km², with the flocking behaviour of these species resulting in local areas of high density (20-60/km2) in the outer half of the bay.

Kittiwakes Kittiwakes are the most pelagic of gull species, dispersing into and even across the North Atlantic during winter (Shealer, 2002). They are surface feeders, taking mostly fish and planktonic invertebrates obtained offshore (Shealer, 2002). The relatively high densities of kittiwakes in the outer half of Bantry Bay may have been feeding on concentrations of prey associated with the nearby Irish Shelf front or from trawler discards just offshore. Pollock et al. (1997) also found high densities of kittiwakes over the shelf break southwest of Ireland and this species is known to associate with a tidal mixing front in the Irish Sea in summer (Durazo et al., 1998). As well as being positively related to seaward distance, kittiwake density was also negatively related to slope and distance from the coast. This was due to the presence of high numbers of kittiwakes (possibly non-breeders) which chose to preen and roost on the steep-sided cliffs of Sheep’s Head during the summer (Roycroft pers. obs.). This headland may have been favoured over others because of its proximity to the rich feeding grounds of the Irish Shelf front. Typical kittiwake densities in the outer half of Bantry Bay were over 5/km2. Previous studies in Irish shelf and inshore waters by Pollock et al.

60 (1997) and Mackey et al. (2004) have found slightly lower densities of kittiwakes (typically <2/ km2, including flying birds) with some areas of the continental slope containing densities of 100/ km2.

Alcidae Razorbill and guillemot density was positively related to seaward distance and distance from the coast. Razorbills and guillemots have a typically inshore distribution, remaining within the boundaries of the Irish Shelf throughout the year (Pollock et al., 1997; Mackey et al., 2004). They are pursuit divers, taking mainly mid-water schooling fish such as sand eels (Ammodytes spp.), herring (Clupea harengus) and sprat (Sprattus sprattus) (Gaston & Jones, 1998). The foraging range of these species is typically less than 30km from the breeding colony (Gaston & Jones, 1998). This may have contributed to the low numbers recorded in the inner portion of Bantry Bay as it is over 40km from the nearest breeding colony on the Bull and Cow Rocks (while the outer bay is within this foraging range). Garthe (1997) also found that guillemots were absent at distances of over 25km from the colony. The prey of this species may also have been higher in the outer bay due to the proximity of the Irish Shelf Front. Razorbills and guillemots have been associated with fronts in the Irish Sea and in the California Current system (guillemots only) for example (Durazo et al., 1998; Hoefer, 2000; Ainley et al., 2005).

Guillemots and razorbills were recorded in higher densities away from the coastline. These species may be wary of foraging close to land due to the presence of predators such as gulls or due to human disturbance. Many auk species are relatively clumsy on land and take off from it with difficulty (Gaston & Jones, 1998) therefore they carry out all maintenance activities at sea or at breeding colonies in the summer. Typical guillemot and razorbill density in the outer half of Bantry bay was between 10 and 60/km2. Previous studies in Irish shelf and inshore waters by Pollock et al. (1997) and Mackey et al. (2004) have found similar densities of razorbills and guillemots (combined) with highest densities occurring in late summer and autumn. That guillemots outnumber razorbills by 13:1 is not surprising as the ratio of breeding guillemots to razorbills on the nearest breeding colony (Bull & Cow Rocks) was 10:1 (General

61 Introduction, Appendix 1). The adult post-breeding moult of guillemots and razorbills is complete, with birds becoming flightless for c. 45-50 days in late summer (Gaston & Jones 1998). For this reason these species are highly vulnerable to surface pollutants at this time of year.

Shearwaters Manx Shearwaters, like most Procellariiformes, are pelagic in distribution for much of the year, only entering inshore Irish waters to breed (manx shearwaters) or forage (sooty shearwaters) in spring and summer (Shealer, 2002; Mackey et al., 2004). Breeding manx shearwaters are known to make foraging trips of some days duration (Gaston & Jones, 1998) with Irish breeding birds often ranging as far as the Porcupine Bank in search of prey (Mackey et al., 2004). Both manx and sooty shearwaters are surface feeders, relying on small fish, cephalopods and crustaceans, many of which are abundant along the Irish Shelf Front (Raine, 1990). The high density of shearwaters at the mouth of Bantry bay may reflect high concentrations of prey associated with the Irish Shelf Front which occurs close to the bay mouth in summer (Edwards et al., 1996). Sooty shearwater distribution in particular has been associated with fronts by a number of authors (Hoefer, 2000; Ainley et al., 2005).

These species are also typically wary of land because of predators such as gulls - only returning to their breeding colonies under the protection of darkness for example (Cramp & Simmons, 1977). It is not surprising therefore that shearwater distribution increased with increasing distance from the coast in this study. Mean shearwater density in the outer bay was very high (typically 15- 75/km2) even when averaged across all seasons. Previous studies in Irish shelf and inshore waters by Pollock et al. (1997) and Mackey et al. (2004) have found much lower densities of manx shearwaters (typically <10/ km2, including flying birds) with sooty shearwater density typically less than 1/km2. This indicates that the outer region of Bantry bay is an important foraging area for these species in summer and autumn.

62 Maximum Species Richness A total of 19 species were recorded in Bantry bay during the course of the study. This is comparable to Pollock's (1994) study, which recorded 18 species in neritic waters around the southwest coast of Ireland in Autumn. Areas of high species richness (max 6/km) were distributed randomly throughout the bay, showing no significant relationship with the physical habitat variables.

Total seabird distribution (excluding gulls) The distribution of total seabird density in Bantry Bay was positively correlated with seaward distance and distance from the coast. Highest seabird density (30- 135/km²) occurred in the outer bay, indicating that prey abundance is high in this region and can supply the energy demands of high densities of seabirds. The high densities of guillemots, razorbills and shearwaters in this study contribute largely to the distribution of these variables. All of these species are highly vulnerable to surface pollutants, indicating that these areas of high seabird density are also areas of high sensitivity and should be of high conservation concern.

Seasonal variation in total seabird distribution Seabird density in Bantry Bay was much lower in winter (overall 10.3±2.6/km²) than in summer (overall 34.6±5/km²) and exhibited a largely different distribution. In summer seabird distribution was positively related to seaward distance and nearest coast, however in winter densities were negatively related to maximum depth and positively related to nearest coast. This difference in distribution may be largely due to the complete absence of shearwaters in winter and the more offshore location of the Irish Shelf Front (up to 30km further west) and its associated rich feeding grounds at this time of year (McMahon et al., 1995). Densities of many species such as kittiwakes, guillemots and razorbills, which enter coastal waters to breed in summer, are also typically much lower in winter (Pollock et al., 1997; Mackey et al., 2004). Great-northern divers are winter migrants to the bay and show a preference for shallow inshore areas. This species, together with other pursuit divers such as cormorants, shags and black guillemots remain in the shallow inner bay over winter and are limited by depth because of their preference for benthic prey. All of these species, particularly

63 great northern divers (an Annex 1 species) and black guillemots (which are flightless for 5 weeks during the post-breeding moult, Gaston & Jones, 1998) are highly vulnerable to surface pollutants and so are at high risk from any spillages from the Whiddy Island oil terminal.

Offshore species Gannets, fulmars and great skuas occurred in relatively low numbers in Bantry Bay and were restricted mainly to its outer regions. These species are typically more offshore in distribution; the former being among the most abundant species recorded by Pollock et al. (1997) and Mackey et al. (2004). Great shearwaters (Puffinus gravis), European storm petrels and Atlantic puffins (Fratercula arctica) were also found in relatively high densities in offshore waters by Pollock et al. (1997) and Mackey et al. (2004) but were never recorded in Bantry bay. However, there is a record of a wreck of great shearwaters in Bantry Bay (at Whiddy Island) in the past (T.Kelly, pers. comm.) indicating that this species may make occasional use of the bay. European storm-petrels are also regularly seen, sometimes in very large numbers, on sea-watches off Cape Clear Island, Mizen Head and Dursey Island (e.g. Sharrock, 1973; Hutchinson, 1989), indicating that this species is likely to occur in outer regions of the study site but may not have been recorded due to its inconspicuous appearance. Gannets, fulmars and puffins all breed in high numbers within 20km of Bantry Bay (on the Bull and Cow rocks and Dursey Island, chapter 1, Appendix 1.1) but did not regularly forage within the bay. It is likely therefore that the prey of these species is more abundant in offshore waters.

Conclusions The outer region of Bantry Bay is clearly a ‘hot-spot’ of seabird distribution in summer, with large concentrations of shearwaters and Alcidae occurring here. All species within these taxonomic groups are listed as Birds of Conservation Concern in Ireland (BoCCI, Amber List) and are highly vulnerable to surface pollutants (OVI of 22-29), particularly auks which are flightless for 5 weeks during this time. Total seabird density is also high in the outer bay indicating that this area provides rich feeding grounds for a wide range of seabird species.

64 Thus an oil-spill event in the outer bay at this time of year would have major consequences for a wide range of vulnerable seabird species.

In winter however, the inner bay appears to be of much greater importance to seabirds than the outer bay. Two highly vulnerable (OVI, 29) Annex 1 Gaviidae species are present in the bay at this time of year, as well as notable concentrations of black guillemots (OVI, 29), cormorants and shags (OVI, 20 & 24). The incidence of oil-pollution events is higher at this time of year (see Cross et al., 1979; Smiddy, 1992; Smiddy, 1998), and in this region of the bay (due to the presence of the Whiddy Island oil terminal) indicating that these species are at high risk from an oil-spill event. The inner bay is also of high importance to the breeding population of Annex 1 Arctic terns in summer.

The results of the modelling indicate that seaward distance appears to be a reliable predictor of overall seabird density (particularly shearwaters, kittiwakes and Alcidae) in summer (April-September) in Bantry Bay, i.e. high seaward distances indicate high seabird density. This does not apply to all species groups however. Maximum depth can be used as an indicator of the limit of Phalacrocoracidae distribution, with very few individuals occurring in waters deeper than 45m. Maximum depth also appears to be a limiting factor in winter (March-October) for a wide range of wintering species. Further research is needed to assess the transferability of these results to other bays in this region.

65 1.7 REFERENCES

Ainley, D.G., Spear, L.B., Tynan, C.T., Barth, J.A., Pierce, S.D., Ford, R.G. & Cowles, T.J. 2005. Physical and biological variables affecting seabird distributions during the upwelling season of the northern California Current. Deep-Sea Research Part II 1-2, 123-143. Akaike, H. 1973. Information theory as an extension of the maximum likelihood principle. In: B.N. Petrov & F. Csaki (eds), Second international Symposium on Information Theory, Akademiai Kiado, Budapest, Hungary. pp 267-281. Begg, G.S. & Reid, J.B. 1997. Spatial variation in seabird density at a shallow sea tidal mixing front in the Irish Sea. ICES Journal of Marine Science 54, 552-565. Clarke, E.D., Spear, L.B., McCracken, M.L., Marques, F.F.C., Borchers, D.L., Buckland, S.T. & Ainley, D.G. 2003. Validating the use of generalized additive models and at-sea surveys to estimate size and temporal trends of seabird populations. Journal of Applied Ecology 40, 278-292. Cramp, S. & Simmons, K.E.L. 1977. Handbook of the Birds of Europe, The Middle East and North Africa. The Birds of the Western Palearctic. Oxford University Press. pp 722. Cramp, S. & Simmons, K.E.L. 1983. Handbook of the Birds of Europe, The Middle East and North Africa. The Birds of the Western Palearctic., Volume 3, Waders to Gulls. Oxford University Press. pp 913. Cross, T., Southgate, T. & Myers, A.A. 1979. The initial pollution of shores in Bantry Bay, Ireland, by oil from the Tanker Betelgeuse. Marine Pollution Bulletin 10, 104-107. Durazo, R., Harrison, N.M. & Hill, A.E. 1998. Seabird observations at a tidal mixing Front in the Irish Sea. Estuarine Coastal and Shelf Science 47, 153-164. Edwards, A., Jones, K., Graham, J.M., Griffiths, C.R., MacDougall, N., Patching, J., Richard, J.M. & Raine, R. 1996. Transient Coastal

66 upwelling and water circulation in Bantry Bay, a Ria on the Southwest Coast of Ireland. Estuarine, Coastal and Shelf Science 42, 213-230. Elphick, C.S. & Hunt, J., G.L. 1993. Variations in the Distribution of Marine Birds with water mass in the Northern Bering Sea. The Condor 95, 33-44. Garthe, S. 1997. Influence of Hydrography, fishing activity, and colony location on summer seabird distribution on the south-eastern North Sea. ICES Journal of Marine Science 54, 566-577. Gaston, A.J. & Jones, I.L. 1998. The Auks, Bird Families of the World. Oxford University Press, New York. pp 349. Haney, J.C. & McGillivary, P.A. 1985. Aggregations of Cory's Shearwaters (Calonectris diomedea) at Gulf Stream Fronts. Wilson Bulletin 97, 191- 200. Hastie, T.J. & Tibshirani, R.J. 1990. Generalized Additive Models. Chapman & Hall, London Hoefer, C.J. 2000. Marine bird attraction to thermal fronts in the California Current System. The Condor 102, 423-427. Hunt, G.L.J. & Schneider, D.C. 1987. Scale dependent processes in the physical and biological environment of marine birds. In: J.P. Croxall (ed). Seabirds: feeding ecology and role in marine ecosystems. Cambridge University Press, Cambridge, England Hutchinson, C.D. 1989. Birds in Ireland. T & A D Poyser, Calton. pp 215. Hyrenbach, K.D., Forney, K.A. & Dayton, P.K. 2000. Marine protection areas and ocean basin management. Aquatic Conservation: Marine and Freshwater Ecosystems 10, 437-458. Mackey, M., O'Cadhla, O., Kelly, T.C., Aguiler de Soto, N. & Connolly, N. 2004. Cetaceans and Seabirds of Ireland's Atlantic Margin. Volume 1 - Seabird distribution, density and abundance. Report on research carried out under the Irish Infrastructure Programme (PIP): Rockall Studies Group (RSG) projects 98/6 and 00/13, Porcupine Studies Group project P00/15 and Offshore Support Group (OSG) project 99/38, Cork. pp 95. McCullagh, P. & Nelder, J.A. 1989. Generalized Linear Models. Chapman & Hall, London

67 McMahon, T., Raine, R. & Boychuk, S. 1995. Some oceanographic features of northeastern Atlantic waters west of Ireland. ICES Journal of Marine Science 52, 221-232. Mitchell, I.M., Newton, S.F., Ratcliffe, N. & Dunn, T.E. 2004. Seabird Populations of Britain and Ireland. Results of the Seabird 2000 Census (1998-2002). T & AD Poyser, London. pp 511. Pollock, C., Reid, J., Webb, A. & Tasker, M. 1997. The distribution of seabirds and cetaceans in the waters around Ireland. JNCC Report, No. 267, Peterborough. pp 167. Pollock, C.M. 1994. Observations on the distribution of seabirds off south-west Ireland. Irish Birds 5, 173-182. Raya Rey, A. & Schiavini, A.C.M. 2000. Distribution, abundance and associations of seabirds in the Beagle Channel, Tierra del Fuego, Argentina. Polar Biology 23, 338-345. Ribic, C.A., Ainley, D.G. & Spear, L.B. 1997. Seabird associations in Pacific equatorial waters. Ibis 139, 482-487. Roycroft, D. 2005. Seabirds at sea in high risk inshore environments. Ph.D Thesis, Department of Zoology, Ecology and Plant Science, University College Cork, Cork. pp 220. Sharrock, J.T.R., (ed). 1973. The Natural History of Cape Clear Island. Poyser, Berkhamsted. Shealer, D.A. 2002. Foraging behaviour and food of seabirds. In: E.A. Schreiber & J. Burger (eds), Biology of Marine Birds. CRC Press, London. pp 137- 177. Skov, H. & Durinck, J. 2000. Seabird distribution in relation to hydrography in Skagerrak. Continental Shelf Research 20, 169-187. Skov, H., Durinck, J. & Andell, P. 2000. Associations between wintering avian predators and schooling fish in the Skagerrak-Kattegat suggest reliance on predictable aggregations of herring Clupea harengus. Journal of Avian Biology 31, 135-143. Smiddy, P. 1992. The effect of the Kowloon Bridge oil spill in east Cork. Irish Birds 4, 559-570. Smiddy, P. 1998. The effect of the Cork Harbour Oil Spill of November 1997 on birds. Irish Naturalists Journal 26, 32-37.

68 Spear, L.B., Balance, L.T. & Ainley, D.G. 2001. Response of seabirds to thermal boundaries in the tropical Pacific: the thermocline versus the Equatorial Front. Marine Ecological Progress Series 219, 275-289. Tasker, M.L., Hope Jones, P., Dixon, T. & Blake, B.F. 1984. Counting seabirds at sea from ships: A review of the methods employed and a suggestion for a standardised approach. The Auk 101. Veit, R.R., Silverman, E.D. & Everson, I. 1993. Aggregation patterns of pelagic predators and their principal prey, Antarctic Krill, near South Georgia. Journal of Animal Ecology 62, 551-564. Wanless, S., Burger, A.E. & Harris, M.P. 1991. Diving depths of shags Phalacrocorax aristotelis breeding on the Isle of May. Ibis 133, 37-42. Webb, A. & Durinck, J. 1992. Counting Birds from ships., Manual for Aeroplane and ship surveys of waterfowl and seabirds. IWRB special publication, Slimbridge. pp 24-37. Wright, P.J. & Begg, G.S. 1997. A spatial comparison of common guillemots and sandeels in Scottish waters. ICES Journal of Marine Science 54, 578- 592. Yen, P.P.W., Huettmann, F. & Cooke, F. 2004. A large-scale model for the at- sea distribution and abundance of Marbled Murrelets (Brachyramphus marmoratus) during the breeding season in coastal British Columbia, Canada. Ecological Modelling 171, 395-413.

69 1.8 APPENDIX

Appendix 1.1. Breeding seabird numbers in Bantry Bay, including the outer bay from Bere Island to Dursey Island. Seabird 2000 results (Newton pers. comm.). Number of Pairs Inner Sheep’s Bere Is - Total Bantry Head Dursey Bantry Northern Fulmar 11 575 586 Great Cormorant 52 52 European Shag 82 12 94 Lesser black-backed gull 4 4 Herring gull 19 21 40 Great black-backed gull 8 1 9 Arctic Tern 104 104 Razorbill* 7 7 Black Guillemot* 71 65 79 215 * Number of individuals

70

CHAPTER 2

SHORE-BASED OBSERVATIONS OF SEABIRDS IN

SOUTHWEST IRELAND.

2.1 ABSTRACT

The community composition and relative abundance of seabirds at sea in the high-risk inshore environment of southwest Ireland was studied from six shore- based observation points over a three-year period. Shearwaters (mainly manx shearwaters, Puffinus puffinus) and gannets (Morus basanus) dominated the species assemblage at the outer headland sites, while auks (mainly guillemot, Uria aalge and razorbill, Alca torda) dominated the species assemblage at the more inshore sites of Bantry Bay. The diversity and species richness of seabirds was high in Bantry Bay due to the presence of both neritic and pelagic species, however the total relative abundance of seabirds at the outer headland sites was over double that of the Bantry Bay sites. Peak numbers of many species occurred in autumn (August - October) indicating that an oil-spill event at this time of year would have a large impact on seabirds in these sites. The tidal cycle did not significantly influence the abundance of any of the species studied, however further studies incorporating seabird behaviour are recommended. The use of a theodolite at selected sites allowed the distance to all seabird sightings (excluding flying birds) to be calculated and the results showed a significant decline in detection rate of seabirds at distances above 2km from the observation point. Densities calculated within this 2km radius did not differ significantly from densities calculated at these sites using boat-based surveys. It was concluded that estimates of seabird density using fixed-point survey techniques are reliable up to a distance of 2km from the observation point, but are likely to be underestimated at larger scales (>2km).

72 2.2 INTRODUCTION

The inshore waters of southwest Ireland are likely to be utilized by large numbers of foraging seabirds from some of the largest breeding colonies in the country (i.e. the Blasket Islands and the Skellig rocks), as well as a number of seasonal and passage migrants, many of which are protected under national or EU legislation (e.g. storm petrel, Hyrdobates pelagicus, great-northern diver, Gavia immer). These inshore waters represent areas of high potential risk to seabirds due to the presence of anthropogenic activities such as mariculture and the shipping and storage of oil (e.g. Bantry Bay). The density, distribution and fine scale habitat use of seabirds in Bantry Bay have been described in detail (chapter 1), however the seasonal use of the Bay and its surrounding waters by individual species groups has not been studied. The influence of the tidal cycle on seabird abundance at these sites is also relatively unknown and may be a significant factor for diving species.

An understanding of the seasonal and tidal variations in use of these areas by vulnerable seabird species is vital for the effective management and impact assessment of major pollution events in high-risk areas. Significant seasonal and/or tidal variations in abundance identified in these areas, could also be applied to similar regions utilized by these seabirds and so are of broad-ranging use.

Land-based observations are a widely used, inexpensive and easily executed method of surveying seabirds at sea. The accuracy of bird density estimation from fixed-point surveys has been discussed in detail by a number of authors (e.g. Bibby et al., 1992; Buckland et al., 2001), however field-based ground- truthing of these density estimates at fine scales has not been attempted to date for seabirds at sea. This study provides a measure of the accuracy of shore-based density estimates of seabirds by a comparison with boat-based density estimates in Bantry Bay.

73 Aims The aims of this study were to: I. examine the abundance, diversity and community composition of seabirds at selected sites in southwest Ireland – identifying hot-spots of abundance and concentrations of species vulnerable to oil spills; II. Investigate seasonal and tidal variations in seabird relative abundance at the six shore-watch sites; and III. Determine the reliability of density estimation from shore-based observations through a comparison with boat-based survey techniques.

2.3 METHODS

2.3.1 Shore-watch techniques Six observation points overlooking Bantry Bay and its approaches were selected in order to represent both inner and outer sections of the study site (see Figure 2, General Introduction for site map). All observation points were located within 200m of the shoreline, had a relatively un-obstructed view of the study site, and were at least 45m above sea level (all apart from the Inner Bantry site were over 80m above sea level). Observations were carried out systematically each month, using a telescope (Swarovski, 20-60x zoom lens) and binoculars (Leica 10 x 42). A theodolite (Sokkia DT500, 30x magnification) was also used at the Sheep’s Head and Inner Bantry sites in order to accurately record positions of birds on the water (for ground-truthing of boat-survey data). This surveying instrument measures a horizontal and a vertical angle in relation to a set reference point (e.g. lighthouse) on each bearing. Using a precisely known observer point, reference point and known observer height these angles can subsequently be used to derive positions of remote objects at sea (Würsig et al., 1991; Lutkebohle, 1995). Ideally the variation in sea level height caused by tidal fluctuations should be accounted for, however for this study all seabird positions were grouped into a 1km grid-system and so fine-scale positioning of seabirds was not essential.

Scans were carried out during sea states of 3-4 or less on the Beaufort Scale from each vantage point, over all tidal states and between the hours of 0900 and

74 1930 (1600 in winter). Typically, one watch (consisting of one or more scans) per month was carried out from each vantage point in the winter (October to March) and at least two per month (weather permitting) in the summer (April to September). Scan duration varied between 30 and 350 minutes depending on the number of seabirds present. The area of sea within a 0-180ºobservation arc (or less depending on the view available) was surveyed as far as the eye could see during each scan. The near-shore area was surveyed first using binoculars, followed by systematic sweeps (e.g. 0º -180º arcs) with the telescope at increasing distance increments from the shore until the horizon was reached. A horizontal and vertical angle was obtained for each seabird sighting using the theodolite, along with the relevant sighting details such as species, group size and behaviour. Theodolite positions can only be obtained for birds on the water, thus flying birds were not included in this analysis. Birds associated with fishing vessels were also excluded from all calculations. Shore-watches were carried out between June 2001 and February 2004 at the Sheep’s Head and inner Bantry sites and between June 2003 and September 2004 at the four outer sites.

2.3.2 Analysis of relative abundance Relative abundance was calculated as the number of birds on the water (for each species group) in each scan divided by the number of scans at that site. The resulting value is the mean number of birds per scan at each site and this was used for analysis of community similarity and diversity. Flying birds and birds associated with fishing vessels were excluded from all analyses.

Community composition and similarity Diversity at the four shore-watch sites was measured using Simpson’s Index of 1 Diversity (Reciprocal /D) and community similarity was investigated using the Bray-Curtis Index of Similarity (Magurran, 1988). Simpson’s Index of Diversity (D) is calculated as:

D = ∑ ni (ni –1) N (N –1)

Where: ni = total number of individuals of a particular species

75 N = total number of individuals of all species 1 Simpson’s Reciprocal Index ( /D) was presented, as it is more intuitive (i.e. the higher the value the greater the diversity).

The Bray-Curtis index incorporates the abundance of species as well as species richness at the different sites. This index reflects the similarity in individuals between habitats and is calculated as:

Cn = 2jn . (An + Bn)

Where: j = number of species in common jn = sum of the lesser values of species in common to both An = total individuals in habitat A Bn = total individuals in habitat B

Similarity between communities in different sites was expressed as a dendrogram using Mountfords classification scheme based on simple average linkage. The ‘BioDiversity Professional’ program (version 2) was used for the calculation of diversity and similarity indices.

Seasonal variation Seasonal variations in seabird abundance (excluding flying birds which did not make contact with the water surface) were displayed graphically at the six shore- watch sites. The mean number of birds per scan in each season was computed for this. The seasons were allocated as follows: Spring = February, March and April; Summer = May, June and July, Autumn = August, September and October and Winter = November, December and January. The same species groups defined in chapter 1 were used here.

Tidal variations Three paired measures of tidal variation were categorized to investigate the influence of the tidal cycle on the abundance of a number of seabird families;

a) Tide height; ‘high’ (3 hours before and 3 hours after high tide), versus ‘low’ (3 hours before and 3 hours after low tide), b) Direction of flow; ‘ebbing’ (6 hours) versus ‘flooding’ (6 hours),

76 c) Rate of flow; ‘slack’ (3 hours before and 3 hours after the turn of the tide) versus ‘flow’ (the remaining 6 hours of faster tidal flow).

Tidal variations in abundance were investigated at the inner Bantry and Sheep’s Head shore-watch sites separately and data from the four outer headland sites (Mizen, Black Ball, Three Castle Heads and Dursey Island) were pooled to improve sample size. The four most numerically abundant species groups at each site were chosen for tidal analysis. This comprised the Alcidae, (guillemot Uria aalge, razorbill Alca torda and black guillemot Cepphus grylle), Laridae (all gull species, not including kittiwake), Phalacrocoracidae (cormorant Phalacrocorax carbo and shag Phalacrocorax aristotelis) and Manx shearwaters (Puffinus puffinus) at the inner Bantry site and Manx shearwaters, gannets (Morus basanus), kittiwakes (Rissa tridactyla) and Alcidae at the outer sites. The mean number of birds per month in each of the six tidal states was calculated for the three shore-watch site groups. These abundances were compared between tidal states using paired t-tests (all data were normally distributed).

2.3.3 Density calculation Using spherical trigonometry, theodolite readings recorded at the inner Bantry and Sheep’s Head shore-watch sites were used to derive geographical positions of seabirds on the water. All sightings of birds on the water were allocated to a 1km² grid square using the grid system devised for boat survey data in chapter 1. This was achieved by linking the grid-reference fields of the sighting data and the grid square data in a Microsoft Access database program. The number of birds on the water in each grid square was divided by the effort (number of scans at that site) and by the area surveyed (1 in this case) to produce a density of birds per kilometre squared. In this instance, densities were calculated for two periods of the year; Summer (April to September) and Winter (October to March) for comparison with the boat survey data. Densities were only calculated for total seabirds pooled (excluded gulls), and not for separate species groups as the aim of this exercise was to compare shore and boat-based survey results. The densities derived from boat-based surveys of separate species groups in Bantry Bay were described previously in chapter 1. All densities were mapped using ArcView (version 8.1).

77

Figure 1 shows that the detection rate of seabirds (all species pooled) was not constant across all grid squares. Mean seabird density was negatively related to distance from the observation point (GLM, P<0.001) when all sites and seasons were pooled. However there was no significant relationship between density and distance from the observation point when only grid squares with midpoints located less than 2km from the observation point were included (GLM, Quasi- Poisson family, log-link function, t-value 4.712, P= 0.371, n=32). For this reason, densities calculated for grid squares over 2km from the observation point were considered as unreliable, and were omitted from further analysis, as they were likely to be underestimated.

The overall density of seabirds in the surveyed shore-watch area (up to a specified distance from the observation point) can be estimated using the DISTANCE program (Buckland et al., 2001). This takes into account the decline in detection rate of seabirds with increasing distance from a fixed point. The area of the visible shore-watch site was calculated by estimating the angle of the viewing arc (i.e. 0-140o ) and dividing it by 360º as the program assumes a circular viewing area. The overall density of seabirds recorded from the Sheep’s Head shore-watch site in summer was estimated using this program1. Overall density values produced for large areas such as the outer Bantry Bay site using this technique should be used with caution however as they are based on the assumption that birds were distributed evenly throughout the outer bay (i.e. that the density of seabirds close to the coast was representative of densities further offshore). For the purposes of this study, densities in a 1km grid-system were considered to be more appropriate as these showed fine-scale variations in density and were directly comparable with boat-survey results. Thus only one shore-watch site was analysed using the DISTANCE program for comparison purposes.

1 A Uniform Key Function with a simple polynomial series expansion was selected as the best model using the AIC statistic. Simple polynomial adjustments were of the order 2, 4 and 6.

78 ) 10

m2 9 /k

s 8

d 7 ir

B 6 (

y 5 it

s 4 n

e 3

D 2 n

a 1 e 0 M 0 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 10 20 30 40 50 60 70 80 90 10 11 12 6 26 49 72 86 100 90 92 74 66 60 44 24 Distance from observation point (m)

Figure 1. Mean (±SE) seabird density (excluding gulls) at all shore-watch sites (all seasons) with increasing distance from the observation point. The number of grid squares (n) per distance category is also shown on the x-axis. Distances were measured from the mid-point of each 1km2 grid square.

2.3.4 Comparison of shore and boat-based densities In order to compare seabird densities calculated from shore-based and boat-based surveys, all grid squares surveyed using the two techniques were identified. Only grid squares with midpoints located less than 2km from the observation point were included in the analysis as seabird detection rates from shore-watch sites were accurate here. Densities in winter and summer at the two Bantry Bay sites were compared using a paired t-test following normalization using log10 (+1) transformation. The SPSS statistical program (Version 11) was used for this data analysis.

79 2.4 RESULTS

A total of 44 shore-watches (37 with theodolite readings) were carried out from the Sheep’s Head observation point and 51 (46 with theodolite) were carried out from the inner Bantry Bay site. A further 22 surveys were carried out at Three Castle Head, 14 from Mizen Head, 13 from Black Ball Head and 9 from Dursey Island. These totals are exclusive of surveys that were carried out during deteriorating weather or poor visibility – these were deleted from the dataset.

2.4.1 Relative abundance Table 1 shows the relative abundance of the five most commonly occurring seabird groups at the six shore-watch sites. Mean seabird numbers per scan were highest at the Black Ball Head site, with a mean of over 776 birds per scan over the entire study period, followed by Three Castle Head (671 birds per scan) and Mizen Head (651 birds per scan). Shearwaters and gannets comprised the bulk of all sightings at these three sites. The inner Bantry Bay site recorded the lowest numbers of seabirds with a mean of 149 birds per scan with Alcidae being the most abundant species. The Dursey Island and Sheep’s Head sites held similar seabird numbers (Mean ~320 per scan), however the species assemblage differed markedly. The highest number of Alcidae at any site occurred at Sheep’s Head, while the lowest occurred at Dursey Island. The lowest number of shearwaters was also recorded at Dursey Island – despite the high abundance of this species at the nearby Black Ball Head. This may reflect the small sample size of surveys at this site however.

Table 1. Relative abundance (Mean number of birds per scan ± Standard Error) of selected seabird groups at the six shore-watch sites. The number of scans per site (N) is also shown). Species Inner Sheep’s Dursey Black Ball Mizen Three Bantry Head Island Head Head Castle Head N = 51 N = 44 N = 9 N = 13 N = 14 N = 22 Shearwaters 23.8± 8.3 110.2±29.2 2.2± 2.2 400.4±209.2 394.6±251.5 247.2±107.8 Gannets 9.5± 2.4 30.1± 7.3 194.4±64.9 192.1± 66.2 113.6± 43.6 269.8± 72.0 Laridae 51.5±21.6 37.6±21.6 3.2± 1.2 61.6± 30.6 32.3± 11.8 8.0± 2.7 Kittiwakes 0.6± 0.5 19.5± 8.3 79.2±59.6 61.9± 42.6 43.1± 18.0 46.9± 15.6 Alcidae 64.1±19.7 124.1±33.1 40.8±25.5 60.8± 20.6 67.4± 42.8 99.6± 44.7

80

Figure 2(a-f) shows the community composition of seabirds at the six shore- watch sites (not including flying birds). There are clear differences in community composition across the sites with Alcidae forming a much larger proportion of the species assemblage at the inner Bantry and Sheep’s Head sites (16-34%) than any of the outer sites (2-8%). Conversely, Gannets formed a large percentage of the overall assemblage (17-60%) at the outer sites but only accounted for 6 to 9% of the species at the inner Bantry and Sheep’s Head sites. Laridae were also much more abundant at the inner sites (12-31%) than at the outer headlands (1-8%). Shearwaters formed a large proportion of the species assemblage at all sites apart from Dursey Island (Figure 2c) where they accounted for only 1% of the total assemblage.

Species richness and diversity Table 2 shows that species richness was highest at the Three Castle Head site (18 species) followed by the inner Bantry and Sheep’s Head sites (16 species) and the Mizen Head site (15 species). The sites with the lowest species richness (Dursey Island and Black Ball Head) were also the sites with the lowest survey effort indicating that this may be an incomplete representation of the species utilizing these sites. Diversity was highest at the Sheep’s Head site (Simpson’s 1 1 1 Index /D = 3.5), followed by Inner Bantry ( /D = 3.3), Three Castle Head ( /D = 1 2.9) and the Black Ball Head ( /D = 2.7). Bray-Curtis cluster analysis showed that the Mizen and Black Ball Head communities were very similar to each other (87% similarity) and also to the Three Castle Head community (74%). The Sheep’s Head and Inner Bantry communities were similar (59% similarity) to each other but distinct from the four outer headland sites (28% similarity). The Dursey Island community was quite distinct from the other sites with only 54% similarity to any other community but low survey effort here may have contributed to this..(Figure 3).

81 Other Other 1% Shearwaters 1% 16% Shearwaters Alcidae 34% 38% Gannets Alcidae 6% 44% Phalacro. 2% Gannets 9% Laridae Laridae Kittiwakes 12% 31% 6%

a) Sheep’s Head b) Inner Bantry

Other 5% Alcidae Alcidae Other Laridae Shearwaters 8% Laridae 2% 1% 8% 1% 1% Kittiwakes Kittiwakes 8% 25% Shearwaters 55% Gannets Gannets 26% 60%

c) Dursey Island d) Black Ball Head

Alcidae Other Alcidae Other 6% Laridae 4% 8% Laridae 3% 1% Kittiwakes 5% Kittiwakes 7% 7% Shearwaters 39% Gannets 17% Shearwaters 59% Gannets 44% e) Mizen Head f) Three Castle Head

Figure 2 (a-f). Community composition (% of total bird numbers) at the Sheep’s Head (a), Inner Bantry (b), Dursey Island (c), Black Ball Head (d), Mizen Head (e) and Three Castle Head (f) shore-watch sites.

82 Table 2. Species presence/absence at the six shore-watch sites during the entire study period. The number of surveys (N) per site is also shown. SPECIES Sheeps Inner Dursey Black Mizen Three Head Bantry Island Ball Head Castle Head Head N = 44 N = 51 N = 9 N= 13 N = 14 N = 22 Red-throated Diver (Gavia stellata) • Great Northern Diver (Gavia immer) • • Northern Fulmar (Fulmarus glacialis) • • • • • • Manx Shearwater (Puffinus puffinus) • • • • • • Sooty Shearwater (Puffinus griseus) • European Storm Petrel (Hydrobates pelagicus) • Northern Gannet (Morus basanus) • • • • • • Great Cormorant (Phalacrocorax carbo) • • • • • • European Shag (Phalacrocorax aristotelis) • • • • • • Great Skua (Stercorarius skua) • • Arctic Skua (Stercorariu parasiticus) • • • Black-headed Gull (Larus ridibundus) • • • • Mew Gull (Larus canus) • Herring Gull (Larus argentatus) • • • • • • Lesser Black-backed Gull (Larus fuscus) • • • • • Great Black-backed Gull (Larus marinus) • • • • • • Black-legged Kittiwake (Rissa tridactyla) • • • • • • Common Guillemot (Uria aalge) • • • • • • Razorbill (Alca torda) • • • • • • Black Guillemot (Cepphus grylle) • • • • • Atlantic Puffin (Fratercula arctica) • • • • • • = present.

HD

Figure 3. Dendrogram produced from the Bray-Curtis Index of community similarity. The scale shows percentage similarity (links drawn close to 100% are highly similar).

83 Seasonal variation in relative abundance Seasonal variations in seabird abundance at the six shore-watch sites are shown in Figure 4(a-l). Mean seabird numbers were highly variable at all sites, with the flocking behaviour of many species contributing to the high standard errors recorded. Some consistent seasonal trends were present in the four most commonly occurring species however (Figure 4). The most obvious seasonal variation in abundance occur in the shearwaters as both manx and sooty shearwaters are summer migrants – being almost completely absent from all sites in winter and spring (Figure 4a-c). Mean numbers of this species group peaked in spring at the outer headlands (apart from Dursey Island where few were recorded) and peaked in autumn at the inner Bantry and Sheep’s Head sites. Highest mean numbers of gannets occurred in either summer or autumn at all sites (Figure 4 d-e) apart from Sheep’s Head where a large feeding flock was recorded in winter. Autumn appears to be the most important time for kittiwakes and Alcidae at most sites, with clear peaks in abundance occurring at this time (Figure 4g-l). Mean Alcidae numbers peaked in winter at the Mizen Head and Three Castle Head sites however with numbers increasing markedly at both sites at this time of year (Figure 4(l)). This indicates that these sites may be important winter foraging areas for these species. Winter migrants such as great-northern and red-throated divers were exclusively recorded in winter and spring.

Tidal variation in relative abundance There was no significant difference between mean Phalacrocoracidae, Alcidae, Laridae or shearwater abundance across any of the tidal categories at the inner Bantry site (paired t-test, P>0.05). Similarly there was no significant difference between mean shearwater, gannet, kittiwake or Alcidae abundance at the Sheep’s Head or the pooled outer headland sites (paired t-test, P>0.05). Some trends were apparent however with gannets occurring in higher mean numbers during the flooding tidal category than during tidal ebbing at Sheep’s head and the outer sites. This difference was not significant however.

84 300 140 an n c a

c 120

r s 250 s e r

p 100 s

200 pe r s e t

e 80 at

w 150 nn r

a 60 ea G h 100

S 40 . no. o n n 50

a 20 n a Me e 0 0 M Spring Summer Autumn Winter Spring Summer Autumn Winter

a) Inner Bantry Sheeps Head d) Inner Bantry Sheeps Head

n 2000 500 n

sca 1800 450 a r c e

p 1600 400 s

r s 1400 350 pe er

s at

1200 t 300 w r

a 1000 250 nne e a h 800 200 G S .

x 600 150 n no

a

400 n 100 M a

. e

o 200 50 M n 0 0

ean Spring Summer Autumn Winter Spring Summer Autumn Winter M b) Dursey Is. Black Ball Hd e) Dursey Is. Black Ball Hd

n 3000 600 n sca a r c e 2500 500 s p

r s er

2000 pe 400 s at t w r

a 1500 300 nne e a h G

S 1000 . 200 x n no a n M a

500 100 . e o M n 0 0

ean Spring Summer Autumn Winter Spring Summer Autumn Winter M c) Mizen Hd Three Castle Hd f) Mizen Hd Three Castle Hd

Figure 4 (a-f). Seasonal abundance (Mean ± SE) of Manx shearwaters (a-c) and Gannets (d-e) at the Sheep’s Head, Inner Bantry, Dursey Island, Black ball Head, Mizen Head and Three Castle Head shore-watch sites.

85 60 n 350 a c

s 50

r 300 scan r pe

40 e s 250 e k a 30 ae p 200 w d i t ci t l i 150 A K

20 .

o 100 n

no. 10 n an

a 50 e M Me 0 0 Spring Summer Autumn Winter Spring Summer Autumn Winter

g) Inner Bantry Sheeps Head j) Inner Bantry Sheeps Head

300 160 an

an 140 250 sc r sc

r 120 e 200 e e p 100 e p a ak d w 150 i 80 i c t l t i

A 60 . K

. 100 o o n n 40 n 50 an a e 20 e M M 0 0 Spring Summer Autumn Winter Spring Summer Autumn Winter

h) Dursey Is. Black Ball Hd k) Dursey Is. Black Ball Hd

140 800 n a c 120 an 700 s

sc r e

r 600 p

100 e

p

es 500 e

80 a ak d

w 400 ci i l tt 60 i

A 300 . K . 40 o o

n 200 n n a

20 e an 100 e M M 0 0 Spring Summer Autumn Winter Spring Summer Autumn Winter

i) Mizen Hd Three Castle Hd l) Mizen Hd Three Castle Hd

Figure 4 (g-l). Seasonal abundance (Mean ± SE) of Kittiwakes (g-i) and Alcidae (j-l) at the Sheep’s Head, Inner Bantry, Dursey Island, Black ball Head, Mizen Head and Three Castle Head shore-watch sites.

86 2.4.2 Density Table 3 shows the mean seabird densities (excluding gulls) calculated at the two Bantry Bay shore-watch sites in winter and summer using shore and boat-based survey techniques. The densities calculated from the Sheep’s Head shore watch site in winter (7.8±2.8/km²) were significantly higher than those calculated for the same grid squares from boat surveys (0.6±0.6/km²) in winter (paired t-test, df=4, t=3.597, P<0.05). There was no significant difference between densities calculated from the two techniques at the inner Bantry site however.

Table 3. Mean seabird density/km² (excluding gulls) in grid squares surveyed using both shore and boat-based survey techniques in summer and winter at the Bantry shore-watch sites. Only grid squares located less than 2km from the land- watch point and surveyed using both techniques were included.

Site Survey Sheeps Head Inner Bantry Summer Shore 17.2±5.8 (5) 5.7±1.5 (2) Boat 23.1±9.8 (5) 4.0±1.0 (2) Winter Shore 7.8±2.8 (5) 6.2±0.4 (2) Boat 0.6±0.6 (5) 7.5±7.5 (2)

Using the DISTANCE program, total seabird density (excluding gulls) at the Sheep’s Head shore-watch site (up to a distance of 8km) was calculated as 6.7/km² (95% CI 4.9-9.2) in summer. This compares to the mean density of 5.6±1/km² (n=77 grid squares) calculated from the same shore-watch data without accounting for the decline in detection rate of seabirds with distance. Both of these densities are lower than those calculated from the near-shore grid squares (n=5) at this site from shore (17.2±5.8/km²) and boat (23.1±9.8km²) - based surveys (Table 3).

The distribution of total seabird density (excluding gulls) at the two Bantry sites in summer and winter up to a distance of 7km from the observation point can be seen in Figure 5(a-d). Total seabird densities were higher at the Sheep’s Head site than at the Inner Bantry site in both seasons, however the difference was less pronounced in winter (see also Table 3). The decrease in detectability of seabirds (see section 2.3.3) can be clearly seen from these maps as highest densities occur close to the observation points at both sites.

87

a) Sheep’s Head Summer b) Sheep’s Head Winter

c) Inner Bantry Summer d) Inner Bantry Winter

Figure 5(a-d). The distribution of mean seabird density (excluding gulls) at the outer Bantry (Sheep’s Head) shore-watch site in summer (a) and winter (b) and at the inner Bantry site in summer (c) and winter (d). Only grid-squares up to 7km from the shore-watch point are shown as detection rates were very low beyond this.

88 2.5 DISCUSSION

Species richness, diversity and relative abundance The high diversity recorded at the Sheep’s head and inner Bantry sites reflects the evenness of abundance of the species assemblages here. These communities were not dominated by any one species, such as at some of the outer sites (e.g. 60% gannets at Dursey Island). Species richness was highest at the Three Castle Head site where European storm petrels (Hydrobates pelagicus) and two skua species (great & Arctic skua) were recorded as well as a variety of more neritic species such as cormorants and shags. Species richness was also high at the Bantry Bay sites due to the mix of neritic and pelagic species utilizing these sites. The high diversity and species richness at these sites indicates that this region is of high importance to a wide range of seabirds.

Although most had a lower species richness and diversity than the Bantry Bay sites, the outer headlands held a higher relative abundance of many species groups. The three outer sites of Three Castle, Mizen and Black-Ball Head held over double the relative abundance of seabirds than the Sheep’s Head site and four times that of the inner Bantry site. This may indicate that foraging conditions are optimal here – or may simply reflect the proximity of these sites to large breeding colonies at the Bull and Cow rocks and Scarriff and Deenish Island. It is likely that the species richness and abundance of seabirds at the Dursey Island site was underestimated due to low survey effort here - indicating that long term monitoring is essential for the accurate assessment of seabird site utilization.

Community composition and similarity The differing seabird communities (<30% similarity) utilizing the Bantry Bay sites compared to the outer headlands may reflect the inshore location of the Bantry sites. The Bantry Bay sites, (particularly the inner site) clearly represent an attractive foraging location for neritic species such as cormorants, shags, divers and gulls but are also utilized by large proportions of Alcidae – setting them apart from the outer sites which were dominated by shearwaters and gannets. The seabird communities at the outer headland sites were remarkably similar (>70% similarity), apart from Dursey Island site which (unlike the other sites) was not utilized by large numbers of

89 shearwaters. It is likely that the low survey effort here resulted in an under- representation of the species assemblage however. The exposed and open nature of the outer headlands as well as their proximity to breeding colonies and the rich foraging grounds of the Irish Shelf Front (Raine et al., 1990; Edwards et al., 1996) makes the outer sites attractive to pelagic species such as shearwaters and gannets.

Vulnerability to surface pollutants The outer Bantry shore-watch site contained the highest proportion (72%) of species vulnerable to surface pollution (i.e. high Oil Vulnerability Index, OVI >20, see General Introduction), followed by the inner Bantry site (62%). The species contributing to this high vulnerability comprised the Alcidae group and shearwaters, all of which are Birds of Conservation Concern in Ireland (BoCCI, amber list) (Table 1, General Introduction). In terms of relative abundance however the outer headland sites should be of conservation priority as these areas were utilized by very high numbers of protected (BoCCI amber list) shearwaters and gannets and well as the Annex 1 European storm petrel.

Seasonal variations The relative abundance of all seabird species was highly variable from season to season and from site to site during this study, reflecting the patchy distribution of these species. Autumn appeared to be the most important season for many species, with high numbers of auks, shearwaters, gannets and kittiwakes occurring at this time. All of these species breed on the islands off southwest Ireland and leave the colonies in large numbers in late summer to forage before dispersing in late autumn (Webb et al., 1995). Auks undergo a total flight-feather moult at this time of year, becoming flightless for approximately six weeks (Gaston & Jones, 1998) and therefore are highly vulnerable to surface pollution events at this time of year.

The high variability in gannet and Alcidae abundance in winter at some sites (e.g. Mizen, Three Castle and Sheep’s Heads) reflects the presence of transient feeding flocks at this time of year. The rich feeding grounds associated with the Irish Shelf Front are located further offshore (up to 30km further west) at this time of year (McMahon et al., 1995). As a consequence, seabird prey in inshore areas at this time of year is likely to be patchier and less reliable than in summer and autumn. For

90 example, herring Clupea harengus are known to spawn in the vicinity of Bantry Bay in October and November (Smith & McLaverty, 1997; Boelens et al., 1999) and may be exploited by large numbers of feeding seabirds during this short time. The Gaviidae group (consisting of the great northern and red-throated divers) occurred exclusively in winter and spring in the inner bay. The frequency of oil pollution events in southwest Ireland is also highest in winter due to the high winds and the exposed nature of the coastline (see Cross et al., 1979; Smiddy, 1992; Smiddy, 1998) indicating that these Annex 1 species are at high risk from future pollution incidents.

The high variability in seabird abundance at sea makes predicting the impacts of a major pollution event difficult, particularly at small scales such as within bays and estuaries. The largest and most vulnerable (high OVI) concentrations of seabirds in these study sites occur in autumn (e.g. moulting auks and feeding shearwaters) however the species of highest conservation concern (i.e. Annex 1 Gaviidae species) occur in winter when the likelihood of an oil spill event is highest. Clearly there are vulnerable (and often unpredictable) concentrations of seabirds in this region throughout the year and longer-term monitoring is needed before reliable predictions of seasonal abundance can be provided to inform management decisions.

Tidal variations The tidal cycle did not significantly influence the abundance of any of the species studied (i.e. Phalacrocoracidae, Alcidae, Laridae, gannets, kittiwakes and shearwaters) at any of the shore-watch sites. The trend towards higher mean abundances of gannets during flooding tides may indicate that these piscivorous birds are following fish brought into these inshore sites on the incoming tide. Further investigations incorporating seabird behaviour (i.e. foraging birds only) may produce clearer patterns. Nevertheless it is clear that seabird abundance did not vary significantly with the tidal cycle at these sites, indicating that this variable cannot be used as a reliable predictor of seabird abundance when responding to a disturbance event in these areas.

The reliability of density estimates from shore-based observation points

This study provided the first known comparison of fine-scale (1km² grid network) seabird density estimates calculated using shore and boat-based survey techniques.

91 The results indicate that only the area of sea within 2km of the observation point can be accurately surveyed using shore-based surveys with the aid of a telescope (max 60x magnification). At distances above this there is a steep decline in detectability of seabirds with increasing distance resulting in an underestimate of seabird density in these areas. Thus, it is difficult to make inferences about seabird distribution from density maps created from shore-watch observations. These results are applicable to all mixed seabird surveys carried out with the aid of a telescope from shore-based observation points in sea states of 3 or less on the Beaufort scale. The area over which seabirds can be accurately surveyed may be greater if small species such as auks and shearwaters are excluded however.

Densities calculated using observations from the first 2km of each shore-watch area did not significantly differ from those calculated from boat-based surveys at the same sites at fine scales (1km grid). However, sample sizes were small and variances high, indicating that these results are not conclusive. The exclusion of all sightings exceeding this distance threshold of 2km means that only a small sample of the seabirds present can be used for density calculations and species known to avoid land when foraging (e.g. shearwaters, see chapter 1, section 1.5.1) may be completely excluded. The DISTANCE program can be used to estimate seabird densities at larger scales, however sightings close to the observation point have a high influence on the density value and may therefore bias results. For this reason shore-watch observations do not provide a useful measure of seabird density and are better suited for studies of relative abundance.

92 2.6 REFERENCES

Bibby, C.J., Burgess, N.D. & Hill, D.A. 1992. Bird Census Techniques. Academic Press Ltd., London. Boelens, R.G.V., Maloney, D.M., Parsons, A.P. & Walsh, A.R. 1999. Irelands Marine and Coastal Areas and Adjacent Seas. An Environmental Assessment. Marine Institute, Dublin, Ireland. pp 388. Buckland, S., Anderson, D.R., Burnham, K., Laake, J., Thomas, L.T. & Borchers, D.L. 2001. Introduction to Distance Sampling: Estimating Abundance of Biological Populations. Oxford University Press, New York. pp 432. Cross, T., Southgate, T. & Myers, A.A. 1979. The initial pollution of shores in Bantry Bay, Ireland, by oil from the Tanker Betelgeuse. Marine Pollution Bulletin 10, 104-107. Edwards, A., Jones, K., Graham, J.M., Griffiths, C.R., MacDougall, N., Patching, J., Richard, J.M. & Raine, R. 1996. Transient Coastal upwelling and water circulation in Bantry Bay, a Ria on the Southwest Coast of Ireland. Estuarine, Coastal and Shelf Science 42, 213-230. Gaston, A.J. & Jones, I.L. 1998. The Auks, Bird Families of the World. Oxford University Press, New York. pp 349. Lutkebohle, T. 1995. Dolphin movements and behaviour in the Kessosck Channel and how these are influenced by boat traffic. Report to Scottish Natural Heritage. pp 37. Magurran, A.E. 1988. Ecological Diversity and its Measurement. Cambridge University Press. pp 179. McMahon, T., Raine, R. & Boychuk, S. 1995. Some oceanographic features of northeastern Atlantic waters west of Ireland. ICES Journal of Marine Science 52, 221-232. Raine, R., O`Mahony, J., McMahon, T. & Roden, C. 1990. Hydrography and Phytoplankton of waters off Southwest Ireland. Estuarine, Coastal and Shelf Science 30, 579-592. Smiddy, P. 1992. The effect of the Kowloon Bridge oil spill in east Cork. Irish Birds 4, 559-570.

93 Smiddy, P. 1998. The effect of the Cork Harbour Oil Spill of November 1997 on birds. Irish Naturalists Journal 26, 32-37. Smith, J. & McLaverty, A. 1997. The South West coast of Ireland. An Environmental Appraisal. BHP, Chevron, Marathon, Occidental, Statoil and Total, Ireland. pp 64. Webb, A., Stronach, A., Tasker, M.L. & Stone, C.J. 1995. Vulnerable Concentrations of Seabirds south and West of Britain. Joint nature Conservation Committee Würsig, B., Cipriano, F. & Würsig, M. 1991. Dolphin movement patterns: information from radio and theodolite tracking studies. In: K. Pryor & K.S. Norris (eds), Dolphin Societies, Discoveries and Puzzles. University of California Press., Berkeley. pp 79-111.

94

CHAPTER 3

Harbour seal adult with suckling pup (Photo. M. Cronin).

TEMPORAL VARIATION IN THE USE OF HAUL-OUT SITES BY HARBOUR SEALS IN BANTRY BAY AND THE KENMARE RIVER

95 3.1 ABSTRACT

The seasonal change in the distribution and abundance of harbour seals at terrestrial haul-out sites in southwest Ireland was investigated. Statistical models of the relationship between seal abundance at haul-out sites and covariates, such as the time of year and time of day and environmental conditions at haul-out sites, provided a means of assessing the influence of the covariates on the seals’ haul out behaviour and if this varied between sites. There was a difference in the seasonal patterns of seal abundance between haul-out sites. The effect of the time of day on seal abundance at haul-out sites varied between the sites and was only significant at sites that also showed a seasonal pattern in abundance. Fewer seals were observed during strong winds and rain. The differences in site use and covariate effects on haul-out behaviour are discussed in relation to the seals annual cycles and the physical characteristics of the sites.

96 3.2 INTRODUCTION

The harbour seal (Phoca vitulina L.) is the most widely-distributed pinniped species, inhabiting cold-temperate and temperate waters in the northern hemisphere on both sides of the north Atlantic and north Pacific oceans (Bigg, 1981). Harbour seals, are semi-aquatic mammals (Pitcher & McAllister, 1981) that spend time ashore at terrestrial sites on which they haul-out to rest, breed, moult, engage in social activity and escape predation (Pitcher & McAllister, 1981; Da Silva & Terhune, 1988; Thompson, 1989; Watts, 1992; Boily, 1995). Haul-out substrate varies across the harbour seals’ geographical range and includes tidal sand and mud bars, sand and gravel beaches, inter-tidal rocks and reefs and ice floes and glacial drift (Stewart, 1984).

Seasonal changes in the numbers of harbour seals ashore at terrestrial haul-out sites have been described (Thompson & Rothery, 1987; Thompson, 1989; Thompson et al., 1989; Thompson & Miller, 1990; Härkönen et al., 1999) explained largely by seasonal changes in haul-out behaviour, however, individual changes in site-use is also a contributory factor (Thompson, 1987, 1989) and seasonal variations in haul-out site use have also been described (Brown & Mate, 1983; Jeffries, 1986; Thompson, 1989; Thompson et al., 1994; Harding, 2000; Härkönen & Harding, 2001; Rehberg & Small, 2001; Reder et al., 2003). Harbour seals, previously considered to be site specific with generally limited movements (Pitcher & McAllister 1981, Brown & Mate, 1983), have been shown to use a range of haul-out sites, even within a particular season (Thompson, 1989) and long distance movements between haul-out sites have been observed (Lowry et al., 2001, Rehberg & Small, 2001; Sharples et al., 2004).

In addition to temporal factors, other factors or ‘covariates’ which influence haul- out behaviour and site use include time of day, tidal effects, disturbance and local weather (Stewart, 1984; Yochem et al., 1987; Thompson et al., 1989, 1994; 1997; Thompson & Harwood 1990; Thompson & Miller, 1990; Grellier et al., 1996; Reder et al., 2003). Examining the effects of these covariates on haul-out behaviour can help explain seasonal patterns in abundance and site use.

97 Haul-out sites on the Irish coast used by harbour seals during the 2003 annual moult were identified during a national aerial survey to determine a minimum population estimate for the species. The resulting information on the distribution and abundance of harbour seals on the Irish coastline was limited to the moult period (Cronin et al., 2004). Hitherto no information was available on the year round patterns in the distribution and abundance of harbour seals at specific haul-out sites in Ireland. The harbour seal is listed as an Annex II species under the EC Habitats Directive (92/43/EEC) which requires member states to designate Special Areas of Conservation (SACs) for the protection of listed species. Identifying the full range of sites used throughout the annual cycle is essential for the identification of and subsequent monitoring and management of SACs for harbour seals in response to the Directive.

Over one third of the national minimum population estimate of harbour seals use terrestrial haul-out sites in southwest Ireland (Cronin et al., 2004). Most of the harbour seal haul-out sites in this region are located within the Kenmare River and Bantry Bay. The Kenmare River and inner Bantry Bay have been designated as SACs with the harbour seal listed as one of the qualifying interests. The National Parks and Wildlife Service of the Department of Environment, Heritage and Local Government have monitored the abundance of harbour seals in both bays since 1985 (Heardman et al., 2006), however survey effort was limited to the summer months and no information is available on the year-round patterns in abundance and site use.

The objectives of the present study were (i) to describe seasonal changes in the abundance and distribution of harbour seals at individual haul-out sites within Bantry Bay and Kenmare River and (ii) to investigate the effects of factors such as month, time of day and weather on seal haul-out behaviour and the resulting distribution and abundance of seals at the haul-out sites.

98 3.3 MATERIALS & METHODS

3.3.1 Study area

Bantry Bay, County Cork (51o 36’N, 9o 50’W), a drowned river valley, is the longest marine inlet in southwest Ireland with a varied coastline ranging from exposed rocky shores to sheltered sediment shores at Whiddy Island. A detailed description of the physical characteristics of Bantry Bay is given in chapter 2. Haul-out sites used by harbour seals within Bantry Bay are predominantly located on the northern side of the bay, the exception being Gerrane rocks off Whiddy Island in the inner part of the bay (figure 1). Haul-out substrate is exclusively rocky and haul-out sites are generally on skerries or islands located adjacent to the mainland shore. The majority of these sites are in the inner part of Bantry Bay, in the northeast corner, Glengarriff harbour. Seals use a number of haul-out sites located further west, the most westerly of these are located in Adrigole harbour approximately 20km from the head of the bay. Ten main discrete haul-out sites have been identified, some comprising of smaller adjacent sites, and are shown in figure 2.

The Kenmare River, County Kerry (51o 43’N, 10o 05’W), is a partially mixed estuary and its inter-tidal areas are dominated by rocky shores. A detailed description of the physical characteristics of the Kenmare River is given in chapter 2. Haul-out substrate is exclusively rocky and haul-out sites are located on skerries and islands situated off the mainland shore. The majority of haul-out sites are located within sheltered bays on the Iveragh peninsula, the exception being Brennel Island and Ormond’s Island off the southerly shore and the most westerly of these sites are in Westcove harbour, approximately 30km from haul-out sites located towards the head of the river. A total of eleven discrete haul-out sites have been identified and shown in figure 3.

99

Figure 1. The study area in southwest Ireland.

100

1. Carrigskye 2. Inner Glengarriff harbour 3. Garinish Island 4. Big point rocks 5. Coolieragh harbour 6. Coulagh rocks 7. Garinish west 8. Orthans Island 9. Adrigole harbour 10. Whiddy Island

Figure 2. Harbour seal haul-out sites in Bantry Bay, Co. Cork, Ireland

1. Templenoe 2. Carrignaronomore 3. Brennel Island 4. Ormonds Island 5. Coongar harbour 6. Brown Island 7. Outer sneem harbour 8. Inner sneem harbour 9. Potato Island 10. Illaunsillagh 11. Westcove harbour

Figure 3. Harbour seal haul-out sites in the Kenmare River, Co. Kerry, Ireland

101 3.3.2 Seal counts Comprehensive scoping surveys of Bantry Bay and the Kenmare River were carried out in April 2003 in a Tornado 5.8m Rigid Inflatable Boat (RIB) and harbour seal haul-out sites were identified. Between April 2003 and November 2005 surveys of both bays were carried out by RIB and numbers of seals at haul-out sites counted using Leica 10 x 42 binoculars and recorded on a Sony Dictaphone. Counts of seals at each haul-out site were carried out independently and simultaneously by two observers and repeated if necessary until consensus was agreed. Counts were initially obtained from a distance of approximately 200m from the haul-out site and at progressively closer ranges whilst preventing disturbance to the seals.

New pups were recorded only when it was obvious they were pups of the year, identifiable by size and dark pelage. All other seals counted were considered as ‘non- pups’ or adults due to the difficulty in differentiating juveniles from adults from a distance.

Surveys were carried out on at least a monthly basis year-round and weekly during the summer and autumn, weather permitting. Surveys were scheduled to occur within two hours on either side of low tide and during daylight hours. Surveys began at the head or mouth of the bay on alternate survey dates so that haul-out sites would not always be surveyed in the same order within the four hour tidal period. On each survey the coastline between ‘established’ haul-out sites was also checked for the presence of seals to ensure that all possible haul-out sites used over the study period were identified.

3.3.3 Statistical modelling The effects of environmental variables on the numbers of adult seals hauled out during surveys were modeled using generalized additive mixed modeling (GAMM), a combination of generalized additive modeling (GAM) and mixed effects modeling (described in chapter 2). All statistical analyses were carried out using Brodgar v 2.5.1 (www.brodgar.com) software package and R v 2.3.1 statistical program.

The environmental variables investigated, as explanatory variables, are month, time of day, wind speed, wind direction and ‘weather’ (sun, cloud and rain). The response

102 variable is harbour seal abundance. The effects of time of day and month were modeled as smooth functions and the other effects modeled as categorical factors. The variable ‘week time’ was included to explore potential significant variation in the patterns in seal abundance over the entire study period. It was calculated as {year + (week number-1)/52}. There was no evidence of collinearity between any of the explanatory variables so they were all included in the subsequent analyses. An examination of validation plots resulting from GAMs showing patterns in the residuals of the explanatory variables, suggesting violation of homogeneity, in addition to high leverages suggested model misspecification and therefore GAMMs were applied.

The optimal GAMM model for the seal abundance data was selected using a two- step approach to search for the most optimal random and fixed components. Once the random components were selected, the most optimal model in terms of fixed components was explored (Fitzmaurice et al., 2004). The restricted maximum likelihood estimation (REML) was used to compare models with the same fixed terms and different random components (Pinheiro & Bates, 2000). The optimal model in terms of random components allowed for heterogeneity of variances at least by season and auto-correlation was added in the form of an auto-regressive model AR-1, allowing for correlation between residuals of sequential weeks.

To determine the most optimal model in terms of fixed terms a backward selection was carried out and those explanatory variables that were found not to be significant in explaining seal abundance in (p values >0.05) were dropped in turn and the AIC (Akaike Information Criteria) (Akaike, 1973) of resulting models compared following each step. To compare models with different fixed effects but with the same random components the maximum likelihood method was used instead of REML (Zuur et al., 2006).

The explanatory variables ‘month’ and ‘time of day’ were fitted using a 2- dimensional smoother allowing the effect of the time of the day to change per month. The reasons for this approach were based on observations made over the study period which suggested that the effect of time of day on seal abundance at haul-out sites may change over the year. A unique smoother per site was used so we could explore if the

103 effects of month and time of day on seal haul-out behaviour varied between sites. A smoother s(X) + s(Y) is nested within s(X, Y). If an s(X, Y) smoother is not significant, it is split into an s(x) and s(Y) smoother and the model re-run, and each smoother is dropped from the model if not significant. Furthermore, due to a ‘by’ command in the R code, the smoothers will not influence each other, enabling modifications to various smoothers simultaneously. This was continued in a stepwise fashion removing non-significant fixed terms and using F tests to confirm the improvements to the models.

104 3.4 RESULTS

3.4.1 Seal counts There was seasonal variation in haul-out site use by harbour seals in Bantry Bay and the Kenmare River over the study period. Seasons are denoted in this study as spring (February-April), summer (May-July) autumn (August-October) and winter (November-January). Patterns of seasonal abundance differed between haul-out sites. Within Bantry Bay sites 1 to 5 were generally used throughout the year with numbers increasing during the summer and autumn months. Highest numbers of seals were observed at sites 2 and 3. Sites 6 and 7 had limited use and generally only during summer months. Site 9 was used throughout the year but unlike the majority of sites within the bay, showed no obvious increase in numbers during summer/autumn and site 10 was only used during summer and autumn (Figure 4). Within the Kenmare River the majority of haul-out sites were used year round and the obvious summer increase in numbers, evident at most of the Bantry Bay sites was apparent only at sites 1, 2, 7 and 10. These sites also had relatively higher counts than other sites within the bay. Site 9 was used only during summer months (Figure 5).

Pups were recorded at all sites within Bantry Bay apart from sites 7 and 8. The most important sites for pupping within the bay, based on highest pup counts, were sites 2, 3 and 4. Within the Kenmare River pups were recorded at all sites apart from sites 8 and 11 and sites 2, 3 and 7 had the highest pup counts (Figures 4 & 5).

The seal count data, modelled as a function of the covariates time of year, time of day and weather describes how significantly different the patterns of seal abundance were between sites and the effect of the covariates on these patterns.

105 3.4.2 Model output and validation The optimal model for the seal abundance data from Bantry Bay contained terms for month, time of day, weather and site:

SAmwsy = fs1(Mm, TDmwsy) + fs2(Mm) + fs4 (Mm, TDmwsy)+fs8(Mm, TDmwsy) + fs10(Mm,

TDmwsy) + Ss + Wmwsy + εmwsy

2 εmwsy ~ N (0, σ s)

|'ww− | cor(,εεmwsy mw'sy )= φ

Where SAmwsy is the abundance of seals in month m, week w, site s, year y. The explanatory variables ‘month’ and ‘time of day’ were fitted using a 2-dimensional smoother f(Mm, TDmwsy) allowing the effect of time of the day to change per month. The effect of covariates on seal abundance at individual sites could be examined by

using a unique smoother per site (fs1 to fs10) and a ‘by’ command in the R code enabled modifications to various smoothers simultaneously. The model accounts for autocorrelation by allowing for correlation between residuals of sequential weeks

|'ww− | cor(,εεmwsy mw'sy )= φ

2 εmwsy ~ N (0, σ s) allows for different variances in each site (s).

The effect of time of day on seal abundance varies across the months however this is only significant at sites 1, 4, 8 and 10 (p< 0.001). At site 2 the effect of month on seal abundance is significant (p<0.01) but not time of day. At the remaining sites neither month nor time of day explain the patterns in seal abundance at the haul-out sites over the study period. The phi (φ) value of 0.294 denotes the correlation between sequential weeks (Table I). The significance of the term ‘site’ confirms the difference in patterns of abundance between sites (p<0.001). Individual p values per site explain which sites differ the most from site 1 and all but sites 5 and 9 show different patterns in abundance (p<0.05). The difference in the influence of ‘site’ on harbour seal abundance within Bantry Bay is shown in figure 6, the larger site labels represent the larger contribution from factor ‘site’ (taken from the ANOVA output). The importance of sites 2, 3 and 10 as haul-out sites for seals is apparent however there is also high variation in abundance and relatively high residual variances at these sites.

106 The residual variance component per site (variance of the ‘noise’ or information unexplained by the model) is shown in Table I and sites 2 and 3 show the largest residual variances. Box-plots of the abundance of seals conditional on site confirm that these sites have higher variation in abundance. The relative difference in residual variances per site is shown in figure 7, with sizes of site labels proportional to the variances per site. The larger residual variances may be explained by other processes influencing abundance at these particular sites that were not included in the models (such as disturbance).

Weather had an effect on the numbers of seals counted, significantly higher numbers (p<0.01) were observed in sunny weather (category 3).

The 2-D smoothing functions for sites where there was a significant effect of month and time of day on seal abundance, are shown in Figure 8. At site 1 the interactive effect of time of day and month is evident. The time of day does not appear to be as influential on abundance of seals at this site in spring and later in the year but in the summer months there appears to be an obvious effect of time of day, with numbers peaking later in the day and decreasing at night. The same summer peak is seen at site 4 but there is not as sharp a rise in abundance as evident in site 1. In addition, the peak in abundance is evident earlier in the day than at site 1, with the probability of high numbers of seals at this site remaining high later in the day. A different pattern is apparent at site 8 with probability of highest abundance in the spring decreasing steadily throughout the year to an autumn/winter low. Time of day appears to only have an effect in the spring at this site, when afternoon peaks are apparent, decreasing rapidly later in the day. The smoothing functions from site 10 show a clear seasonal pattern similar to site 1 with highest abundance in the summer months falling rapidly following this peak. The time of day has a significant effect during the summer months also, as in site 1, with abundance increasing steadily throughout the day to an evening peak and falling at night. Overall, significant seasonal patterns in abundance are apparent for sites 1, 2, 4, 8 and 10 and in all but site 2 there is also an effect of time of day on seal abundance. This effect changes across the year and both the effect of time of day and its interaction with month varies between sites.

107 Exploration of the seal abundance data from the Kenmare River involving co-plots, lattice plots and scatter-plots led to a slightly different approach than with the Bantry Bay seal abundance data. The co-plots and lattice-plots suggested including the same fixed terms as in the initial model of the Bantry Bay dataset. However, patterns in the residuals suggested different sites have different residual spread and in addition there was also apparent heterogeneity both between and within seasons. Using site as a variance component as in the previous analysis, will not account for the heterogeneity. Spatial correlation in the data was checked for by recoding site, based on regional distances (figure 9), into four areas and the heterogeneity of residual variance examined (figure 10). No difference in the spread of residuals based on spatial proximity was apparent. Square-root transformation of the response variable i.e. the count data can be used to stabilise the relationship between the mean and the variance (Zuur et al., 2006).

The optimal model for the seal abundance data from Kenmare River contained terms for month, time of day, wind speed, year and site:

√SAmwsy = fs1(Mm, TDmwsy) + fs2(Mm, TDmwsy) + fs4(Mm) + fs5(Mm, TDmwsy) + fs6(Mm) +

fs7(Mm, TDmwsy) + fs9(Mm, TDmwsy) + WindSpeedmwsy + Ss + Yy + εmwsy

2 εmwsy ~ N (0, σ s)

|'ww− | cor(,εεmwsy mw'sy )= φ

Where √SAmwsy is the square root of the abundance of seals in month m, week w, site s, year y. The explanatory variables ‘month’ and ‘time of day’ were fitted using a 2-

dimensional smoother f(Mm, TDmwsy) allowing the effect of time of the day to change

per month. Unique smoothers per site are denoted by fs1 to fs10. The model accounts for autocorrelation by allowing for correlation between residuals of sequential weeks and allows for different variances in each site (s). Seal abundance varies significantly over the year at all sites apart from sites 3, 8, 10 and 11. As was evident in the analyses of the abundance data for the entire bay, the time of day has an influence on seal abundance, however not at all sites. The time of day significantly effects seal abundance only at sites 1, 2, 5, 7 and 9 and the effect

108 varies over the year at these sites (P<0.01) (Table II). The significance of the term ‘site’ confirms the difference in patterns of abundance between sites (p<0.001).

Individual p values per site explain which sites differ the most from site 1 and sites 2, 7 and 10 show different patterns in abundance (p<0.01). The difference in the influence of ‘site’ on harbour seal abundance within the Kenmare River is shown in figure 11, the larger site labels represent the larger contribution from factor ‘site’ (taken from the ANOVA output). The importance of sites 7 and 10 as haul-out sites for seals within the Kenmare River is apparent. The residual variances at all sites are relatively similar (figure 12).

Wind speed had an effect on the numbers of seals counted, significantly fewer seals (p<0.05) were observed in strong winds (Beaufort 5 or above). A significant year effect was evident with significantly more seals counted in 2005 (P<0.01).

The 2-D smoothing functions for sites where there was a significant effect of month and time of day on seal abundance, are shown in figures 13 & 14. At site 1 the interactive effect of time of day and month is evident. This effect changes significantly over the year, with higher numbers of seals earlier in the day during the summer months but later in the day in the autumn and winter, as seen in the ‘flip’ in the smoother. Site 2 shows the same pattern that was apparent at some of the Bantry Bay sites (sites 1 and 10), with time of day influencing seal numbers only during summer with a steady increase from morning until evening. There was no significant effect of the time of day on seal numbers at site 4 and a weak effect of month with numbers increasing gradually with highest counts at the end of the year; confidence intervals are large however particularly at the start and end of the year. Seal numbers at site 5 were highest early in the day at the start of the year and the time of day had progressively less of an effect throughout the year. At site 6 seal abundance was highest during the summer months, confidence intervals are also large at the start and end of the year and the time of day did not significantly influence abundance. Abundance peaked early in the year also at site 7 but late in the day and the time of day was not influential throughout the remainder of the year. A summer peak in abundance occurred at site 9 and a diurnal time of day effect was apparent with early morning and late afternoon peaks in abundance. At other times of the year the time of

109 day did not appear to effect seal abundance at this site. Overall, significant seasonal patterns in abundance are apparent for sites 1, 2, 4, 5, 6, 7 and 9 and in all but sites 4 and 6 there is also an effect of time of day on seal abundance. This effect changes across the year and both the effect of time of day and its interaction with month varies between sites

110

140 Site 1 140 Site 6

120 Adults and juveniles 120 100 Pups 100 80 80 60 60 40 40

20 20 *** 0 0 ***

140 Site 2 140 Site 7 120 120 100 100 80 80

60 60 40 40 20 20 0 * ** 0 * * *

140 Site 8 140 Site 3 120 120 100 100 80 80 60 60 40 40 20 20 0 * * * 0 * * * *

140 Site 9 140 Site 4 120 120 100 100 80 80 60 60 40 40 20 20 0 * * * ** 0 * * *

Site 10 Site 5 140 140 120

t 120 n

u 100 o 100 c l 80 a 80 60 r se 60 u o 40 rb 40 a

H 20 20 0 * * * 0 * * * 3 4 5 3 4 3 5 4 5 3 3 4 4 4 5 5 3 3 4 4 5 4 5 3 4 5 5 4 3 5 3 4 4 3 5 4 3 4 5 3 0 0 0 5 3 4 5 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 03 0 04 04 0 05 05 0 0 0 ------05 04 03 0 0 0 ------0 0 0 0 0 05 04 03 04 - - - 0 0 0 05 04 03 05 04 - - - l l l - - - t t r-0 t r-0 r-0 ------r-0 r-0 c c n n n l l l v v p- p- b- p- b- t t t r-0 r-0 r-0 r-0 r-0 g g g c c n- n- p p p ug- ug- ug- p- p- p- b- b- e e e e e Ju Ju Ju p p p a un- un- a un- e e ov- e ov- e e Ju Jan Ju Jan Ju Ju Ju Ju Oc Oc Oc S A S F A S F A J J J J J A De A De A Ma Ma No No Oc Oc Oc May May May S S S F A F A A De Au De Au Au Ma Ma N N May May May Month

Figure 4. Maximum harbour seal counts at haul-out sites in Bantry Bay, Co. Cork April 2003 to October 2005 (*= no count)

111

100 Site 1 100 90 Adults and juveniles 90 Site 6 80 80 70 Pups 70 60 60 50 50 40 40 30 30 20 20 10 10 0 * * 0 * *

100 Site 7 90 100 Site 2 80 90 70 60 80 50 70 40 60 30 50 20 40 10 0 * * 30 20 10

0 * * 100 90 Site 8 80 70 60 100 Site 3 50 90 40 80 30 70 20 * 60 10 0 * 50 40 30 20 10 100 Site 9 90 0 * * 80 70 60 50 40 30 100 Site 4 90 20 10 80 0 * * 70 60 50 40 100 30 Site 10 90 20 80 10 70 * * 0 60 50 40 30 Site 5 20 10 100 0 **** 90

t 80

un 70

co 60 Site 11 50 100

seal 90

r 40 80

ou 30

b 70 r 20 a 60

H 10 50 0 * * 40 3 4 5 3 3 4 4 4 5 5 3 3 4 4 5 3 4 4 5 30 03 04 05 0 0 0 03 04 05 0 0 0 0 0 0 0 0 0 0 0 0 0 0 03 04 04 05 05 ------l l l t t r r r r-0 r-0 c c v v p- b- p- b- p- 20 a a ug- ug- ug- un- un- un- e e e e e Ju Ju Ju J Jan J Jan J Oc Oc Ap S F Ap S F Ap S A De A De A M M No No 10 May May May 0 * ** * 3 4 5 3 4 3 4 5 3 4 4 Month 5 03 04 05 0 0 0 03 04 05 0 03 04 0 04 05 05 03 04 0 0 0 03 04 0 04 05 05 ------l l l t r-0 t r-0 r-0 r g- r g- g- c c y- y- n- y- n- p- b- p- b- p- p p p u u u a un- a a un- a a un- e e ov- e e ov- e Ju Ju Ju J J J J J Oc Oc A S F A S F A S A De A De A Ma Ma N N M M M

Figure 5. Maximum harbour seal counts at haul-out sites in Kenmare River, Co. Kerry, April 2003 to September 2005 (* = no count)

112

0 00 22222222222222222222222222222222222222222222222222 56

0 33333333333333333333333333333333333333333333333333 0 0

4 4444444444444444444444444444444444444444444444444 5 s

g n i 0 11111111111111111111111111111111111111111111111111 h t 00

r o

52

N 55555555555555555555555555555555555555555555555555

66666666666666666666666666666666666666666666666666 77777777777777777777777777777777777777777777777777 000 999999999999999999999999999999999999999999 50 888888888888888888888888888888888888888888 1010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010

80000 85000 90000 95000

E astings

Figure 6. The difference in the influence of ‘site’ on harbour seal abundance within Bantry Bay, the larger site labels represent the larger contribution from factor ‘site’.

22222222222222222222222222222222222222222222222222 56000 33333333333333333333333333333333333333333333333333

54000 4444444444444444444444444444444444444444444444444

ngs 11111111111111111111111111111111111111111111111111

hi t r o

52000 55555555555555555555555555555555555555555555555555

N

66666666666666666666666666666666666666666666666666 77777777777777777777777777777777777777777777777777

999999999999999999999999999999999999999999 50000 888888888888888888888888888888888888888888 1010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010

80000 85000 90000 95000

Eastings Figure 7. The relative differences in variances of harbour seal abundance per site within Bantry Bay with sizes of site labels proportional to the variances per site.

113

Site Site

s s e e a a l l

a a

b b u u n n d d a a n n c y c e y a e d a f d o f e o e m i M m on T M i th o T nth

Site Site 10

s s e e a a l l a a b b u u n n d d a a n n c c e y e y a a d d f f o o e e m m M i M i o T on T nth th

Figure 8. 2-D smoothing function describing the partial effect of month (1 to 12) and time of day(08.00 to 18.00) on the abundance of harbour seals at haul-out sites in Bantry Bay.

114

11111111111111111111111111111111111111111111111 0

0 22222222222222222222222222222222222222222222222 0 8 6 33333333333333333333333333333333333333333333333

000

66 55555555555555555555555555555555555555555555555 s g

n 44444444444444444444444444444444444444444444444 i h t

000 88888888888888 r 4 6 No 77777777777777777777777777777777777777777777777 66666666666666666666666666666666666666666666666 9999999999999999999999999999999999999999999999

0 200 6

00 1010101010101010101010101010101010101010101010101010101010101010101010101010101010101010 600 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111

60000 65000 70000 75000 80000 85000

Easti ngs

Figure 9. Spatial proximity of sites within the Kenmare River

40

20 s al du i s e

R 0

0 2 -

0 204060

Colors Sibyte si codete redcode by dcolours Figure 10. Residual values versus sites within the Kenmare River (colours denote sites recoded into four regions based on spatial proximity)

115

11111111111111111111111111111111111111111111111 0 0

0 22222222222222222222222222222222222222222222222 8 6 33333333333333333333333333333333333333333333333

0 600

6 55555555555555555555555555555555555555555555555

s g 44444444444444444444444444444444444444444444444 ng tin hi 88888888888888 t 000 r o 64 N

77777777777777777777777777777777777777777777777 66666666666666666666666666666666666666666666666 North 9999999999999999999999999999999999999999999999 0 0

0 2 6

0

000 1010101010101010101010101010101010101010101010101010101010101010101010101010101010101010 6 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111

60000 65000 70000 75000 80000 85000 Eastings Eastings Figure 11. The difference in the influence of ‘site’ on harbour seal abundance within the Kenmare River, the larger site labels represent the larger contribution from factor ‘site’

11111111111111111111111111111111111111111111111

0 22222222222222222222222222222222222222222222222

6800 33333333333333333333333333333333333333333333333

0

6600 55555555555555555555555555555555555555555555555 ng hi s rt 44444444444444444444444444444444444444444444444 ng No hi 00 t

r 88888888888888

o 640 N 77777777777777777777777777777777777777777777777 66666666666666666666666666666666666666666666666 9999999999999999999999999999999999999999999999 00 620

00 1010101010101010101010101010101010101010101010101010101010101010101010101010101010101010 600 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111 60000 65000 70000 75000 80000 85000 Eastings Eastings Figure 12. The relative differences in variances of harbour seal abundance per site within the Kenmare River with sizes of site labels proportional to the variances per site

116

Site Site

s e s a e l a a l

b a u b

n u

d n a d n a

c n e c y e a d y f a o d e f o m Mo i e nt T M m h o i nth T

Site Site

s e s

a e l a

a l

b a

u b n u d n

a d n a c n

e c y e a y d f a d o e of M m e o i T M m nth o i nth T Site

s e

a

l

a

b

u n d

a

n c e y a d f o e M m o i nth T

Figure 13. 2-D smoothing function describing the partial effect of month (1 to 12) and time of day (08.00 to 18.00) on the abundance of harbour seals at haul-out sites in Kenmare River

117

Site 4 Site 6

3 3

2 2

1 1 ) ) 4 97 2 . 1. 2 0 0 h,

h, nt nt o o M M ( ( s s

1 - -1

-2 -2 -3 3 -

24681012 24681012 Month Month

Figure 14. Smoothing functions describing the partial effect of month on the abundance of harbour seals at haul-out sites within the Kenmare River.

118

Table I. Summary of optimum generalized additive mixed models of seal abundance within Bantry Bay. For explanatory variables fitted as smoothers the estimated degrees of freedom (edf) are shown and for parametric terms the degrees of freedom (df) shown. For all explanatory variables, including 2D smoothing functions, the associated probability value (p) is given. The Phi value resulting from the auto-correlation structure and the residual variance co mponent by site are shown.

Site Month/ Month Weather Site Phi Variances Time of day Overall 0.294 d.f. 2 9 p value p<0.01 p<0.001 Site 1 1.000 e.d.f. 10.73 p value p<0.001 p<0.001 Site 2 2.437 e.d.f. 1 p value p<0.01 p<0.001 Site 3 3.723 e.d.f. p value p<0.001 Site 4 1.236 e.d.f. 13.81 p value p<0.001 p<0.001 Site 5 1.161 e.d.f. p value p=0.797 Site 6 1.097 e.d.f. p value p<0.05 Site 7 0.717 e.d.f. p value p<0.001 Site 8 0.751 e.d.f. 13.67 p value p<0.001 p<0.05 Site 9 1.102 e.d.f. p value p=0.219 Site 10 1.575 e.d.f. 11.44 p value p<0.001 p<0.05

119

Table II. Summary of optimum generalized additive mixed models of seal abundance within the Ken mare River. For explanatory variables fitted as smoothers the estimated degrees of freedom (edf) are shown and for parametric terms the degrees of freedom (df) shown. For all explanatory variables, including 2D smoothing functions, the associated probability value (p) is given. The Phi value resulting from the auto-correlation structure and the residual variance

Site Month/ Month Wind speed Year Site Phi Variances Time of day Overall 0.227 d.f. 5 4.08 10 p value p<0.001 p<0.05 p<0.001 Site 1 1.000 d.f. 7.71 p value p<0.001 Site 2 0.857 d.f. 11.68 p value p<0.001 p<0.01 Site 3 1.049 d.f. p value p=0.559 Site 4 0.749 d.f. 1.07 p value p<0.05 p=0.617 Site 5 0.596 d.f. 2 p value p<0.001 p=0.051 Site 6 0.701 d.f. 2.24 p value p<0.05 p=0.981 Site 7 0.958 d.f. 7.2 p value p<0.01 p<0.001 Site 8 1.129 d.f. p value p=0.157 Site 9 0.869 d.f. 12.06 p value p<0.001 p=0.06 Site 10 0.624 d.f. p value p<0.001 Site 11 0.832 d.f. p value p=0.546

120 3.5 DISCUSSION

Harbour seals use a number of haul-out sites within Bantry Bay and the Kenmare River throughout their annual cycle. There was a difference in the seasonal patterns of seal abundance between these sites. Modeling the numbers of harbour seals at the sites, as a function of variables or covariates such as the time of day, time of year and environmental parameters provided information on what factors were driving or influencing the patterns in abundance observed. Within Bantry Bay a significant seasonal pattern in abundance was evident at five of the ten haul-out sites. At Carrigskye (site 1), Whiddy Island (site 10) and Big Point rocks (site 4) a late summer peak in abundance occurred, more markedly at the former two sites. A spring peak occurred at Orthan’s Island (site 8) and within inner Glengariff harbour (site 2) a steady rise in abundance was evident over the year. The other haul-out sites showed some seasonal changes in attendance, although not significantly so, apart from inner Adrigole harbour (site 9), which was used by relatively consistent numbers of seals throughout the annual cycle. All sites were used, albeit to different extents, throughout the year with the exception of Whiddy Island which was not used between the months of November to March.

The observed patterns in the abundance of seals at haul-out sites can largely be explained by events in the seals’ annual cycle. The late summer increase in abundance at Carrigskye, Whiddy Island and Big Point rocks is likely a result of these sites being used for the annual moult. Moulting seals were observed at these sites during three consecutive annual moults during August and September. Haul-out sites within Glengarrif harbour, including Garinish Island (site 3) were also used by seals as moulting sites. The increase in abundance around the time of moult was not as apparent in Glengarrif harbour but this is probably due to the fact that the sites within the harbour are used throughout the year and the relative increase in abundance during the moult is not as apparent as at other sites. Haul-out sites within Glengarrif harbour, including the inner harbour, Garinish Island and the rocks at Big Point in the outer harbour, were used as breeding sites during the study period, evident from the higher numbers of pups observed at these sites.

121 In general site use was lower in winter, probably resulting from seals spending a higher proportion of their time at sea, suggested by the behaviour of tagged seals in the study area during this period (chapter 4). Glengarrif harbour and Adrigole harbour, in inner and outer Bantry Bay respectively were used by seals throughout the year, possibly because of the shelter they afford to seals during adverse weather. All other haul-out sites within Bantry Bay are located on islands or rocky skerries outside of sheltered harbours and exposed to strong winds and swell in winter. Within Adrigole harbour a spring increase in numbers at Orthan's Island may be explained by the sites relative proximity to potential foraging grounds, it is the nearest site to the mouth of Bantry Bay and the open sea. Studies on the activity patterns of tagged seals in Scotland suggest that they spend less time in inshore waters in winter (Thompson et al., 1989; Sharples, 2005) and using haul-out sites closer to foraging areas would be energetically less demanding. However in the absence of information on foraging behaviour of harbour seals in southwest Ireland, the potential association of haul-out site use with proximity to foraging areas should be interpreted with caution.

Overall within Bantry Bay significantly higher numbers of seals were observed throughout the study period within Glengarrif harbour, including the inner harbour and Garinish Island than at other haul-out sites in the bay. These sites, important pupping sites, were also used at other times during the annual cycle. The sites physical characteristics provide optimal haul-out conditions for practically all stages of the annual cycle; steeply shelving rocky skerries providing immediate access to deep water located within a sheltered harbour in the innermost part of Bantry protected from adverse weather conditions. The high variation in counts observed at these sites and high variances in residuals seen when abundance was modelled as a function of covariates, suggests other factors not measured may be influential in determining patterns of abundance at this location. Disturbance is a probable one as Glengarrif harbour has a busy tourism industry, in particular during the summer months. Ferries servicing Garinish Island pass within 20m of the haul-out sites within the harbour and along with pleasure craft, including yachts, kayaks and power-boats, have been observed to disturb seals from haul-out sites on occasion. Anthropogenic disturbance, or potential of, has been shown to influence the selection of haul-out site by harbour seals (Montgomery, 2005). As disturbance can lead to seals avoiding or abandoning haul-out sites (Pauli & Terhune, 1987; Da Silva & Terhune, 1988), the

122 importance of Glengarrif harbour for breeding seals along with the relatively high levels of anthropogenic disturbance during the breeding period should be important considerations in the management of this SAC.

Within the Kenmare River a significant seasonal pattern in abundance was evident at seven of the eleven haul-out sites. At Carrignaronomore (site 2) and Potato Island (site 9) a summer peak in abundance occurred. Both of these sites are used as moult sites and the former is an important pupping site within the Kenmare River. Brennel Island (site 3) and outer Sneem harbour (site 7) are also used as pupping sites, but as both of these sites are used throughout the year, the significant increase in seal numbers during summer is not as apparent. A winter peak in abundance was evident at the most easterly haul-out site within the Kenmare River at Templenoe (site 1) and numbers in Coongar harbour (site 5) and outer Sneem harbour peaked in the spring. All haul-out sites within the Kenmare River are used by seals, albeit to different extents, throughout the year with the exception of Potato Island which is used exclusively between June and September. Pups have been recorded at this site but it appears to be primarily associated with the moult. Most haul-out sites within the Kenmare River, apart from those located in the inner bay, are situated in sheltered harbours apart from Potato Island and the exposed nature of this site may explain the limited use of it by seals during the winter months. Overall significantly higher numbers of seals used outer Sneem harbour and Illaunsillagh compared to other sites within the Kenmare River throughout the study. Both of these sites also provide optimal haul-out conditions for all stages of the annual cycle and afford significant shelter from the prevailing winds as well as the heavy swell that frequently occurs during winter months in the outer Kenmare River region.

No seasonal effects were obvious at some haul-out sites within Bantry Bay and the Kenmare River and other factors may be influential in determining patterns in abundance at these sites. The seasonal shift in availability of food resources is another factor potentially influencing site use. Harbour seals have been observed to switch sites to move closer to feeding grounds on west coast US (Brown & Mate, 1983; Jeffries, 1986; Montgomery, 2005) however this has not been observed in studies of harbour seal site use in Orkney, Scotland (Thompson, 1989). It is likely that temporal and spatial changes in food availability, varies widely geographically, as does the

123 subsequent influence of this on haul-out site selection. Ongoing research into the foraging ecology and habitat use of harbour seals in the study area involving telemetry, dietary studies and photo-identification of individuals will explore this.

Other general factors considered to play an important role in the year-round selection of haul-out sites are substrate type, distance from human disturbance, shelter from prevailing winds and immediate access to deep water (Bigg, 1981; Scheffer & Slipp, 1944; Bjorge et al., 2002; Montgomery, 2005). Haul-out site selection by seals across the year may be determined by the physical characteristics of a site fulfilling particular physiological or behavioural requirements. The use of haul-out sites exclusively for pupping, have been observed (Vaughan, 1971; Jeffries, 1986). Selected pupping sites generally have immediate access to deep water and are away from human disturbance and other con-specifics, as females drive other seals away from their pups (Thompson 1987, 1989, Montgomery, 2005). Interestingly the main sites used for pupping in Bantry Bay and the Kenmare River are those most exposed to human disturbance, primarily from ferries, boat-based eco-tourism and leisure craft. These haul-out sites, generally found near the head of the bays, are the most sheltered sites relative to all haul-out sites within the two bays, affording seals protection from large swell that frequently occurs in the bays. Additionally these sites are located in deeper water than that found in the immediate vicinity of other haul-out sites within the bays. Such advantages may outweigh the costs of potential disturbance and the reaction of seals to passing boats varied largely between haul-out sites, suggesting potential habituation to disturbance at some sites.

Similar patterns of seasonal variation in abundance differing between sites have been shown in other studies, with some sites used predominantly during the breeding season, peak numbers occurring at other sites during winter (Brown & Mate, 1983; Payne & Schneider, 1984; Thompson, 1989) and other sites used throughout the whole year (Riseborough et al., 1980; Thompson, 1989). Seasonal changes in haul- out behaviour and subsequent abundance at haul-out sites in the area, together with changes in individuals’ use of particular sites are likely to explain different trends in abundance between sites (Thompson, 1987). The actual composition of haul-out groups has been shown to vary seasonally, with the heterogeneity in behaviour explained largely by the different requirements of individuals of different age and sex

124 throughout various phases of the annual cycle (Thompson et al., 1989; Härkönen et al., 1999). Information on the sex and age composition of haul-out groups in the study area would be useful for further explaining the patterns in haul-out behaviour at sites within the area.

The effect of the time of day on seal abundance varied between the haul-out sites and was only significant at sites that also showed a seasonal pattern in abundance. The effect was observed to change over the year. At sites where abundance peaked in the spring, at Orthan’s Island in Bantry Bay and Coongar harbour and outer Sneem harbour in the Kenmare River afternoon peaks in abundance were evident at that time of year. At sites where there was a summer or autumn peak in abundance, at Carrigskye, Big Point rocks, and Whiddy Island in Bantry Bay and at Carrignaronomore, and Potato Island in the Kenmare River, this occurred later in the day and remained high in the evening. Generally those sites where there was a summer or autumn peak in abundance were used for moulting and haul-out habitat was available during all tidal states. Sites used during the moult typically have habitat available above high tide allowing seals to spend more time ashore (Jeffries, 1986; Thompson, 1987, 1989). Excessive heat loss can occur if animals remain in the water during moult (Boily, 1995) and seals generally spend more time ashore during moult (Stevik et al., 2002); this may explain the temporal change in diurnal haul-out patterns observed during the moulting period.

Wilson (1978) suggested that in areas where habitat for hauling out is available above the high water level, diurnal cycles may be more influential than tidal cycles on haul-out behaviour. At all sites that had a seasonal pattern in abundance, the significant effect of the time of day on haul-out behaviour was evident apart from inner Glengarrif harbour in Bantry Bay and at Ormond’s Island and Brown Island in the Kenmare River. This might be explained by the inter-tidal nature of the rocky skerries used at these haul-out sites, and the tidal cycle may possibly be the main factor influencing seals’ use of these sites. As all counts were carried out at low tide the potential influence of the tidal cycle on haul-out site use could not be explored. Counts of seals conducted throughout the full tidal cycle at a range of sites in the study area would identify tidally influenced site use.

125 Wind speed influenced seal abundance at haul-out sites in the Kenmare River only, where significantly fewer seals were observed in strong winds. When the effects of wind speed on the combined seal counts from all sites within both bays was explored, the strength of the wind appeared not to significantly affect seals’ haul-out behaviour (chapter 2); combining data from both bays apparently masked the effect of wind speed on seal haul out behaviour and therefore numbers at sites in the Kenmare River. Haul-out sites within Bantry Bay are mostly located in the inner part of the bay and are afforded more shelter from the prevailing west/south-westerly winds relative to haul-out sites in the outer Kenmare River area. The lower numbers of seals observed during rain could be related to increased difficulties associated with counting in rain however seals were observed on a number of occasions entering water in heavy downpours. Both of these covariates have been shown to effect harbour seal haul-out behaviour in a number of other studies with higher numbers hauled out when winds are not strong (Venables & Venables 1955; Bishop, 1968; Boveng et al., 2003) and rain not heavy (Pauli & Terhune, 1987; Olesiuk et al., 1990; Grellier et al., 1996; Boveng et al., 2003).

For the effective designation and monitoring of potential and existing SACs for which the harbour seal is a qualifying interest, it is essential to identify the full range of sites used during the seals annual cycle. Additionally, determining the year round patterns in site use highlights sites of particular importance during different parts of the annual cycle. Local declines in harbour seal abundance in Orkney, Scotland may have resulted from local redistribution, suggesting site use may be flexible over long time periods and highlighting the importance of identifying and protecting a broad range of sites within an SAC (Thompson et al., 2001).

126 3.6 REFERENCES

Akaike, H. (1973). Information theory as an extension of the maximum likelihood principle. In B.N. Petrov & F. Caski (eds) Second International Symposium on Information Theory. Akademiai Kiado, Budapest, Hungary. pp 267-281. Adkinson, M. D., Quinn, T. J. & Small, R. J. (2003). Evaluation of the Alaska harbour seal (Phoca vitulina) population survey: A simulation study. Marine Mammal Science, 19, 764-790. Bigg, M. A. (1981). Harbour seal Phoca vitulina Linnaeus, 1758, and Phoca largha, Pallas, 1811. In: Ridgeway, S.H. and Harrison, R.J. (eds.), Handbook of Marine Mammals, Seals, 2, 1-77, Academic Press Inc., Ltd, London. Bishop, R. H. (1968). Reproduction, age determination and behaviour of the harbour seal (Phoca vitulina L.) in the Gulf of Alaska. Unpubl M.Sc. Thesis, University of Alaska. Boily, P. (1995). Theoretical heat flux in water and habitats selection of phocid seals and beluga whales during the annual moult. Journal of Theoretical Biology, 172, 235-244. Bonner, W. N. (1972) The Grey seal and Common seal in European waters. Oceanographic Marine Biology Annual Review, 10, 461-507. Boveng, P. L., Bengston, J. L., Withrow, D. E., Cesarone, J. C., Simpkins, M. A., Frost, K. J., & Burns, J. J. (2003). The abundance of harbor seals in the Gulf of Alaska. Marine Mammal Science, 19, 111-127. Brown, R. F. & Mate, B. R. (1983). Abundance, movements and feeding habits of the harbour seal, Phoca vitulina, at Netarts and Tillamook Bays, Oregon. Fisheries Bulletin, U.S. Nat. Ocean. Atmos. Admn. 81, 291-301. Cronin, M, Duck, C. O’Cadhla, O., Nairn, R., Strong, D. & O’Keefe, K. (2004) An assessment of population size and distribution of harbour seals (Phoca vitulina vitulina) in the Republic of Ireland during the moult season in August 2003. Biological Conservation( in review) Croxall, J. P., Everson, I., Kooyman, G. L., Ricketts, C. & Davis, R. W. (1985). Fur seal diving behaviour in relation to vertical distribution of krill. Journal of Animal Ecology, 54, 1-8. Da Silva, J. & Terhune, J. M. (1988). Harbour seal grouping as an anti-predator strategy. Animal Behaviour, 36, 1309-1316.

127 Dalgaard, P. (2002). Introductory statistics with R. Statistics and Computing. Springer. Faraway, J. J (2006). Extending the linear model with R. Chapman & Hall/CRC. Fitzmaurice, G. M., Laird, N. M. & Ware, J. H. (2004). Applied Longitudinal Analysis. Wiley. Frost, K. J., Lowry, L. F. & Ver Hoef, J. M. (1999). Monitoring the trend of harbour seals in Prince William Sound, Alaska after the Exxon Valdez oil spill. Marine Mammal Science, 15, 494-506. Grellier, K., Thompson, P. M. & Corpe, H. M. (1996). The effect of weather conditions on harbor seal (Phoca vitulina) haul-out behaviour in the Moray Firth, northeast Scotland. Canadian Journal of Zoology, 74, 1806-1811. Harding, K.C. (2000). Population dynamics of seals: the influences of spatial and temporal structure. Unpublished. PhD thesis, University of Helsinki, Helsinki. 35pp. Härkönen, T. K., Harding, C. & Lunneryd, S. G. (1999). Age and sex specific behaviour in harbour seals Phoca vitulina leads to biased estimates of vital population parameters. Journal of Applied Ecology, 36, 825-841. Härkönen, T. & Harding, K.C. (2001). Spatial structure of harbour seal populations and the implications thereof. Canadian Journal of Zoology, 79, 2115-2127. Hastie, T. J. & Tibshirani, R. J. (1990). Generalised Additive Models. Chapman & Hall, London. Heide- Jorgensen, M. P. & Härkönen, T. (1988). Rebuilding seal stocks in the Kattegat-Skagerrak. Marine Mammal Science, 4, 231-246. Huber, H.R., Jeffries, S.J., Brown, R.F., Delong, R.L., & Vanblaricom, G. (2001). Correcting aerial survey counts of harbor seals (Phoca vitulina richardsi) in Washington and Oregon. Marine Mammal Science, 17, 276-293. Jeffries, S. J. (1986). Seasonal movements and population trends of harbour seals (Phoca vitulina richardsi) in the Columbia River and adjacent waters of Washington and Oregon: 1976-1982. Report to the US Marine Mammal Commission, Contract No: MM30793575. Jemison, L.A. & Kelly, B.P. (2001). Pupping phenology and demography of harbor seals (Phoca vitulina richardsi) on Tugidak Island, Alaska. Marine Mammal Science, 17, 585-600.

128 Mathews, E.A. & Kelly, B.P. (1996). Extreme temporal variation in harbor seal (Phoca vitulina richardsi) numbers in Glacier Bay, a glacial fjord in S.E. Alaska. Marine Mammal Science, 12, 483-489. Olesiuk, P. F. (1999). An assessment of the status of harbour seals (Phoca vitulina) in British Columbia. Canadian Stock Assessment Secretariat Research Document 99/33. Fisheries and Oceans Canada, Ottawa, Ontario, Canada. 130 pp. Pinheiro, J. C & Bates, D. M. (2000). Mixed-effects models in S and S-Plus. New York Springer. Pitcher, K. W. & McAllister, D. C. (1981). Movements and haul-out behaviour of radio-tagged harbor seals, Phoca vitulina. Canadian Field Naturalist, 95, 292- 297. Reder, S., Lydersen, C., Arnold, W. & Kovacs, K. M. (2003). Haul-out behaviour of high Arctic harbour seals (Phoca vitulina vitulina) in Svalbard, Norway. Polar Biology, 27, 6-16. Rehberg, M. J. & Small, R. J. (2001). Dive behaviour, haulout patterns and movements of harbour seal pups in the Kodiak archipelago, 1997-2000. In: Harbor seal investigations in Alaska. Annual report for NOAA, award NA87FX0300. Alaska Department of Fish & Game, Division of Wildlife Conservation, Anchorage, AK. pp. 209-238. Rejenders, P., Abt., K., Brasseur, S., Tougaard, S., Siebert, U. & Vareschi, E. (2003). Sense and sensibility in evaluating aerial counts of harbour seals in the Wadden Sea. Wadden Sea Newsletter, 1, 9-12. Simpkins, M. A., Withrow, D. E., Cesarone, D. E. & Boveng, P. L. (2003). Stability in the proportion of harbour seals hauled out under locally ideal conditions. Marine Mammal Science, 19(4), 791-805. Small, R. J. G., Pendleton, W. & Pitcher, K. W. (2003). Trends in abundance of Alaska harbour seals, 1983-2001. Marine Mammal Science, 19, 344-362. Stewart, B.S. (1984). Diurnal patterns of harbour seals at San Miguel Island, California. Journal of Wildlife Management, 48, 1459-1461. Stevik, P. T., McConnell, B. J. & Hammond, P. S. (2002). Patterns of movement. In: Marine Mammal Biology, An Evolutionary Approach. Hoelzel, A.R. (Eds). Blackwell Science Ltd.

129 Thompson, P. M. (1987). The effect of seasonal changes in behaviour on the distribution and abundance of common seals (Phoca vitulina), in Orkney, Unpublished PhD thesis, University of Aberdeen. Thompson, P.M. & Rothery, P. (1987). Age and sex differences in the timing of moult in the common seal, Phoca vitulina. Journal of Zoology London, 212, 597-603. Thompson, P.M. (1989). Seasonal changes in the distribution and composition of common seal (Phoca vitulina) haul-out groups. Journal of Zoology London, 217, 281-294. Thompson, P.M., Fedak, M., McConnell, B., & Nicholas, K.S. (1989). Seasonal and sex-related variation in the activity patterns of common seals (Phoca vitulina). Journal of Applied Ecology, 26, 521-535. Thompson, P.M. & Harwood, J. (1990). Methods for estimating the population size of common seals Phoca vitulina. Journal of Applied Ecology, 27, 924-938. Thompson, P.M. & Miller, D. (1990). Summer foraging activity and movements of radio-tagged common seals (Phoca vitulina) in the Moray Firth, Scotland. Journal of Applied Ecology, 27, 492-501. Thompson, P.M., Miller, D., Cooper, R., & Hammond, P.S. (1994). Changes in the distribution and activity of female harbour seals during the breeding season: implications for their lactation strategy and mating patterns. Journal of Animal Ecology, 63, 24-30. Thompson, P.M., Tollit, D.J., Wood, D., Corpe, H.M., Hammond, P.S., & Mackay, A. (1997). Estimating harbour seal abundance and status in an estuarine habitat in north-east Scotland. Journal of Applied Ecology, 34, 43- 52. Thompson, P. M., Van Parijs, S. & Kovacs, K. (2001). Local declines in the abundance of harbour seals: implications for the designation and monitoring of protected areas. Journal of Applied Ecology, 38, 117-125. Venables, U. M. & Venables, L. S. V (1955). Observations on a breeding colony of the seal Phoca vitulina in Shetland. Proceedings of the Royal Zoological Society of London, 125, 521-532. Watts, P. (1992). Thermal constraints on hauling-out by harbour seals (Phoca vitulina). Canadian Journal of Zoology, 70, 553-560.

130 Watts, P. (1996). The diel hauling out cycle of harbour seals in an open marine environment: Correlates and constraints. Journal of Zoology, London, 240, 175-200. Yochem, P.K., Stewart, B.S., Delong, R.L., & DeMaster, D.P. (1987). Diel haul- out patterns and site fidelity of harbour seals (Phoca vitulina richardsi) on San Miguel Island, California in Autumn. Marine Mammal Science, 3, 323-332. Zuur, A. F., Ieno, E. N & Smith, G. M. (2006). Analysis of ecological data. Springer Verlag. 688 pp.

131 CHAPTER 4

Plate 1 Harbour seal with phone tag glued to fur at base of skull

HAUL-OUT BEHAVIOUR OF HARBOUR SEALS IN THE KENMARE RIVER, CO. KERRY.

132 4.1 ABSTRACT

The haul-out behaviour of ten harbour seals in the Kenmare River in southwest Ireland was investigated using a telemetry system based on Global Systems for Mobile Communications (GSM) technology. Statistical modeling techniques were used to examine the influence of covariates tidal level, tidal state, time of day, and month on the haul-out behaviour of tagged seals. The haul-out behaviour of tagged seals varied over the tagging period with animals spending a higher proportion of time ashore post moult in October, decreasing over the winter months to a minimum in February. A strong tidal influence on haul-out behaviour was evident throughout the tagging period, tagged seals hauled out more frequently at low tide. There was variation between tagged seals in the influence of the time of day on their haul-out behaviour. A cyclic pattern with lunar periodicity was evident in the haul-out behaviour of seals tagged in October and the pattern varied between tidal periods. There was overall large variation in the patterns in behaviour over the tagging period (i) between individuals and (ii) between tidal periods for each individual. The large variation in the behaviour between individual seals suggest caution should be exercised when making inferences on the haul-out behaviour of the ‘population’ based on the behaviour of tagged individuals.

133 4.2 INTRODUCTION

The harbour seal (Phoca vitulina L.) is the most widely-distributed pinniped, inhabiting cold-temperate and temperate waters in the northern hemisphere on both sides of the north Atlantic and north Pacific oceans (Bigg, 1981). Harbour seals spend time ashore at terrestrial sites on which they ‘haul-out’ to rest, breed, moult, engage in social activity and escape predation (Pitcher & McAllister, 1981; Boily, 1985; Da Silva & Terhune, 1988; Thompson, 1989; Watts, 1992). The haul-out behaviour of harbour seals has been studied using telemetry (Yochem et al., 1987; Thompson et al., 1989; Thompson & Miller, 1990; Thompson et al., 1997; Rehberg & Small, 2001; Reder et al., 2003; Sharples, 2005), time lapse photography (Stewart, 1984; Thompson & Harwood, 1990) and modelling count data (Adkinson & Small, 2001; Jemison & Pendleton 2001; Simpkins et al., 2003; Boveng et al., 2003; Small et al., 2003; Montgomery, 2005). Their haul-out behaviour has been shown to be influenced by environmental and climatic factors or ‘covariates’ including the time of year, time of day, tidal effects, disturbance and local weather (Stewart, 1984; Yochem et al., 1987; Thompson et al., 1989, 1994; 1997; Thompson & Harwood 1990; Thompson & Miller, 1990; Grellier et al., 1996; Reder et al., 2003)

Estimates of population size are derived from counting the numbers of individuals ashore at haul-out sites, however counts of seals at terrestrial sites can only be considered as minimum population estimates as a fraction of the population will be at sea and unavailable to count. Minimum population estimates, as opposed to abundance estimates, although sufficient for investigating population trends, are inadequate for conservation and management requirements such as the identification of Special Areas of Conservation (SACs) for seals required under the EC Habitats Directive (92/43/EEC) and in determining the predation pressures on fish stocks by a seal population. Information on harbour seals haul-out behaviour has been used to derive a correction factor to account for the missing element of the population during counts and obtain a true abundance estimate (Yochem et al., 1987; Thompson & Harwood, 1990; Thompson et al., 1997; Ries et al., 1998; Huber et al., 2001, Sharples, 2005). Understanding the effects of covariates on seal haul-out behaviour and therefore numbers at haul-out sites helps to enhance the design of surveys and covariates can be factored into the statistical analyses to improve the accuracy of

134 resulting population estimates (Frost et al., 1999; Adkinson et al., 2003; Boveng et al., 2003; Small et al., 2003).

Recent efforts in Ireland have addressed the shortfall in information on harbour seal abundance and distribution at haul-out sites on the Irish coastline, including a national census to establish a minimum population estimate (Cronin et al., 2004) and local studies on the year round changes in the terrestrial distribution and abundance of harbour seals in southwest Ireland (Cronin, 2006, in prep). Information on the haul- out behaviour of individual seals helps explain seasonal changes in numbers of seals at haul-out sites and could potentially be used to derive a correction factor for haul- out counts to estimate the size of the population.

A telemetry system based on Global Systems for Mobile Communications (GSM) technology (McConnell et al., 2004) was used in this study to provide detailed information on the haul-out behaviour of tagged individuals. The mobile phone tag is programmed to send text messages to a base phone, incorporating data on haul-out events, thereby providing information on the amount of time tagged seals spent at sea and ashore. The large global investment in GSM networks have resulted in a telemetry technology with low capital and running costs and even though phone tags do not have the global coverage of Argos satellite system (Fedak et al., 2002), the lower cost allows for larger sample sizes and more powerful inferences to estimate population parameters (McConnell et al., 2004). To acquire basic fundamental data on haul-out behaviour, the mobile phone tags provide a cost-effective approach, overcoming constraints associated with radio and satellite telemetry, which are respectively labour intensive and costly.

Hitherto no information was available on the haul-out behaviour of harbour seals in Ireland. The aims of the study are therefore to (i) determine the activity patterns of harbour seals in the study area and how these change over the annual cycle, (ii) determine what factors affect the haul-out behaviour of tagged seals and (iii) explore the potential of deriving a correction factor to apply to haul-out counts of seals to estimate the total number of seals in the study area.

135 4.3 METHODS

4.3.1 Study site Capture of harbour seals and deployment of tags was attempted in Bantry Bay Co. Cork in October 2004 and the Kenmare River Co. Kerry in southwest Ireland in October 2004 and April 2005. Harbour seals haul-out on rocky islands and skerries generally off the north shores of both bays. Seal capture for tag deployment was attempted at haul-out sites in both bays selected on the basis of haul-out group size, group size greater than 10 and the bathymetry in the vicinity of haul-out sites, water depth less than 3m (figure 1).

Figure 1 Study area showing haul-out sites selected for seal capture and tag deployment

4.3.2 Capturing and handling procedure and tag deployment Tagging was staggered over the interval between moults (between early October and August) as the anticipated length of attachment was approximately four months (Sharples, SMRU pers comm.). The technique employed for catching seals was similar to that described in Jeffries (1993). Two custom made nets, 60m long x 3m

136 deep, with a buoyant headline and lead weighted sink line were used. Deployment was at speed from two rigid inflatable boats (RIBs) which approached the haul-out site from opposite sides and aimed to deploy the net around and as close to the haul- out as possible. Seals entered into the water as soon as deployment commenced and the nets formed a barrier in which they were entangled or trapped. Hoop nets were used to assist in getting the captured animals into the boats. These consist of a 1m diameter hoop made of 20mm plastic hosing and a funnel net of 10mm mesh attached. Captured seals were brought ashore and remained in the hoop nets throughout the administration of the anaesthetic and prior to the tagging procedure.

Seals were weighed and anaesthetised using 0.05ml of Zoletil per 10kg delivered intravenously. If intravenous administration of the anaesthetic was difficult (as with a struggling animal) an intra-muscular dose of 0.1ml of Zoletil per 10kg was delivered. Length and girth of the animal were measured and the sex recorded. The fur was dried with paper towels and degreased using acetone and the tag was secured in place using fast setting epoxy resin (Fedak et al., 1983) at the base of the skull. Seal handling and tagging procedures were carried out under National Parks & Wildlife Service licence no. C18/2005.

4.3.3 Tag operation The tag was designed by the Sea Mammal Research Unit (SMRU), St Andrews University, Scotland and is based on Global Systems for Mobile Communications (GSM) mobile phone technology. Details of the hardware design can be found in McConnell et al (2004). The controlling software is designed to minimize energy consumption while still obtaining a reasonable rate of text messages. The wet/dry sensor in the tag is interrogated every 2.3 seconds. A haul-out event starts when the tag is continuously dry for 10 min and ends when it is continuously wet for 2 min. The start and end times of the haul-out event as well as a unique incremental number are appended to a 160 character long buffer. When the buffer is full a short message service (SMS), also know as a text message, is created and stored in the SIM card. Every four hours the tag ‘wakes’ from sleep mode, waits until it is dry and then attempts to send a text message. To do this it registers with a GSM network for a maximum of 95 seconds. If registration is unsuccessful only diagnostic data for that attempt are appended to the buffer. If registration is successful the delay from dry to

137 registration, the ID code of the radio cell with which it has registered are also appended to the buffer (after McConnell et al., 2004).

4.3.4 Information relay and interpretation Successful registration requires that the phone is within radio contact of a GSM radio cell. The maximum theoretical range is 35km but it is often less than this as a result of obstruction of line of sight or radio interference (McConnell et al., 2004). It is not a prerequisite that the tagged animal remains in the GSM coastal corridor of coverage rather that it returns at some stage to this area resulting in successful registration with the network and subsequent relaying of stored information. Text messages were relayed via the GSM network to a base phone located at the SMRU and details (including haul-out event start/end times, haul-out number, message id, radio cell id, time to register) described in an Access database.

4.3.5 Statistical modelling Statistical modelling techniques were used to examine the influence of covariates tidal level, tidal state, time of day, and month on the haul-out behaviour of tagged seals. The probability of a tagged seal being hauled out was modelled as a function of explanatory variables (or covariates) using generalized additive models (GAM) with logistic link function. These use smoothing curves to model the relationship between the response variable and explanatory variables and allow for non-linear relationship in the data (Hastie & Tibshirani, 1990). A GAM with a binomial distribution and logistic link was used to model the response variable (0 = seal in water, 1 = seal hauled out) as a function of the covariates tidal state, tidal level, time of day and month. Tidal data (tide level, time, speed, direction) for the tagging period at the study site was obtained from tide prediction software Polpred V.2 and an hourly tidal ‘state’ value (-6 to +6) and tidal level (m) assigned to each hour on both the ebb and flood tides. Colinearity was expected between some of the variables, in particular those related to tide and a scatter-plot was used to confirm this.

Cross-validation was applied to determine the optimal degrees of freedom for each smoother. Once the optimal degrees of freedom were determined Pearson and Deviance residuals were plotted versus the original explanatory variables. Any patterns in such a plot may indicate problems with the cross-validation. An

138 assumption in the GAM model is that the errors (εij) are independent, but as the observations are made sequentially over time the independence assumption is violated. Adding a correlation structure on the errors (εij) is a means of dealing with this, such as an auto-regressive error structure AR (1), allowing for auto-correlation between the residuals of sequential hours (Zuur et al., 2006). The length of each data series is given in Table 1. Due to the length of the time series of data that resulted from tags 4, 5, 6, 7, 11 and 20 the auto-regressive error structure in the gamm function in the mgcv library could not be added, the algorithm is halted due to error messages related to lack of computer memory. In the presence of auto-correlation, p-values of smoothers can be seriously inflated (Ostrom, 1990). Instead of incorporating a temporal correlation structure within the gam model to get ‘better’ p-values, it is also possible to use bootstrapping for this (Davison & Hinkley, 1997).

Table 1 Details of tag deployments and measurements of 10 harbour seals captured on the Kenmare River during 2004/2005. Tag/ Date of Location Sex Weight/ Length/ Girth/ Duration Duration Seal tagging kg cm cm tagging/ tagging/ no. days hours 6 16/10/04 Illaunsillagh male 80 156 97 187 4486 5 17/10/04 Illaunslea male 67 127 97 99 2385 4 18/10/04 Illaunslea male 40 112 84 104 2496 3 27/04/05 Sneem male 95 159 125 11 263 10 27/04/05 Sneem male 47 138 83 39 956 9 27/04/05 Sneem male 44 127 87 43 1024 7 28/04/05 Sneem male 87 154 118 92 2213 20 28/04/05 Sneem female 42 120 85 52 1242 11 29/04/05 Sneem male 63 143 103 73 1741 2 29/04/05 Sneem male 103 161 114 29 696

A search for the optimal model was carried out for seals 4, 5, 6, 7, 11 and 20. As the hour smoothers require long time series the length of the data from seals 2, 3, 9 and 10 was too short for inclusion in the models and the gamm function in the mgcv library, with an auto-regressive error structure series was applied.

139 4.3.6 Bootstrap variance estimation Confidence intervals for the haul-out status were obtained using a non-parametric bootstrap approach (Davison & Hinkley, 1997). Bootstrapping involves creating new data sets from the original sample, analysing the new samples in the same way as the original and the distribution of the statistic of interest is estimated from its empirical distribution among the bootstrap sample (Davison & Hinkley, 1997). The following bootstrap approach was applied:

1. The optimal GAM model (in terms of degrees of freedom and explanatory variables) is determined and fitted values and residuals obtained 2. The time series of the residuals is divided into blocks of M days 3. The residuals within each block are permutated 4. The permutated residuals are added to the fitted values in step 1 and new data obtained by rounding to 0 or 1 5. A GAM is applied to the new data obtained in step 4 using the same degrees of freedom for each smoother as in step 1 6. Steps 2 to 5 are repeated 1000 times 7. 95% quartile confidence intervals are obtained by sorting the 1000 bootstrapped values at each point and using observation 25 and 975

In step 2 the time series is divided into blocks the length of which is such that within a block the auto-correlation is captured, but points beyond this length are not auto- correlated. Based on the auto-correlation function blocks of length M=2 or M=3 days were used depending on the tag. The motivation to use the same degrees of freedom in step 5 is motivated by computing time. Allowing the GAM algorithm to determine the optimal degrees of freedom in each bootstrap drastically increases computing time.

The method described above was used for two purposes: (i) to obtain 95% point- wise confidence bands for each smoother, and (ii) to assess the importance of each covariate in the model. The models with, and without a particular covariate were compared using the difference in deviance. This difference was calculated for the original data, but also for each bootstrapped data set. The null-hypothesis is that the covariate has no effect (hence it is removed from the model) so, the null-model is the

140 GAM without the covariate. The alternative hypothesis is that the covariate does have an effect. In this approach, the residuals from the null-model are used for permutation. The data are permutated and the deviance calculated and the process repeated 1000 times. The difference in deviances between the full and null models divided by 1001 represents the p value for the omitted explanatory variable. The entire bootstrapping approach is carried out for each covariate in turn. A more detailed description can be found in Algorithm 7.4 in Davison & Hinkley (1997).

4.4 Results

4.4.1 Capture and tag deployment A total of ten animals were successfully captured at haul-out sites in the Kenmare River and tagged in October 2004 and April 2005. Nine males and one female were captured and weights ranged from 40kg to 103kg (Table 1). Four of the ten seals tagged weighed less than 60kg and based on harbour seal growth curves (Härkönen & Heide-Jorgensen, 1990; Lydersen & Kovacs, 2005) are considered as juveniles.

4.4.2 Duration of transmission Details of tag transmission durations are given in Table 1. The tags transmitted for a period of between 11 and 187 days with an average duration of transmission of 75.5 days, approximately two and a half months. The last transmission during the study period was from seal 7 on August 2nd 2005.

4.4.3 Examination of haul-out data The duration of every haul-out event was calculated in an Access database using the start time and end times of each haul-out event. A sequential haul-out number system allocated to haul-out events and sent in the SMS enables the identification of lost haul-out records. This happens as a result of ‘dropped’ messages which occurred infrequently as a result of the tag failing to send a text message. It is important to identify these as failure to do so would imply that the animal has spent a longer period at sea than is the case.

Haul out records for each seal are shown in figures 2a and 2b. The blue lines indicate haul-out records, the red lines indicate the period between sequential haul-out

141 events and absence of a line indicates that there is no information available for that period. These charts enable the visualisation of the entire haul-out records for each seal over the tagging period. The amount of time the tagged animals spent ashore varied. The average length of a haul-out event was 215 minutes and the longest and shortest haul-out events recorded were 1660 and 10 minutes respectively (10 minutes was the set time programmed in the tag to constitute a haul-out event). 98% of all haul-out events were less than 12 hours. The longest recorded period a tagged individual spent at sea between haul-out events was 12607 minutes which amounts to almost nine days. This individual, seal 6 made seven separate trips to sea of durations of more than four days over the tagging period. This behaviour of spending extended periods at sea was also evident in seal 7 but less than 1% of all trips to sea made by tagged individuals over the tagging period were of duration longer than 4 days. All tagged seals, apart from seal 3, made trips of one day or more, with variation between individuals in the frequency of such trips. However most periods spent at sea by all tagged animals were for hours as opposed to days, between daily haul-out events, with 91% of all trips to sea lasting less than one day. The mean proportion of time tagged seals spent hauled out monthly is shown in figure 3 ranging from 25% to 11% of total time hauled out during October and February respectively. The larger standard error in the value for April, relative to other months, is a result of a small number of haul-out records for that month.

4.4.4 Model outputs and validation An initial analysis was carried out on the data from tag 4 to validate the modeling approach as it was one of the longest time series of data. The haul-out response of seal 4 was modeled as a function of tidal level, tidal state and time of day. Cyclic patterns in the residuals with positive correlations with a time lag of 24 hours suggested the model was missing a fixed term(s). The possibility that the effect of tidal state and/or level was changing during different tidal cycles was explored by differentiating between the first and second rising and falling tides per 24 hour period.

A nominal variable Ts with four levels was created:

1 if observation s is during the rise of tide 1

Ts = 2 if observation s is during the fall of tide 1

3 if observation s is during the rise of tide 2

142

An interaction term between the various tidal periods and explanatory variables tidal level and state was added. Interactions between nominal variables and smoothers can be implemented using the ‘by’ command in the gam function in the mgcv R library.

For example, a model that describes haul-out status as a function of tidal state and level, plus an interaction between these two smoothers and Ts is formulated as:

logit(ps) = α + f (TidalState) * Ts + f(Level) * Ts

Where Status is Binomial distributed: Statuss ~ B (ps,1).

The interaction term between tidal state or level and tidal cycle allows for a different trend in the rise of the first tide, fall of the first tide, rise in the second tide, and fall in the second tide.

The adequacy of the model is explored by examining the residuals and the autocorrelation function values. The residuals represent the difference between the data and the model; the Pearson residual is used in GLM and GAM and comparable to the standardized residuals used for linear models (Faraway, 2006). The autocorrelation function (ACF) looks for correlations between residuals separated by various amounts of time. Specifically its ith value expresses the amount of correlation between pairs of residuals separated by i-1 time points. If no correlation amongst residuals exists then the autocorrelation function values should be random (Gore, 2000).

The model was only a starting point and suggested an immediate problem. It was applied on the data of tag 4 and the ACF of the Pearson residuals showed a clear residual pattern (figure 4). Each time lag represents an hour. The ACF suggested that the status at time s depends on s-1, s-2, but also on s-12, s-24, etc. Hence, the model was missing a component that deals with the daily rhythms.

One option was to extend the model with the following components:

143 α ++f ()TidalState *T33ssf (Level)*T + TimeOfDay TimeOfDay Logit(ps) = βπsin(2)+ βcos(2)π + 1223 23

f ()Hour *T3ss+ Month

The time of the day was converted into a daily cyclic pattern using cos(2*pi*(Timeofday)/23) and sin(2*pi*(Timeofday)/23). To simplify notation, the sin and cosine term are labelled Z1 and Z2 respectively: The term f (Hour) is the long term smoother allowing potential patterns in the haul-out behaviour of the tagged seal over the duration of the tagging period to be identified. For the longer time series, a term Month (nominal) was added to the model.

Logit()p =+α f (TidalState)*T +f (Level)*T +ββZ +Z + ss33s11s21s f (Hour)*T3ss+ Month

As collinearity was confirmed in a scatter-plot of tidal level versus tidal state for seal 4 (figure 13), these variables were not used in the same model. The following two models were therefore applied to the seal 4 data and compared to identify the optimal model:

Model 1: Logit()ps =+α f (Level)*T31ss+ββZ1+2Z1s+f (Hour)*T3s+Months

Model 2: Logit()ps =+α f (TidalState)*T31ss+ββZ1+2Z1s+f (Hour)*T3s+Months

The Akaike Information Criteria (AIC) values for the two models were compared. The AIC enables a measure of the goodness of fit and also employs a penalty for the number of parameters in the model (Zuur et al., 2006) and the model with the smallest AIC chosen as the most optimal. Examination of residual plots confirmed model 1 as the optimal model for tag 4:

Logit()ps =+α f (Level)*T31ss+ββZ1+2Z1s+f (Hour)*T3s+Months

This model still contained a residual auto-correlation structure. The AR-1 structure in the residuals is mainly from the tidal period 1 residuals. Hence, processes are in place

144 for this time frame that are not explained well within the current model. Bootstrapping techniques were therefore required to assess the true significance of the smoothers.

The smoothers for hour on the haul-out status of seal 4 over the entire tagging period in four tidal periods are shown in figure 5. The coloured lines represent the long term smoothers for the rising and falling parts of tide 1 and tide 2, the black dots denote the full moon and the following patterns are evident: (i) a tidal pattern, a high probability of the seal being at sea during the rising first or second daily tide but never during both, (ii) a long term pattern, higher probability of the seal being at sea more- so on the first rising tide for approximately the first 1500 hours and more-so on the second rising ride from then onwards and (iii) a pattern with lunar periodicity, the highest probability of being at sea occurred just after the full moon each month and this occurred on the first rising tide, additionally the highest probability of being hauled out on the second rising tide occurred at this time.

The probability of haul-out at different tidal levels during the four tidal periods and during the first 400 hours of the tagging duration is shown in figure 6. The highest probability of being hauled-out occurs during low tidal level or height (i.e. around low tide) of all tidal periods apart from the second rising tide, suggesting the seal spent more time at sea during the second rising tide (and at all levels of this tide) during the first 400 hours. The hourly pattern of haul-out probability for the different tidal periods during this time is shown in figure 7. This pattern changes later in the tagging period when highest probability of haul-out occurs during the second rising tide and lowest probability during all levels of the rising tide (figure 8).

The optimal GAMs for tags 5, 6, 7, 11 and 20 were variations on the optimal model for tag 4: Logit()ps =+α f (Level)*T31ss+ββZ1+2Z1s+f (Hour)*T3s+Months The difference in the optimal model for each tag was associated with the nominal

variable Ts which represents the four daily tidal periods. For some tags a level and/or month smoother combining some or all of the four tidal periods was optimal.

The smoothers for hour per tidal period from GAMs of the haul-out data from the tags on seals 5 and 6 are shown in figures 9 and 10. The coloured lines represent the long term smoothers for the rising and falling parts of tide 1 and tide 2, the black dots

145 denote the full moon. The patterns in the haul-out behaviour of seal 5 are similar to those of seal 4. The tidal pattern, a high probability of the seal being at sea during the rising first or second daily tide but never during both; the long term pattern, higher probability of the seal being at sea more-so on the first rising tide for approximately the first 1500 hours and more-so on the second rising ride from then onwards; and a lunar pattern, the highest probability of being at sea occurred just after the full moon each month and this occurred on the first rising tide with the exception being in the third month when the highest probability of being at sea occurred during the falling part of tide 1 following a full moon. Although a cyclic pattern is also seen in the haul- out behaviour of seal 6 it differs to that from seals 4 and 5. The alternating probability of haul-out between the first rising and falling tides seen in the data from the tags on the latter two seals is not obvious in the data from the tag on seal 6. A possible lunar influence on haul-out behaviour is suggested by highest probabilities of being either at sea or hauled out occurring after a full moon during all tidal periods and not just during the first rising tide as with seals 4 and 5.

The cyclic patterns evident in the haul-out behaviour of seals 4, 5 and 6 throughout the tagging period were not obvious in those seals tagged in a different period, seals 7, 11 and 20. Seal 7 showed a higher probability of being at sea after the first full moon of its tagging period during all tidal periods however there was no cyclic pattern evident and after approximately the first 1000 hours of the tagging period the probability of this seal being at sea gradually increased (figure 11). Seal 11 showed a similar pattern to seal 7 with a possible lunar influence during the first 1000 hours of tagging followed by a gradual increase in the probability of being at sea; additionally the tidal patterns seen in the behaviour of seals 4 and 5, a high probability of the seal being at sea during the rising first or second daily tide but never during both is also evident in the early stage of this seals tagging period (figure 12). Seal 20 showed a gradual increase in haul-out probability over its tagging period (figure 13).

The patterns in hour smoothers for the different tidal periods for tags attached in the latter part of the tagging period i.e. during the spring and early summer do not show the cyclic patterns evident in the tags attached at the beginning of the tagging period in the autumn. Apart from the similarity in the long term pattern in haul-out behaviour between seals 4 and 5 there appears to be otherwise large variation in these long term

146 patterns in behaviour (i) between individuals and (ii) between tidal periods for each individual and (iii) over the tagging period.

Bootstrapping provided a means of dealing with the problems of autocorrelation and figures 14-19 show the bootstrapped hour and level smoothers with confidence bands for individual seals. The smoothing functions describing the partial effect of level on haul-out behaviour suggest all tagged seals had a higher probability of hauling out at low tide and this decreased with rising tide. The confidence bands resulting from bootstrapping the ‘hour’ data from the tag on seal 6 were slightly larger. Data from the other tags suggested differences in the long-term patterns in the haul-out behaviour of tagged seals between the autumn/winter and the spring/summer. The tag on seal 6 remained attached for a longer period than other tags and included data on haul-out patterns that potentially differed across the seasons. When only the first 2400 hrs of data from tag 6 were bootstrapped, ensuring that the residuals from the data from this period were not fitted to data from later in the tagging period, the confidence bands around the hour smoothers improved considerably. The confidence bands around the smoothing functions describing the effect of both level and hour on the haul-out behaviour of seal with tag 7 are wider (figure 17) suggesting additional factors other than those included in the model are influencing the haul-out behaviour of this animal.

The p values for the explanatory variables included in the GAMs of data from seals 4, 5, 6, 7, 11 and 20 are given in table II. The p values for explanatory variables level and time of day resulting from GAMMs applied to the data from the tags with shorter time series of data, on seals 2, 3, 9 and 10, are shown in table III. There was a significant effect of tidal level on the haul-out behaviour of all tagged seals during all tidal cycles throughout the tagging period (p<0.001 to p<0.01). The ‘hour’ smoother depicts the long term pattern in haul-out behaviour over the tagging period. A significant change in haul-out patterns over the tagging period was evident in seals 4, 6, 11 and 20 (p<0.001 to p<0.05). In the latter two this was evident over all tidal periods. Seals 4 and 6 showed significant change in their haul-out pattern over the tagging period during tidal periods other than the second falling and rising tides respectively. Seals 5 and 7 did not significantly change their haul-out behaviour over the tagging period. The long-term patterns in the haul-out behaviour of seals 2, 3, 9

147 and 10 could not be explored as the tagging periods were too short. A significant change in the haul-out behaviour of tagged seals between months was apparent in seal 6 only (P<0.05). The time of day had a significant effect on the haul-out behaviour of all seals (P<0.001 to P<0.05) apart from seals 4 and 7. The phi (φ) values in table III denote the correlation between sequential hours. The AR(1) structure in the GAMM dealt with the auto-correlation problem as no patterns were obvious in the auto- correlation function of the Pearsons residuals.

148 Tagi2 i3Tag

01-Aug 01-Aug

01-Jul 01-Jul

01-Jun 01-Jun

01-May 01-May

01-Apr 01-Apr

00:00 04:00 08:00 12:00 16:00 20:00 00:00 00:00 04:00 08:00 12:00 16:00 20:00 00:00

Tagi4 i5Tag

01-Jul 01-Jul

01-May 01-May

01-Mar 01-Mar

01-01-Jan 01-Jan

01-Nov 01-Nov

00:00 04:00 08:00 12:00 16:00 20:00 00:00 00:00 04:00 08:00 12:00 16:00 20:00 00:00

Tag Tagi6 i7 01-Aug

01-Jul 01-Jul 01-Jun 01-May 01-May

01-Apr

01-Mar

01-01-Jan

01-Nov

00:00 04:00 08:00 12:00 16:00 20:00 00:00 00:00 04:00 08:00 12:00 16:00 20:00 00:00 Figure 2a. Haul-out records from tagged seals numbers 2-7 over the tagging period. The blue line indicates haul-out periods, the red line indicates periods when the seals were known to be not hauled out. White space indicates periods from which there is no information.

149

Tagi9 9 i1Tag0 10 01-Aug 01-Aug

01-Jul 01-Jul

01-Jun 01-Jun

01-May 01-May 01-Apr 01-Apr

00:00 04:00 08:00 12:00 16:00 20:00 00:00 00:00 04:00 08:00 12:00 16:00 20:00 00:00

Tagi1 111 Tagi20 20

01-Aug 01-Aug

01-Jul 01-Jul

01-Jun 01-Jun 01-May 01-May 01-Apr 01-Apr

00:00 04:00 08:00 12:00 16:00 20:00 00:00 00:00 04:00 08:00 12:00 16:00 20:00 00:00

Figure 2b. Haul-out records from tagged seals numbers 9,10,11,20 over the tagging period. The blue line indicates haul-out periods, the red line indicates periods when the seals were known to be not hauled out. White space indicates periods from which there is no information.

150

0.3

0.25 t ou d e ul 0.2 t ha n e p s me i 0.15 t f on o i t por

o 0.1 pr n a e M 0.05 4 6 45 90 93 87 28 31 35 18 11 34 3 D= 3 D= 3 D= 3 D= 1 D= 1 D= 8 D= 7 D= 5 D= 2 D=

0 N= N= N= N= N= N= N= N= N= N= October November December January February March April May June July month Figure 3. Mean proportion of time tagged seals spent ashore ((±1 s.e) (N= number of tagged seals D= number of days of haul-out data)

Auto-correlation .0 1 8 . 0 .6 0 4 . ACF 0 .2 0 0 . 0

0 10203040506070

Lag

Figure 4. Auto-correlation function of the Pearsons residuals resulting from GAM on haul-out data from tag 4

151 Table II. Summary of optimum generalized additive models of haul-out status of tagged seals within the Kenmare River. For all explanatory variables, the associated bootstrapped probability value (p) is given; explanatory variables level and hour were fitted as separate smoothers for four tidal periods (TP1-TP4) unless otherwise stated. The estimated degrees of freedom (edf) are shown for variables fitted as smoothers and for parametric terms the degrees of freedom (df) shown

Smoother Tag 4 5 6 7 11 20 Level TP 1 edf 1.005 1.776 1.887 (TP1-4) 2.596 (TP1-4) 1.001 (TP1-4) p value p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 Level TP 2 edf 1.005 (TP2 & 3) 1.906 1.007 p value p<0.001 p<0.001 p<0.001 Level TP 3 edf 1.987 1.925 p value p<0.01 p<0.001 Level TP 4 edf 4.405 3.024 1.008 p value p<0.001 p<0.001 p<0.01 Hour TP 1 edf 9.000 8.625 8.939 7.193 5.787 (TP1-4) 8.681 (TP1 & 2) p value p<0.05 p=0.122 p<0.001 p=0.890 p<0.05 p<0.001 Hour TP 2 edf 7.638 1.000 8.822 8.205 p value p=0.075 p=0.313 p<0.05 p=0.525 Hour TP 3 edf 9.000 8.481 8.866 6.787 7.200 p value p<0.05 p=0.244 p=0.158 p=0.811 p<0.001 Hour TP 4 edf 1.001 1.508 7.229 0.991 1.909 p value p=0.259 p=0.237 p<0.05 p=0.633 p<0.01` Month d.f 4.000 6.000 4.000 3.000 p value p=0.355 p<0.05 p=0.890 p=0.063 Cosine Time of day d.f 1.000 1.000 1.000 1.843 1.000 1.000 p value p=0.237 p<0.001 p<0.001 p=0.207 p<0.001 p<0.001 Sine Time of day d.f 1.000 1.000 1.000 1.000 1.000 1.000 p value p=0.369 p=0.976 p=0.086 p=0.177 p=0.081 p<0.001

152 Table III. Summary of optimum generalized additive mixed models of haul-out status of tagged seals within the Kenmare River. Probability values of all explanatory variables, estimated degrees of freedoms (edf) for smoothers and degrees of freedom (df) for nominal variables and Phi (φ) values are given. Explanatory Variable Tag 2 3 9 10 Level edf 1.000 1.000 1.000 2.689 p value p<0.001 p<0.01 p<0.001 p<0.001 Cosine Time of day df 1.000 1.000 1.000 1.000 p value p<0.05 p<0.001 p<0.01 p<0.001 Sine Time of day df 1.000 1.000 1.000 1.000 p value p<0.05 p=0.233 p=0.617 p<0.05 Phi (φ) value 0.7519 0.5809 0.6438 0.7488

153 Hour effect

3

s es e u 2 l u l a a v v

d HT 1 1 LT 1

e t fitted t i HT 2 f e

o

0 t th

o n t

o LT 2 i

t u -1 tion b u i r b t ri n t o n -2 C Co

-3 0 500 1000 1500 2000 2500

TimTimee/hours

Figure 5. Smoothing functions describing the partial effect of ‘Hour’ on the haul-out status of seal with tag 4 over the entire tagging period in four tidal periods. The black line, labelled LT1, is the long term smoother for the rising part of tide 1, the green line LT2 the rising part of tide 2, the red line HT1 the falling part of tide 1 and blue line HT2 the falling part of tide 2. The black dots denote full moon.

Probabilities first 400 hours

0 . 1

8 0.

t -ou 6 l . u 0

a h O O of 4 0. ility ab

2 Prob . 0

0 0.

0.51.01.52.02.53.0

Tide level/m LL

Figure 6. Probability of seal with tag 4 being hauled out during the first 400 hours of the tagging period for different tidal levels (0=not hauled out, 1= hauled-out; colours represent tidal period: black = rise in tide 1, blue=fall in tide 1, green = rise in tide 2 red=fall in tide 2)

154 Probability

0 .

1

8

0. t

-ou l 6 u . a 0 h

O O of 4 0. ility

ab 2 . 0 Prob 0

0. 0 100 200 300 400 Time/hou HH Figure 7. Probability of seal with tag 4 being hauled out during the first 400 hours of the tagging period; colours present tidal period (0=not hauled out, 1= hauled-out; black = rise in tide 1, blue=fall in tide 1, green = rise in tide 2, red=fall in tide 2)

Probabilities between 1200 and 1400 hours

0 .

81 .

0

t 6 . -ou l

u a O h O 40 of . 0

ility ab 2

. Prob

00 . 0

0.51.01.52.02.53.03.5

Tide level/m LL

Figure 8. Probability of seal with tag 4 being hauled out during the hours of 1200 and 1400 of the tagging period for different tidal levels (0=not hauled out, 1= hauled-out; black = rise in tide 1, blue=fall in tide 1, green = rise in tide 2, red=fall in tide 2)

155 Hour effect 3

HT 1

2

LT 1 HT 2 ues l es

1 LT 2 a u l a ed v t v t i f

o 0 fitted on t i e but th i r o 1 - t

ont C tion u

b -2 ri t n Co

-3

0 500 1000 1500 2000 2500 Time Time/hours Figure 9. Smoothing functions describing the partial effect of ‘Hour’ on the haul-out status of seal with tag 5 over the entire tagging period in four tidal periods. The black line, labelled LT1, is the long term smoother for the rising part of tide 1, the green line LT2 the rising part of tide 2, the red line HT1 the falling part of tide 1 and blue line HT2 the falling part of tide 2. The black dots denote full moon.

Hour effect

3

2 es u l a ues v

1 LT 1 val ed t fitted t HT 1 i

f HT 2 e o 0 t th LT 2 o on i t t u b i r tion -1 nt u o b

C ri t n -2 Co

3 - 0 1000 2000 3000 4000 Time Time/hours

Figure 10. Smoothing functions describing the partial effect of ‘Hour’ on the haul-out status of seal with tag 6 over the entire tagging period in four tidal periods. The black line, labelled LT1, is the long term smoother for the rising part of tide 1, the green line LT2 the rising part of tide 2, the red line HT1 the falling part of tide 1 and blue line HT2 the falling part of tide 2. The black dots denote full moon.

156 H ou r effect

3

2 es u l s a e u l v 1 a v d e t t fitted fi e

0 to HT 2 th n HT 1 o LT 1 o

ti LT 2 t u b i r t 1 - n tion o u

C b ri t n

-2 Co

-3 5000 5500 6000 6500 7000

Time Time/Ho

Figure 11. Smoothing functions describing the partial effect of ‘Hour’ on the haul-out status of seal with tag 7 over the entire tagging period in four tidal periods. The black line, labelled LT1, is the long term smoother for the rising part of tide 1, the green line LT2 the rising part of tide 2, the red line HT1 the falling part of tide 1 and blue line HT2 the falling part of tide 2. The black dots denote full moon.

Hour effect

3

2 es

u l a s v e

u l d 1

a e t v t i d f

tte

e

fi HT 2

h t 0

to o

n HT 1 t o LT 1

i t n u o b i i LT 2 r t t u -1

n o b i C r t

n Co -2 3

-

5000 5500 6000 6500 Time Time/Ho Figure 12. Smoothing functions describing the partial effect of ‘Hour’ on the haul-out status of seal with tag 11 over the entire tagging period in four tidal periods. The black line, labelled LT1, is the long term smoother for the rising part of tide 1, the green line LT2 the rising part of tide 2, the red line HT1 the falling part of tide 1 and blue line HT2 the falling part of tide 2. The black dots denote full moon.

157

3

2 es u l a 1 v fitted 0

e LT 2

th o t -1

HT 1 LT & HT 1 tion u b -2

ri t n

Co -3

4500 5000 5500 6000 Time/Hours Figure 13. Smoothing functions describing the partial effect of ‘Hour’ on the haul- out status of seal with tag 20 over the entire tagging period in four tidal periods. The red line, labelled LT and HT 1, is the long term smoother for the rising and falling parts of tide 1, the green line the rising part of tide 2, and blue line the falling part of tide 2. The black dots denote full moon.

158

Level TP1 Level TP2 & 3 Level TP4 4 4 4

2 2 2 2

0 0 0 0

-2

2 -2 -2

-4 4 -4 -4

0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Tidal level/m Level 1 Level 2 & 3 Level 4

4 4 4 4 Hour TP1 Hour TP2 Hour TP3 Hour TP4

2 fitted

2 2 2 2 e

th 0

o

t 0 0 0 0

tion -2 u 2 2 2 2

- - - b ri

t n -4 4 -4 -4 -4 Co 0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500 Hour 1 Hour 2 Hour 3 HourHours 4

Figure 14. Smoothing functions with 95% confidence intervals obtained by bootstrapping describing the partial effect of ‘Hour’ and ‘Level ‘on the haul-out status of seal with tag 4 over the entire tagging period in the tidal periods 1-4.

4 4

Level TP1 4 Level TP2 Level TP3 Level TP4 4

2 2 2 2 2

0 0 0 0 0

-2

2 -2 -2 -2

-4 4 -4 -4 -4

4

es 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.51.01.52.02.53.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 u l Tidal level/m a Level 1 Level 2 Level 3 Level 4 v

4 4 4 4 Hour TP1 Hour TP2 Hour TP3 Hour TP4 fitted

e 2 2 2 2 2 th

o

t 0 0 0 0 0 tion

u -2 b 2

-2 -2 -2 ri t

n 4 Co -4 -4 -4 -4

0 50 0 1000 1500 2000 0 500 1000 1500 2000 0 500 1000 1500 2000 0 500 1000 1500 2000 Hour 1 Hour 2 Hour 3 Hour 4 Hours Figure 15. Smoothing functions with 95% confidence intervals obtained by bootstrapping describing the partial effect of ‘Hour’ and ‘Level ‘on the haul-out status of seal with tag 5 over the entire tagg ing period in the tidal periods 1-4. 159 2 2 Level TP2 Level TP3 1 1

0 0 -1 -1

-2 -2

0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Level 2 Level 3

2 Level TP4 1

0

-1

-2

0.5 1.0 1.5 2.0 2.5 3.0 3.5 Tidal level/m Level 4

Hour TP1 Hour TP2 68 68

024 024 2 2 - - 4 4 - -

0 500 1000 1500 2000 0 500 1000 1500 2000

Hour 1 Hour 2 es lu

a v

Hour TP3 Hour TP4

fitted o

t 2468 2468 tion 0 0 u b ri 2 2 t - - n 4 4 - - Co 0 500 1000 1500 2000 0 500 1000 1500 2000 Hour 3 Hour 4 Hours

Figure 16. Smoothing functions with 95% confidence intervals obtained by bootstrapping describing the partial effect of ‘Hour’ and ‘Level ‘on the haul-out status of seal with tag 6 over the first 2400 hours of the tagging period in the tidal periods denoted 1-4.

160

4 Level TP1-

2 0 -2

-4

0.5 1.0 1.5 2.0 2.5 3.0 3.5

Level Tidal

4 4 Hour TP1 4 Hour TP2

2

2 2

0 0 0

-2 2 -2

4 - 4 -4

5000 5500 6000 6500 7000 5000 5500 6000 6500 7000

es Hour 1 Hour 2 lu

a v

Hour TP3 Hour TP4 4 4

2

2 2 fitted

o

t 0 0 0

-2 tion 2 2 u -

b ri -4 t 4 -4 n

Co 5000 5500 6000 6500 7000 5000 5500 6000 6500 7000

Hour 3 Hour 4 Hours

Figure 17. Smoothing functions with 95% confidence intervals obtained by bootstrapping describing the partial effect of ‘Hour’ and ‘Level ‘on the haul-out status of seal with tag 7 over the entire tagging period in the tidal periods 1-4.

161

4 Level TP1-

es -4 --2 0 2 lu a v

0.5 1.0 1.5 2.0 2.5 3.0 fitted o Level Tidal

t tion 4 u b ri t n Hour TP1-4 Co

-4 --2 0 2

5000 5500 6000 6500

Figure 18. SmoothingHou r functions1 with 95% confidence intervals obtained by bootstrapping describing the partial effect of ‘Hour’ and ‘Level ‘on the haul-out status of seal with tag 11 over the entire tagging period in the tidal periods 1-4.

Hours

4 4 Level TP1-4 Hour TP1&2 2 2 0 0

2 2 - -

-4 -4

es 0.5 1.0 1.5 2.0 2.5 3.0 4600 4800 5000 5200 5400 5600 5800 6000 lu a Level Tidal Hour 1 & 2 Hours v

Hour TP3 Hour TP4 fitted o

t tion 024 024 u b ri t 2 2 - - n Co 4 4 - -

4800 5000 5200 5400 5600 5800 6000 4800 5000 5200 5400 5600 5800 6000 Hour 3 Hour 4 Hours

Figure 19. Smoothing functions with 95% confiden ce intervals obtained by bootstrapping describing the partial effect of ‘Hour’ and ‘Level ‘on the haul-out status of seal with tag 20 over the entire tagging period in the tidal periods 1-4.

162 4.5 DISCUSSION

4.5.1 Effects of time of day and tidal cycle on haul-out behaviour The effect of the tidal cycle and time of day on the haul-out behaviour of harbour seals has been well studied across their geographical range using telemetry (Thompson & Miller, 1990; Thompson et al., 1989; Thompson et al., 1997; Yochem et al., 1987; Rehberg & Small, 2001; Reder et al., 2003; Sharples, 2005), time lapse photography (Stewart 1984; Thompson & Harwood, 1990,) and modelling count data (Adkinson & Small, 2001; Jemison & Pendleton 2001; Boveng et al., 2003; Simpkins et al., 2003; Small et al., 2003; Montgomery, 2005). Prior to this study no information was available on the effects of the time of day and the tidal cycle on the haul-out behaviour of harbour seals in Ireland.

A significant tidal influence on the haul-out behaviour of tagged seals was evident throughout the study period, tagged seals hauled out more frequently at low tide. Haul-out sites in the study area are generally tidally influenced rocky skerries, with haul-out habitat submerged at high tide. Tagged seals spent less time ashore on a rising tide than a falling tide, possibly responding to local increases in food availability on the incoming tide. The time of day had a significant influence on the haul-out behaviour of all tagged seals other than seals 4 and 7, with variation between individuals’ daily haul-out patterns. A distinct diurnal pattern in haul-out behaviour was evident from six of the tagged seals, spending more time ashore during early to mid afternoon. Seal 5 appeared to have a preference for hauling out at night while seal 4 displayed a bimodal pattern, hauling out at night as well as mid-afternoon. There were no diurnal patterns evident in the haul-out behaviour of seals 2 and 7. Overall, tagged seals spent more time ashore during the day than at night, possibly returning to the water to feed at night when foraging may be more profitable (Croxall et al., 1985). Similar patterns have been observed in other studies of harbour seal haul- out behaviour (Bouvla & McLaren, 1979; Thompson et al., 1989; Thompson & Miller, 1990; Watts, 1993).

163 The two seals that showed no diurnal patterns in haul-out behaviour in the present study were the only adult males tagged prior to and during the breeding season. It is possible that these were breeding males spending more time in the water prior to and during the mating period, increasing their chances of intercepting females. As these animals were tagged in late April it is not possible to say if this was a mating related shift in haul-out behaviour. A staggered approach to tagging across the annual cycle resulted in a small sample size. This together with a male sample bias and the fact the tagged animals were not aged, makes it difficult to identify possible sex and age related differences in haul-out behaviour. Considering the evidence for heterogeneity in harbour seal haul-out behaviour among different population segments (Thompson & Rothery, 1987; Härkönen et al., 1999; Daniel et al., 2003; Reder et al., 2003), it is unlikely that limited numbers of study animals of unknown ages can be used to describe the population (Härkönen et al., 1999). Considering the difficulty in catching seals in the study area it is unfeasible to obtain a representative sample by age and sex of the population; increasing the sample size of tagged animals in the study area over time will however provide a less biased sample. Haul-out bouts have been shown to be correlated with the tidal cycle in an estuarine environment in Scotland (Thompson & Miller 1990; Thompson et al., 1994; Thompson et al., 1997). In contrast, haul-out bouts of more than 24 hours have been recorded in other areas where haul-out sites are available throughout the tidal cycle (Yochem et al., 1987). Wilson (1978) suggested that in areas where habitat for hauling out is available above the high water level, diurnal cycles may be more influential than tidal cycles on haul-out behaviour. Haul-out sites in the study area are generally tidally influenced rocky skerries, however some habitat is available even at high tide, evident from one haul-out record of over 27 hours.

4.5.2 Seasonal changes in haul-out behaviour Most studies on harbour seal haul-out behaviour have focused on breeding and moult periods mainly for the purpose of improving the accuracy of population census (Stewart & Yochem, 1983; Yochem et al., 1987; Thompson & Harwood, 1990; Thompson & Miller, 1990; Thompson et al., 1997; Ries et al., 1998;

164 Huber et al., 2001; Simpkins et al, 2003; Reder et al., 2003) and apart from some exceptions (Thompson et al., 1989; Rehberg & Small, 2003; Sharples, 2005), there is limited information available on the seasonal change in the haul-out behaviour of harbour seals.

The haul-out behaviour of tagged seals in this study varied over the tagging period with animals spending a higher proportion of time ashore post moult in October, decreasing over the winter months to a minimum in February, increasing until April and remaining relatively constant through the proceeding months until July. Changes in the haul-out behaviour of the tagged seals are reflected in the seasonal change in abundance of seals at haul-out sites in the study area (Cronin, 2006 in prep). The behaviour of individuals tagged over the winter months differed from the behaviour of those tagged during spring and early summer. Unfortunately because the tags transmitted for only two and a half months on average, such information was obtained from two separate samples of tagged seals, the first tagged in October and the second in April and the potential seasonal change in the haul-out behaviour of individuals could not be fully explored. A significant change in the haul-out behaviour of tagged seals between months was apparent in seal 6 only; the tag on this animal transmitted from mid October to late April providing the longest data series from any tag and suggests a possible seasonal related change in behaviour.

Studies of the seasonal variation in body condition of harbour seals have shown that they are at their fattest during winter (Drescher, 1979; Pitcher, 1986). We have no information on the offshore behaviour or movement patterns of the tagged animals in the present study and as a result cannot be certain that the longer periods spent at sea in winter represent successful foraging strategy. Winter activity patterns of radio tagged seals around Orkney in Scotland suggested that they spend less time in inshore waters at this time of year (Thompson et al., 1989); data from harbour seals tagged with satellite relay data loggers in St. Andrews Bay suggests that the proportion of time tagged seals spent near the haul-out increased steadily from winter through to summer and the probability of being hauled out is much lower in winter months (Sharples, 2005). Harbour seal pups tagged with satellite linked time depth recorders in Alaska

165 rapidly increased the proportion of time spent at sea, from deployment in August, remaining constant through until February to April when a slight decrease was seen (Rehberg & Small, 2003).

There were cyclic patterns apparent in the haul-out behaviour of the seals tagged over the autumn/winter period. The two tagged juvenile males, seals 4 and 5, in particular had very similar patterns between October and January. These seals showed a higher probability of being at sea during the rising first or second daily tide but never during both and a higher probability of being at sea more-so on the first rising daily tide for approximately the first 60 days and more-so on the second rising daily tide from then onwards. There appeared also to be a lunar influence on the seals activity and for both seals the highest probability of being at sea occurred just after the full moon each month and this occurred on the first rising tide only. A possible lunar influence on the haul-out behaviour of the third seal tagged during this period, seal 6, an adult male, was suggested by highest probabilities of being either at sea or hauled out evident after a full moon during all tidal periods and not just during the first rising tide as was the case with the younger seals.

Lunar cycles have been reported in many marine animals. Patterns of movement, feeding and reproduction in inter-tidal organisms are closely associated with the tidal regime and therefore the lunar cycle (Christy, 1986; Berry, 1986). Many tropical fish species have lunar reproductive cycles and most hypotheses concern the dispersal of planktonic eggs or larvae (Robertson et al., 1990). Lunar patterns in zooplankton density observed in southeast Africa was shown to be induced by predation by sardines that crop zooplankton more efficiently on nights when the full or nearly full moon rises after sunset. Zooplankton approaching the surface during darkness, become vulnerable in the light of the rising moon and a sudden decrease in density occurs (Gliwicz, 1986). Diel vertical migration by zooplankton is a common feature in marine and freshwater environments and predators at higher trophic levels may modify their behaviour to optimize the exploitation of vertically migrating prey (Hays, 2003). Nocturnal feeding has been reported in harbour seals (Boulva & McLaren, 1979; Thompson et al., 1989; Thompson & Miller, 1990) and it has been suggested that

166 seals feed nocturnally in response to changes in the vertical distribution or schooling behaviour of their prey (Croxall et al., 1985; Thompson et al., 1989).

The possible lunar cycle in the behaviour of the seals tagged post-moult observed in the present study may be linked to a lunar periodicity in the food chain; fish abundance in surface layers may increase responding to enhanced foraging opportunities of zooplankton near the surface as a result of the light from the moon. Local increases in food availability on the incoming tide may explain predilection shown by the tagged juvenile seals to go to sea on the first rising tide following a full moon.

Tidal cycles are influenced by the gravitational forces of the moon and the sun and ‘spring tides’, particularly strong tides, occur during the full moon and the new moon each month. At intertidal haul-out sites more habitat will be available for seals to haul out on during spring tides and a lunar cycle in haul-out behaviour related to haul-out habitat availability may be expected. However the periodicity of the cycle evident in the haul-out behaviour of the tagged seals early in the tagging period was on a monthly basis and if it was habitat availability driven a bi-monthly pattern of increased haul-out probability on spring tides would be expected. The difference in the haul-out patterns of seals between tidal periods identified, such as the preference shown by the two juvenile seals for going to sea on the first rising tide following a full moon, further suggests that the observed cycles in the haul-out behaviour of these seals are possibly driven by enhanced foraging opportunities as opposed to habitat availability. The latter may however be a factor in the cyclic patterns in behaviour observed in the adult male seal; a potential lunar influenced cycle of increased probability of being at sea was apparent, as seen in the juveniles, however not every month and during some months there was instead an increased probability of hauling out around the time of the full moon. This seal spent several extended periods at sea during the tagging period followed by long haul-out events, possibly taking advantage of increased availability of haul-out habitat exposed during spring tides. Moreover, the extended trips to sea were possibly further offshore and covariates other than those that were included in the model (the covariates that influence haul-out behaviour such as the tidal cycle

167 and the time of day) are likely to be influential when the seal is not in the vicinity of a haul-out site for extended periods of time.

The haul-out behaviour of seals tagged in the latter part of the tagging period i.e. during the spring and early summer do not show the cyclic patterns in behaviour shown by seals tagged in the autumn. What may be the end of a cyclic pattern was evident in the behaviour of seals 7 and 11 at the beginning of the second tagging period late April/early May, after which their probability of being at sea increased. Both of these seals were adult males and the patterns observed may be due to breeding season associated changes in their haul-out behaviour. Studies of haul-out patterns of radio-tagged harbour seals in Svalbard, Norway suggest that males adjust their haul-out behaviour to follow female distribution and movement patterns during the breeding period (Reder et al., 2003). Copulation takes place in the water (Van Parijs et al., 1997) and older adult males have been shown to spend more time at sea during summer (Härkönen et al., 1999); evidence from acoustic surveys suggests that breeding males restrict their range to areas where they are most likely to intercept females, such as females foraging grounds, near haul-out sites and in between both areas (Van Parijs et al., 1997, 1999, 2000). The only female tagged in the present study, seal 20, did not show a cyclic pattern in haul-out activity, rather a gradual increase in her probability of hauling out from late April until mid June when the tag stopped transmitting. This seal was a juvenile and probably immature. Sexual maturity of female and male harbour seals is reached between the ages of three to six years and four to six respectively (Bigg, 1969; Härkönen & Heide-Jorgensen, 1990; Lydersen & Kovacs, 2005). Immature seals have been shown to haul-out more frequently during the summer compared with older seals (Thompson & Rothery, 1987; Härkönen et al., 1999). Haul-out behaviour has been found to vary with age, sex and locality (Thompson, 1989, Thompson et al., 1989, Härkönen et al., 1999, Frost et al., 2001; Reder et al., 2003). The small sample size of tagged seals precludes the possibility of determining if the cyclic patterns observed in the present study is a common phenomenon in harbour seal behaviour, at least in the study area, and from exploring any potential age and/or sex related differences in such a pattern.

168 Exploring the possibility of seasonal change in the cyclic patterns in the haul- out activity of the tagged seals was only possible with data from six seals as the tagging periods of four of the seals were too short. This could not be explored fully unfortunately as the tagging of seals was ‘staggered’ over the entire study period. Considering the data from the six seals, three of which were tagged in October and April respectively, it appears that there might be a lunar influenced pattern in foraging and haul-out behaviour between October and May. It is suggested that in the months following the annual moult harbour seals haul-out behaviour may be predominantly influenced by their foraging activity, following which haul-out patterns are perhaps driven more by breeding related behaviour, and that this foraging activity could be influenced by lunar cycles.

Prey-predator interactions may have been responsible for selecting for intrinsic monthly rhythms in behaviour and physiology of animals with long life spans (Gliwicz, 1986). If prey species change their behaviour in response to the lunar cycle then seals foraging behaviour and haul-out behaviour may change accordingly. Seasonal changes in the availability of prey species may affect this pattern. Moreover as the diet of the harbour seal can vary considerably between areas (Brown & Mate, 1983; Härkönen, 1987) it is suggested that the influence of the lunar cycle on prey behaviour and the subsequent effects on seal behaviour may vary both temporally and spatially. Heretofore there is no information on the diet of the harbour seal in the study area. Such information together with data on prey availability and temporal changes in this and on the foraging behaviour of harbour seals in the study area would contribute to the interpretation of the observed possible lunar patterns in behaviour.

A possible lunar influence on the hauling-out behaviour by the Pacific harbor seal (Phoca vitulina richardsi) at a haul-out site in British Columbia was reported by Watts (1993) who observed a significant reduction in the proportion of seals hauled out during a full moon and suggests seals spend more time foraging on bright moonlight nights. This hypothesis was based exclusively on changes in attendance of seals at haul-out sites. The haul-out behaviour of pinnipeds can now be more closely examined with the advances in telemetry and statistical modeling techniques. Simpkins et al., (2003) used such techniques to

169 try and identify patterns in and influences on harbour seal haul-out behaviour in Alaska. They suggest a 28 day periodicity evident in the date function may represent haul-out peaks of different demographic classes however they also acknowledged that the periodicity of the date function coincided with the lunar tidal cycle. The haul-out data modeled was data from the period of mid August to mid September only, so a monthly cyclic pattern, if present, was not identified. Using telemetry and statistical modeling techniques, the present study is the first to establish possible lunar patterns in the haul-out behaviour of individual seals and to identify differences in behavioural patterns between individuals and between tidal periods. Differentiating between tidal periods in the data exploration accounts for the fact that a particular tidal level will differ between periods e.g. a 1.5m tidal level on rising and falling first and second daily tides present different environmental and ecological parameters such as current speed and direction, daylight, food availability. The data suggest different haul-out patterns between the four tidal periods. Not differentiating between tidal periods when examining the effect of tidal level and other covariates such as time of day and month on the seals haul-out behaviour would conceal these patterns.

Overall apart from the similarity in the cyclic pattern in haul-out behaviour between seals 4 and 5 there appears to be otherwise large variation in the cyclic patterns in behaviour (i) between individuals, (ii) between tidal periods for each individual and (iii) over the tagging period. This has implications for using information on the behaviour of tagged individuals to derive correction factors for count data. Identifying an optimal model of the haul-out behaviour of a small sample of tagged seals as a function of covariates, using mixed modeling techniques and treating tag as a random factor are only applicable if all seals behave with random variations around the main pattern. The data resulting from this study shows large variation in behaviour between individuals. If the reason for the variation in haul-out behaviour between individuals was established (e.g. age, sex, location of haul-out site, season and possible interactions of such variables) this could be accounted for in a random effects model and haul-out probabilities under ‘ideal’ conditions or during surveys could be estimated, providing a means for correcting count data. Increasing the sample size of tagged

170 seals, with a more balanced age and sex ratio and including as many covariates as possible in the analysis would help to achieve this.

4.6 REFERENCES

Adkinson, M. D. &.Small, R. J. (2001). Evaluation of Alaska harbour seal (Phoca vitulina) population surveys: A simulation study. In: Harbor seal investigations in Alaska. Annual report for NOAA, award NA87FX0300. Alaska Department of Fish & Game, Division of Wildlife Conservation, Anchorage, AK. pp. 88-127. Adkinson, M. D., Quinn, T. J. & Small, R. J. (2003). Evaluation of the Alaska harbour seal (Phoca vitulina) population survey: A simulation study. Marine Mammal Science, 19, 764-790. Berry, A. J. (1986). Semi-lunar and lunar spawning periodicity in some tropical littorinid gastropods. Journal of Molluscan Studies, 52, 144–149. Boness, D. J., Bowen, W. D., Oftedal, O. T. (1994). Evidence of a maternal foraging cycle resembling that of otariid seals in a small phocid, the harbour seal. Behavioural Ecology and Sociobiology, 34, 95-104. Bonner, W.N. (1972) The Grey seal and Common seal in European waters. Oceanographic Marine Biology Annual Review, 10, 461-507. Boulva, J. & McLaren, I. A. 1979. Biology of the harbour seal Phoca vitulina in Eastern Canada. Fisheries Research Board of Canada Bulletin, 200, 24 pp. Boveng, P. L., Bengston, J. L., Withrow, D. E., Cesarone, J. C., Simpkins, M. A., Frost, K. J., & Burns, J. J. (2003) The abundance of harbor seals in the Gulf of Alaska. Marine Mammal Science, 19, 111-127. Brown, R. F. & Mate, B. R. (1983). Abundance, movements and feeding habits of harbor seals Phoca vitulina, at Netarts and Tillamook Bays, Oregon. Fishery Bulletin, 81, 292-301. Christy, J. H. (1986). Timing of larval release by intertidal crabs on an exposed shore. Bulletin of Marine Science, 39, 176-191.

171 Croxall, J. P., Everson, I., Kooyman, G. L., Ricketts, C. & Davis, R. W. (1985). Fur seal diving behaviour in relation to vertical distribution of krill. Journal of Animal Ecology, 54, 1-8. Davidson A. C. & Hinkley D. V. (1997). Bootstrap Methods and Their Applications. Cambridge University Press: Cambridge. Drescher, H. E. (1979). Biology, ecology and conservation of harbour seals in the tidelines of SchleswigHolstein. Beitrage zur Wildbiologie, 1, 1-73. Duck, C. D., Thompson, D. & Cunningham, L. (2005). The status of British common seal populations. In: Scientific Advice on Matters Related to the Management of Seal Populations. Briefing Paper, 05/4, 54-65. Daniel, R., Jemison, L. A., Pendleton, G. W. & Crowley, S. M. (2003). Molting phenology of harbour seals on Tugidak Island, Alaska. Marine Mammal Science, 19, 128-140. Ebling, F. J. & Hale, P. A. (1970). The control of the mammalian moult. Memoirs to the Society of Endocrinology, 18, 215-235, Faraway, J. J. (2006). Extending the linear model with R. Chapman & Hall/CRC. Fedak, M. A., Anderson, S. S. & Curry, M. G. (1983). Attachment of a radio tag to the fur of seals. Journal of Zoology London, 200, 298-300. Fedak, M., Lovell, P., McConnell, B., & Hunter, C. (2002). Overcoming the constraints of long range radio telemetry from animals: getting more useful data from smaller packages. Integrative and Comparative Biology, 42, 3-10. Frost, K. J., Lowry, L. F. & Ver Hoef, J. M. (1999). Monitoring the trend of harbour seals in Prince William Sound, Alaska after the Exxon Valdez oil spill. Marine Mammal Science, 15, 494-506. Frost, K.J., Simpkins, M.A., & Lowry, L.F. (2001) Diving behaviour of subadult and adult harbor seals in Prince William Sound, Alaska. Marine Mammal Science, 17, 813-834. Gliwicz, Z. M. (1986). A lunar cycle in zooplankton. Ecology, 67, 883-897. Gore, M. G. (2000). Spectrophotometry and spectrofluorimetry: A practical approach. Oxford University Press, 363pp. Grellier, K., Thompson, P. M. & Corpe, H. M. (1996). The effect of weather conditions on harbor seal (Phoca vitulina) haul-out behaviour in the

172 Moray Firth, northeast Scotland. Canadian Journal of Zoology, 74, 1806- 1811. Harcourt, R. G., Bradshaw, C. J., Dickson, K. & Davis, L. S. (2002). Foraging ecology of a generalist predator, the female New Zealand fur seal. Marine Ecology Progress Series, 227, 11-24. Härkönen, T. (1987). Seasonal and regional variations in the feeding habits of the harbour seal Phoca vitulina in the Skaggerak and Kattegat. Journal of Zoology, 213, 535-543. Härkönen, T. K., Harding, C. & Lunneryd, S. G. (1999). Age and sex specific behaviour in harbour seals Phoca vitulina leads to biased estimates of vital population parameters. Journal of Applied Ecology, 36, 825-841. Härkönen, T. & Heide-Jorgensen, M.P. (1990) Comparative life histories of East Atlantic and other harbour seal populations. Ophelia, 32, 211-235. Hastie, T. J. & Tibshirani, R. J. (1990). Generalised Additive Models. Chapman & Hall, London. Hays, G. C. (2003). A review of the adaptive significance and ecosystem consequences of zooplankton diel vertical migrations. Hydrobiologia, 503, 163-170. Hayward, J. L., Henson, S. M., Logan, C. J., Parris, C. R., Meyer, M. W. & Dennis, B. (2005). Predicting numbers of hauled-out harbour seals: a mathematical model. Journal of Applied Ecology, 42, 108-117. Heide-Jorgensen, M. P. & Härkönen, T. (1988). Rebuilding seal stocks in the Kattegat-Skagerrak. Marine Mammal Science, 4, 231-246. Horning, M. & Trillmich, F. (1999). Lunar cycles in diel prey migrations exert a stronger effect on the diving of juveniles than adult Galapagos fur seals. Proceedings of the Royal Society of London B, 266, 1127-1132. Horning, M. & Hill, R. D. (2005). Designing an archival satellite transmitter for life-long deployments on oceanic vertebrates: the life history transmitter. IEEE Journal of Oceanic Engineering, 30, 807-817. Huber, H. R. (1995). The abundance of harbour seals (Phoca vitulina richardsi) in Washington, 1991-1993. Unpublished M.Sc. thesis. University of Washington, Seattle, WA. Huber, H. R., Jeffries, S. J., Brown, R. F., Delong, R. L., & Vanblaricom, G. (2001). Correcting aerial survey counts of harbor seals (Phoca vitulina

173 richardsi) in Washington and Oregon. Marine Mammal Science, 17, 276- 293. Jeffries, S. J. (1986). Seasonal movements and population trends of harbour seals (Phoca vitulina richardsi) in the Columbia River and adjacent waters of Washington and Oregon: 1976-1982. Report to the US Marine Mammal Commission, Contract No: MM30793575. Jeffries, S., Huber, H., Calambokidis, J. & Laake, J. (2003). Trends and status of harbour seals in Washington State: 1978-1999. Journal of Wildlife Management, 67, 207-218. Jemison, L. A. & Pendleton, G. W. 2001. Harbour seal population trends and factors influencing counts on Tugidak Island, Alaska. In: Harbor seal investigations in Alaska. Annual report for NOAA, award NA87FX0300. Alaska Department of Fish & Game, Division of Wildlife Conservation, Anchorage, AK. pp. 31-52. Lander, M. E., Haulena, M., Gullan, F. M. D. & Harvey, J. T. (2005). Implantation of subcutaneous radio transmitters in the harbour seal (Phoca vitulina). Marine Mammal Science, 21, 154-161. Ling, J. K. (1970) Pelage and molting in wild mammals with special reference to aquatic forms. The Quarterly Review of Biology, 45, 16-54. Ling, J. K. & Bryden, M. M. (1981). Southern elephant seal Mirounga leonina Linneaus, 1758. In: Ridgeway, S. H. & Harrison, R. (Eds.) Handbook of marine mammals. Volume 2. Seals. Academic Press, New York, NY. pp. 297-327. Lowry, L. F., Frost, K. J., Ver Hoef, J. M. & Delong, R. A. (2001). Movements of satellite-tagged subadult and adult harbour seals in Prince William Sound, Alaska. Marine Mammal Science, 17, 4, 835-861. Lyderson, C & Kovacs, K. (2005). Growth and population parameters of the

world s northernmost harbour seals Phoca vitulina residing in Svalbard, Norway. Polar Biology, 28, 2 156-163. Mathews, E. A. & Kelly, B. P. (1996). Extreme temporal variation in harbor seal (Phoca vitulina richardsi) numbers in Glacier Bay, a glacial fjord in S.E. Alaska. Marine Mammal Science, 12, 483-489.

174 Moran, J. R., Adkinson, M. D. & Kelly, B. P. (2004). Counting seals, estimating the unseen fraction using a covariate and capture-recapture model. Unpublished M.Sc. thesis. University of Alaska Fairbanks. 32 pp. Montogomery, R. A. (2005). Modeling the terrestrial habitat use of harbour seals. Unpublished. M.Sc. thesis, University of Washington. 46 pp. McConnell, B., Chambers, C., Nicholas, K.S., & Fedak, M. (1992) Satellite tracking of grey seals (Halichoerus grypus). Journal of Zoology London, 226, 271-282. McConnell, B. J., Fedak, M. A., Lovell, P. & Hammond, P. S. (1999). Movements and foraging areas of grey seals in the North Sea. Journal of Applied Ecology, 36, 1-19. McConnell, B., Fedak, M., Burton, H. R., Engelhard, G. H., & Reijenders, P. J. H. (2002) Movements and foraging areas of naive recently weaned southern elephant seal pups. Journal of Animal Ecology, 71, 65-78. McConnell, B. J., Beaton, R., Bryant, E., Hunter, C., Lovell, P. & Hall. A. (2004). Phoning home- a new GSM mobile phone telemetry system to collect mark-recapture data. Marine Mammal Science, 20, 2, 274-283. Olesiuk, P. F., Bigg, M. A. & Ellis, G. M. (1990). Recent trends in abundance of harbour seals, Phoca vitulina, in British Columbia. Canadian Journal of Fisheries and Aquatic Sciences, 47, 992-1003. Ostrom, C. W. (1990). Time series analysis: Regression Techniques. 2nd edition. Sage Publications Inc. Reder, S., Lydersen, C., Arnold, W. & Kovacs, K. M. (2003). Haul-out behaviour of high Arctic harbour seals (Phoca vitulina vitulina) in Svalbard, Norway. Polar Biology, 27, 6-16. Rehberg, M. J. & Small, R. J. (2001). Dive behaviour, haulout patterns and movements of harbour seal pups in the Kodiak archipelago, 1997-2000. In: Harbor seal investigations in Alaska. Annual report for NOAA, award NA87FX0300. Alaska Department of Fish & Game, Division of Wildlife Conservation, Anchorage, AK. pp. 209-238. Reijnders, P. J. H., Ries, E. H., Tougaard, S., Norgaard, N., Heidemann, G., Schwartz, J., Vareshi, E. & Traut, I. M. (1997). Population development of harbour seals Phoca vitulina in the Wadden Sea after the 1988 virus epizootic. Journal of Sea Research, 38, 161-168.

175 Ries, E.H., Hiby, L.R., & Reijecders, P.J.H. (1998) Maximum likelihood population size estimation of harbour seals in the Dutch Wadden Sea based on a mark-recapture experiment. Journal of Applied Ecology, 35, 332-339. Sharples, R. J., Mattiopoulos, J. & Hammond, P. S. (2004). Distribution and movements of harbour seals around Orkney, Shetland and the Wash. Sea Mammal Research Unit, University of St. Andrews, St. Andrews, Fife, UK. Final report to Geotek. 23 pp. Sharples, R. (2005) phd thesis –get title from ruth Simpkins, M. A., Withrow, D. E., Cesarone, J. C. & Boveng, P. L. (2003). Stability in the proportion of harbor seals hauled out under locally ideal conditions. Marine Mammal Science, 19, 792-805. Stewart, B. S. (1984). Diurnal patterns of harbour seals at San Miguel Island, California. Journal of Wildlife Management, 48, 1459-1461. Stewart, B. S. & Yochem, P. K. (1983). Seasonal abundance of pinnipeds at San Nicolas Island, California, 1980-1982. Bulletin of the Southern California Academy o Sciences, 83, 121-132. Small, R. J., Pendleton, G. W. & Wynne, K. M. (2001). Harbor seal population trends in the Ketchikan, Sitka and Kodiak areas of Alaska 1983-1999. In: Harbor seal investigations in Alaska. Annual report for NOAA, award NA87FX0300. Alaska Department of Fish & Game, Division of Wildlife Conservation, Anchorage, AK. 8-21. Small, R. J., Pendleton, G. W. & Pitcher, K. W. (2003). Trends in abundance of Alaska harbor seals, 1983-2001. Marine Mammal Science, 19, 2, 344- 362. Thompson, P. M. & Rothery, P. (1987). Age and sex differences in the timing of moult in the common seal, Phoca vitulina. Journal of Zoology London, 212, 597-603. Thompson, P. M., Fedak, M., McConnell, B., & Nicholas, K. S. (1989). Seasonal and sex-related variation in the activity patterns of common seals (Phoca vitulina). Journal of Applied Ecology, 26, 521-535. Thompson, P. M., Miller, D., Cooper, R., & Hammond, P. S. (1994). Changes in the distribution and activity of female harbour seals during the

176 breeding season: implications for their lactation strategy and mating patterns. Journal of Animal Ecology, 63, 24-30. Thompson, P .M., Tollit, D. J., Wood, D., Corpe, H. M., Hammond, P. S., & Mackay, A. (1997). Estimating harbour seal abundance and status in an estuarine habitat in north-east Scotland. Journal of Applied Ecology, 34, 43-52. Thompson, P. M. & Harwood, J. (1990). Methods for estimating the population size of common seals Phoca vitulina. Journal of Applied Ecology, 27, 924-938. Thompson, P. M. & Miller, D. (1990). Summer foraging activity and movements of radio-tagged common seals (Phoca vitulina) in the Moray Firth, Scotland. Journal of Applied Ecology, 27, 492-501. Thompson, D., Moss, S. E., & Lovell, P. (2003) Foraging behaviour of South American fur seals Arctocephalus australis; extracting fine scale foraging behaviour from satellite tracks. Marine Ecology Progress Series, 260, 285-296. Thompson, D., Lonergan, M., & Duck, C. (2005). Population dynamics of harbour seals Phoca vitulina in England: Monitoring growth and catastrophic declines. Journal of Applied Ecology, 42, 638-648. Trillmich, F. & Mohren, W. (1981). Effects of the lunar cycle on the Galapagos fur seal Arctocephalus galapagoensis. Oecologia, 48, 85-92. Van Paris, S. M., Thompson, P. M., Tollit, D. J., & Mackay, A. (1997) Distribution and activity of male harbour seals during the mating season. Animal Behaviour, 54, 35-43. Van Parijs, S. M., Hastie, G. D. & Thompson, P. M. (1999). Geographic variation in temporal and spatial patterns of aquatic mating male harbour seals. Animal Behaviour, 58, 1231-1239. Van Paris, S. M., Hastie, G. D., & Thompson, P. M. (2000) Individual and geographical variation in display behaviour of male harbor seals in Scotland. Animal Behaviour, 59, 559-568. Watts, P. (1993). Possible lunar influence on hauling-out behaviour by the Pacific harbour seal (Phoca vitulina richardsi). Marine Mammal Science, 9, 1, 68-76.

177 Watts, P. (1996). The diel hauling-out cycle of harbour seals in an open marine environment: Correlates and constraints. Journal of Zoology, London, 240, 175-200. Wilson, S. C. (1978). Social organisation and behaviour of harbor seals, Phoca vitulina concolor, in Maine. Final report to Marine Mammal Commission. Contract MM6AC013. NTIS PB-280188. 103pp. Winkle, J., Edwards, R. M., McConnell, B. J. & Bryant, E. (in press). Design, fabrication and measurement of an encapsulated inverted F dual band antenna for the gathering of data on seals at sea over a GSM system. IEE Transactions on Antennas and Propagation. Yochem, P. K., Stewart, B. S., Delong, R. L., & DeMaster, D. P. (1987). Diel haul-out patterns and site fidelity of harbour seals (Phoca vitulina richardsi) on San Miguel Island, California in Autumn. Marine Mammal Science, 3, 323-332. Zuur, A. F., Ieno, E. N & Smith, G. M. (2006). Analysis of ecological data. Springer Verlag. 688 pp.

178

CHAPTER 5

CETACEAN DISTRIBUTION AND RELATIVE ABUNDANCE IN SOUTHWEST IRELAND

179 5.1 INTRODUCTION AND METHODOLOGY

All cetaceans present in European waters are protected under EU law and listed in Annex IV of the EU Habitats Directive, as species of community interest in need of strict protection. In addition bottlenose dolphins (Tursiops truncatus), harbour porpoises (Phocoena phocoena) and two seal species (harbour and grey seal, Phoca vitulina & Halichoerus grypus) are listed in Annex II as requiring the designation of Special Areas of Conservation (SAC’s). In 1991 the Irish government declared Irish waters a whale and dolphin sanctuary including the State’s 200-mile exclusive fishery limit (Rogan & Berrow, 1995).

A number of cetacean species were observed during the course of this study. Results from both shore and boat-based surveys are presented here. Boat-based line-transect techniques are detailed in chapter 1 and shore-watch methodologies are described in chapter 2. Incidental sightings from boat-based surveys of seals in Kenmare River were also included. Groups were defined as: an aggregation of animals whose members were within 100m of one another, engaged in similar activities and, if moving, heading in the same direction (Wells et al., 1987). All observations were carried out in sea states of 4 or less on the Beaufort Scale.

5.2 SPECIES ACCOUNTS

A total of six cetacean species were recorded over the course of the study, including four odontocete species and at least two mysticete species (Table 1). Species richness was highest at the Mizen Head shore-watch site where six species, including a large unidentified whale (probably fin Balaenoptera physalus or sei Balaenoptera borealis whale) were recorded. Species richness was also high at Dursey Island where 5 cetacean species were observed. Species richness was lowest at the inner Bantry Bay site and incidental records from the Kenmare river show a minimum of 2 cetacean species utilize this site. Harbour porpoises were recorded at all sites and common dolphins (Delphinus delphis)

180 and minke whales (Balaenoptera acutorostrata) were recorded at all sites apart from the Kenmare River.

Table 1. Presence/absence of cetacean species recorded in the RAMSSI study area from shore and boat-based surveys, including incidental sightings from seal surveys. ● = present SPECIES Sheeps Inner Dursey Black Mizen Three Ken- Head Bantry Island Ball Head Castle mare Head Head River ODONTOCETES Harbour Porpoise (Phocoena phocoena) ● ● ● ● ● ● ● Common Dolphin (Delphinus delphis) ● ● ● ● ● ● Bottlenose Dolphin (Tursiops truncatus) ● ● ● Risso’s Dolphin (Grampus griseus) ● ● ● ● MYSTICETES Minke Whale (Balaenoptera acutorostrata) ● ● ● ● ● ● Fin Whale (Balaenoptera physalus) ● Large Whale (Balaenoptera sp.) ●

5.2.1 Harbour porpoise (Phocoena phocoena) The harbour porpoise is the smallest of all the cetacean species, and probably the most abundant in Irish waters (Rogan & Berrow, 1996). This species was recorded at all sites in relatively high numbers e.g. 181 were recorded from the Three Castle Head shore-watch site between July 2003 and September 2004, an average of 8 per survey. An average of 19 porpoises per survey were recorded at Dursey Island where only 9 surveys were carried out. Calves or juveniles were sighted at all sites, particularly in summer in Bantry Bay but also in January and February at the outer headlands. These are likely to have been born the previous summer as this species is thought to have a summer breeding season in Irish waters (Rogan and Berrow, 1996). The mean group size of porpoises was 2.1±0.1 with a maximum group size of 25 occurring at Sheep’s Head.

Boat-based surveys of Bantry Bay revealed that harbour porpoises were widely distributed throughout the bay, occurring as far inshore as Whiddy Island (Figure 1). This is confirmed from plots of shore-based observations which showed widespread use of the inner Bantry survey area (Figure 2b). Figure 3 (a-c) shows the seasonal abundance of harbour porpoises at all shore-watch sites. This species was present all year round at all sites apart from winter at the Bantry Bay sites. Peak numbers occurred in autumn at most sites however numbers peaked

181 in summer at Dursey Island. Mean numbers remained relatively constant all year round at Black Ball Head indicating the importance of this region to Harbour porpoises throughout the year. Clearly Bantry Bay and its approaches represent an important habitat for this Annex II species, and as such should be considered for designation as an SAC. Previous studies in Irish waters have also identified southwest Ireland as an area of high porpoise abundance in summer (Northridge et al., 1995; Pollock et al., 1997; Mackey et al., 2004).

Figure 1. The average number per kilometre of harbour porpoises seen from boat survey transects in Bantry Bay during the study. N

5.2.2 Common Dolphin (Delphinus delphis) Common dolphins were recorded at all shore-watch sites in relatively high numbers. A total of 621 individuals were recorded during the course of the study with the highest total number (313) recorded at the Three Castle Head site. Figure 2 (c-d) shows the position of common dolphin sightings recorded from the Bantry Bay shore-watch sites and reveals that group sizes of 15-30 individuals were typical here. The mean group size of common dolphins over the entire study area was 11.3±1.9 with a maximum group size of 80 occurring at Black Ball Head in July. Figure 3 (d-f) shows that mean common dolphin numbers peaked in autumn at most sites. A summer peak in abundance was recorded at Dursey Island and Black Ball Head however. Common dolphins were not recorded at any of the sites in spring and appeared largely absent in winter apart from a group of 15 which was sighted from the Sheep’s Head site in January (Figure 3d). Many of the groups sighted appeared to be actively foraging, with frequent milling and surface-rush activities observed. These foraging groups were often accompanied by feeding seabirds, most commonly gannets (Morus bassanus) and manx shearwaters (Puffinus puffinus). On one occasion common dolphins were seen feeding in close proximity to a minke whale. At least six

182 calves were observed in the survey area all between the months of August and October with 2 of these occurring at the inner Bantry site. This indicates that the entire survey area represents an important habitat for this species both as a foraging and nursery ground. Common dolphins were the most abundant cetacean species encountered in Irish shelf and offshore waters by (Pollock et al., 1997; Mackey et al., 2004).

Common Dolphin (Photo. M. Mackey).

5.2.3 Risso’s Dolphin (Grampus griseus) A total of 11 Risso’s dolphins were recorded from shore-watch observations in the survey area, with a further 3 observed during boat-based surveys of Bantry Bay. All sightings were of 1-3 individuals with no large groups observed. Risso’s dolphins were exclusively recorded in the months of September and October and were never recorded in the inner portion of Bantry Bay or at Black Ball Head. This species is thought to prey exclusively on squid and is typically offshore in distribution for much of the year, generally only coming inshore between August and February (Leatherwood & Reeves, 1983). Risso’s dolphins were recorded in relatively high numbers in waters off southwest Ireland by Pollock et al. (1997) and Hammond et al. (2002), indicating that this region may be an important local concentration of the species.

183 5.2.4 Bottlenose Dolphin (Tursiops truncatus) The low sighting rate of bottlenose dolphins in this study is surprising considering the abundance of this species in inshore waters on the west coast (e.g. the Shannon Estuary and Tralee Bay) in summer (Ingram & Rogan, 2002; Ingram et al., 2005). Studies in Ireland’s Atlantic Margin have also indicated that the waters around the Porcupine Bank and Porcupine Seabight off southwest Ireland may be important for this species (Mackey et al., 2004). Small numbers (2-3) of this species were recorded from both the Mizen Head and Three Castle Head shore-watch sites, with highest numbers (8-10) recorded on three occasions during boat-based seal surveys of the Kenmare River. It is likely that bottlenose dolphins were foraging for Atlantic salmon (Salmo salar) associated with the Kenmare River as they are know to do so in the Shannon Estuary further north on the west coast (Ingram & Rogan, 2002). All bottlenose dolphin sightings took place in summer.

5.2.5 Minke whales (Balaenoptera acutorostrata) Minke whales were recorded in relatively high numbers at all shore-watch sites as well as from boat-based surveys in Bantry Bay. A total of 85 minke whales were recorded during the course of the study, with 27 of these recorded at the Three Castle Head site. A total of 19 minke whales were sighted from the outer Bantry shore-watch site and one was sighted from the inner site (Figure 2 e-f). Most sightings were of single individuals however several pairs were also recorded. Boat-based surveys of Bantry Bay confirmed that minke whale distribution was almost entirely restricted to outer regions of Bantry Bay. A total of six minke whales were recorded from boat-based surveys, with all sightings occurring near the mouth of the bay (Figure 4). Minke whale numbers peaked in autumn at all sites and this species was never recorded in spring (Figure 3g-i). Minke whales were largely absent in winter, however two individuals were sighted at Three Castle Head in November. One calf and 20 immature minke whales were observed in the study area over the course of the survey indicating that this is an important habitat for this species. Pollock et al. (1997) found relatively large numbers of minke whales in inshore waters around southwest Ireland. It is likely that this species is attracted by the rich feeding grounds

184 associated with the Irish Shelf Front near the mouth of Bantry Bay in summer (Raine et al., 1990).

A number of minke whale/seabird feeding associations were observed during the course of this study. The species most commonly recorded in association with minke whales were manx shearwaters, which alone accounted for 66% of all associations. The remaining associations consisted of single-species feeding flocks of gannets or kittiwakes (Rissa tridactyla) and mixed feeding flocks of gannets, manx shearwaters, Alcidae (auks) and Larus (gull) species. Minke whales were also commonly associated with harbour porpoises.

Figure 4. Numbers and positions of minke whale sightings in Bantry Bay from boat-based observations.

5.2.6 Fin Whale (Balaenoptera physalus) One fin whale was observed from the Black Ball Head shore-watch site during the course of this study. This whale was observed feeding and breaching in close proximity to two minke whales. A further two large unidentified whales were sighted from the Mizen Head shore-watch site and it is likely that these were either fin or sei whales. All sightings took place in October. Incidental sightings from shore-based observation points on the south coast of Ireland (e.g. Galley Head and the Old Head of Kinsale) have shown that this species regularly occurs along the south coast from October to December (www.IWDG.ie). There is evidence to suggest that Irelands Atlantic Margin forms part of the migratory pathway of a number of large baleen whales, including fin, sei, humpback (Megaptera novaeangliae) and even blue whales (Balaenoptera musculus) as they move from winter calving grounds in the south to summer feeding grounds at high latitudes(Charif et al., 2001; Harwood & Wilson, 2001; Mackey et al., 2004).

185 a) Sheep’s Head Porpoise b) Inner Bantry Porpoise

c) Sheep’s Head Common Dolphins d) Inner Bantry Common Dolphins

e) Sheep’s Head Minke Whales f) Inner Bantry Minke Whales

Figure 2 (a-f). Location and number of harbour porpoise’s (a-b), common dolphin’s(c-d) and minke whales (e-f) at the Sheep’s Head and Inner Bantry shore-watch sites during the entire survey period. Positions calculated using a N theodolite.

186

n a

9 c 7 s r can 8 6 s pe r

7 s e n

i 5

s p 6 ph se i 5 4 o p r 4 3 o on dol p 3 m . 2 o m n 2 o

c n

. 1 a 1 e no M

0 n 0 Spring Summer Autumn Winter a Spring Summer Autumn Winter Me

a) Inner bantry Sheeps Head d) Inner bantry Sheeps Head n n a a

45 c 60 c s s

r r 40 e 50 pe p 35 e ns s i

o 30 40 phi p l r

o 25 do 30 P 20 on m

bour 15 20 r m a 10 o c H 10 5 no. no. n n

a 0 0 a e e

M Spring Summer Autumn Winter Spring Summer Autumn Winter M

b) Dursey Island Black Ball Head e) Dursey Island Black Ball Head n n a a

18 c 50 c s s

r r 16 45 e pe p 14 40 e ns s

i 35

o 12 phi p l

r 30

o 10 do 25 P 8 on 20 m

bour 6

r 15 m a 4 o c

H 10

2 5 no. no. n n

a 0 0 a e e

M Spring Summer Autumn Winter Spring Summer Autumn Winter M

c) Mizen Head Three castle Head f) Mizen Head Three castle Head

Figure 3 (a-f). Seasonal abundance (Mean ± SE) of harbour porpoises (a-c) and common dolphins (d-f) at the Sheep’s Head, Inner Bantry, Dursey Island, Black ball Head, Mizen Head and Three Castle Head shore-watch sites.

187

0.9 an 0.8 sc r e 0.7 p

es 0.6 al h 0.5 w e 0.4 k n i 0.3 m . 0.2 o

n 0.1 n a e 0 M Spring Summer Autumn Winter

Inner bantry Sheeps Head g)

n 3.5 a c s

r 3 pe

s 2.5 e l a

h 2 W

e 1.5 nk i 1 M no.

0.5 n a e 0 M Spring Summer Autumn Winter

h) Dursey Island Black Ball Head

n 4 a c

s 3.5 r

pe 3 s e l

a 2.5 h

W 2 e

nk 1.5 i

M 1 no. 0.5 n a e 0 M Spring Summer Autumn Winter

i) Mizen Head Three castle Head

Figure 3 (g-i). Seasonal abundance (Mean ± SE) of minke whales at the Sheep’s Head & Inner Bantry (g), Dursey Island & Black ball Head (h), Mizen Head & Three Castle Head (i) shore-watch sites.

188

5.3 CONSERVATION RECOMMENDATIONS

Two of the six cetacean species recorded in this study are listed under Annex II of the EU Habitats Directive as species of Community Interest, whose conservation requires the designation of Special Areas of Conservation (SAC). Glengarriff Harbour in Bantry Bay and The Kenmare River have already been designated as SAC’s in order to protect the resident harbour seal populations occurring there. The remainder of Bantry Bay and it’s approaches out to Dursey Island and Mizen Head have no such designation however and should be considered for SAC status because of it’s widespread use by the annex II harbour porpoise throughout the year. This designation would also protect the large numbers of feeding minke whales and common dolphins (including their young) that frequent the region in summer and autumn.

189 5.4 REFERENCES

Charif, R., Clapham, P.J. & Clarke, C. 2001. Acoustic detections of singing humpback whales in the deep waters off the British Isles. Marine Mammal Science 17, 751-769. Hammond, P.S., Heimlich, S., Benke, H., Berggren, P., Borchers, D.L., Buckland, S.T., Collet, A., Heide-Jorgensen, M.P., Hiby, A.R., Leopold, M.F. & Oien, N. 2002. Distribution and abundance of the Harbour porpoise and other small cetaceans in the North Sea and adjacent waters. Journal of Applied Ecology 29, 361 - 376. Harwood, J. & Wilson, B. 2001. The implications of developments on the Atlantic Frontier for marine mammals. Continental Shelf Research 21, 1073–1093. Ingram, S. & Rogan, E. 2002. Identifying critical areas and habitat preferences of bottlenose dolphins Tursiops truncatus. Marine Ecology Progress Series 244, 247-255. Ingram, S.N., Englund, A., O'Donovan, M., Walshe, L. & Rogan, E. 2005. A survey of marine wildlife in Tralee Bay and adjacent waters. Report to Tuatha Chiarraí Teo. University College Cork, Cork. pp 20. Leatherwood, S. & Reeves, R.R. 1983. The Sierra Club handbook of whales and dolphins. Sierra Club Books, San Francisco. pp 302. Mackey, M., O'Cadhla, O., Kelly, T.C., Aguiler de Soto, N. & Connolly, N. 2004. Cetaceans and Seabirds of Ireland's Atlantic Margin. Volume 2 - Cetacean distribution and abundance. Report on research carried out under the Irish Infrastructure Programme (PIP): Rockall Studies Group (RSG) projects 98/6 and 00/13, Porcupine Studies Group project P00/15 and Offshore Support Group (OSG) project 99/38, Cork. pp 89. Northridge, S.P., Tasker, M.L., Webb, A. & Williams, J.M. 1995. Distribution and relative abundance of harbour porpoises (Phocoena phocoena L.), white-beaked dolphins (Lagenorhynchus albirostris Gray), and minke whales (Balaenoptera acutorostrata Lacapede) around the British Isles. ICES Journal of Marine Science 52, 55-66.

190 Pollock, C., Reid, J., Webb, A. & Tasker, M. 1997. The distribution of seabirds and cetaceans in the waters around Ireland. JNCC Report, No. 267, Peterborough. pp 167. Raine, R., O`Mahony, J., McMahon, T. & Roden, C. 1990. Hydrography and Phytoplankton of waters off Southwest Ireland. Estuarine, Coastal and Shelf Science 30, 579-592. Rogan, E. & Berrow, S. 1995. The management of Irish waters as a whale and dolphin sanctuary. In: A.S. Blix, L. Walloe & O. Ulltang (eds), Developments in Marine Biology, 4. Whales, seals, fish and man. Elsevier, Amsterdam. pp 671-683. Rogan, E. & Berrow, S. 1996. A review of harbour porpoises, Phocoena phocoena, in Irish waters. Report of the International Whaling Commission 46, 595-605. Wells, R.S., Scott, M.D. & Irvine, A.B. 1987. The social structure of free- ranging bottlenose dolphins. In: H.H. Genoways (ed). Current mammology Vol. 1. Plenum Press, New York. pp 247-305.

191

CONCLUSIONS

192 The information on seabird and marine mammal distribution and relative abundance recorded in this study provides a valuable baseline for future monitoring programs. The impacts of future disturbances on the communities in these sites can now be reliably assessed through a comparison of ‘before’ (i.e. this study) and ‘after’ (post-disturbance) datasets using standardized boat-survey techniques. The use of shore-based monitoring of disturbed sites is also recommended in order to establish seasonal trends and to accurately survey species such as storm petrels, Alcidae and harbour porpoises which are known to avoid survey vessels (see Tasker et al., 1984). This baseline data can also aid the identification of future climate-induced shifts in species distribution.

The physical habitat variable ‘seaward distance’ provides a crude indication of high overall seabird abundance for use in reactive situations following pollution events in similar un-surveyed embayments. However further study of seabird habitat-use in alternative sites is needed to establish the reliability of physical habitat variables in determining seabird distribution for use in, for example, oil- pollution management plans.

The high density and species richness of seabirds in the entire RAMSSI study area, coupled with the presence of a number of Annex 1 species and species with a high Oil Vulnerability Index (OVI>20) indicates that this site should be considered for designation as a Marine Protected Area (MPA) for seabirds. According to Hyrenbach et al. (2000), MPA’s designed to protect specific foraging grounds could mitigate certain forms of habitat degradation (e.g. oil spills). However larger-scale designations (encompassing foraging ranges and migration routes) are needed to adequately protect far-ranging species such as Manx shearwaters (Hyrenbach et al., 2000).

The widespread and year-round use of the study area by large numbers of the Annex II Harbour porpoise indicates that this region, particularly Bantry Bay, should be considered for designation as a Special Area of Conservation (SAC). This designation would also protect the large numbers of feeding minke whales and common dolphins (including their young) that frequent the region in summer and autumn.

193

This study provided information, hitherto unavailable, on the seasonal changes in the distribution and abundance of harbour seals within two SACs in southwest Ireland and identified sites important for breeding and moulting and sites that are used year round. The importance of Glengarriff harbour for breeding seals along with the relatively high levels of anthropogenic disturbance during the breeding period should be important considerations in the management of this SAC.

The GSM telemetry system proved an effective means of obtaining information on the haul-out activity of harbour seals in the study area that was respectively less labour intensive and cheaper than VHF and satellite telemetry. It also provided a means of testing the efficiency of using GSM technology to relay data ashore in southwest Ireland and served as a feasibility study for future work in the area. This study identified large variation in the behaviour of individual seals and suggests that caution should be exercised when making inferences on the haul-out behaviour of the ‘population’ based on the behaviour of tagged individuals.

RECOMMENDATIONS FOR FUTURE WORK

Further studies of seabird habitat-use in both estuarine and marine embayments (e.g. Cork Harbour and Dingle Bay) should be carried out in order to determine the reliability of physical habitat variables in determining seabird distribution. Comparable techniques (i.e. RIB-based surveys using modified JNCC survey methodology) should be employed and the use of a 1km² grid network is recommended for data analysis. Ideally, information on oceanographic parameters such as salinity, productivity and current flow should also be recorded. Satellite data on sea surface temperatures (SST) and chlorophyll a concentrations could also be used to identify seasonal variations in the position of the Irish Shelf Front and other areas of high productivity (see Jaquemet et al., 2005).

194 Investigations of seabird diet, e.g. from regurgitated pellets (Montevecchi et al., 1988; Barrett et al., 1990; Zijlstra & Van Erden, 1995; Carss, 1997) stable isotope analysis (Bearhop et al., 1999) or mist-netting (Corkhill, 1973) at seabird colonies adjacent to the study sites could provide a useful indication of important prey species for seabirds in these areas and reveal seasonal shifts in diet. Direct relationships between seabird distribution and prey abundance could also be determined using a towed bioacoustics instrument during boat surveys (Ainley et al., 2005).

Similar investigations into the year-round patterns in Harbour seal abundance and site use should be carried out in other areas along the Irish coastline, that include a selection of harbour seal haul-out sites and represent the full range of haul-out substrate type, both within and outside of protected areas. Statistical modelling of the counts as a function of covariates will provide a means of determining which processes are influencing the seals’ haul-out behaviour at these sites and if this varies spatially and temporally.

A covariate modelling approach such as that used in this study could be included in future national harbour seal surveys in Ireland to help improve the accuracy of population estimates over a wide geographical area by ‘controlling’ for or standardising environmental conditions and accounting for covariate associated variation in counts.

The potential seasonal change in Harbour seal behaviour suggested in the present study could be explored further by obtaining information on the haul-out behaviour of individuals throughout the entire annual cycle. Flipper mounted telemetry devices (Huber et al., 2001; Simpkins et al., 2003) or implantable sub- cutaneous VHF tags (Horning & Hill, 2005; Lander et al., 2005) could be a possible solution to moult associated tag loss but the information from these tags would be limited to haul-out patterns. The approach should be dependent on the study objectives: if the main objective is to devise a correction factor for count data, information on haul-out behaviour during the moult is critical and flipper mounted tags would be appropriate; if the objective is to study movements and behaviour of the seals offshore in order to identify home range and offshore

195 foraging areas, then pelage mounted tags are necessary as the antennae need to break the water surface for successful data capture and relay.

The potential lunar effect on the haul-out behaviour of seals observed in this study warrants further investigation. The approach described in the present study could be used on existing and future data on the haul-out behaviour of individual harbour seals across their geographical range to determine if the patterns in behaviour observed in the present study are common phenomena and to identify any potential demographical or geographical variation.

196 REFERENCES

Ainley, D.G., Spear, L.B., Tynan, C.T., Barth, J.A., Pierce, S.D., Ford, R.G. & Cowles, T.J. 2005. Physical and biological variables affecting seabird distributions during the upwelling season of the northern California Current. Deep-Sea Research Part II 1-2, 123-143. Barrett, R.T., N., R., Loen, J. & Montevecchi, W.A. 1990. Diets of Shags Phalacrocorax aristotelis and cormorants P.carbo in Norway and possible implications for gadoid stock recruitment. Marine Ecology Progress Series 66, 205-218. Bearhop, S., Thompson, D.R., Waldron, S., Russell, I.C., Alexander, G. & Furness, R.W. 1999. Stable Isotopes indicate the extent of freshwater feeding by cormorants Phalacrocorax carbo shot at inland fisheries in England. Journal of Applied Ecology 36, 75-84. Carss, D.N. 1997. Techniques for assessing Cormorant diet and food intake: towards a consensus view. Suppl. Ric. Biol. Selvaggina xxvi, 197-230. Corkhill, P. 1973. Food and feeding ecology of puffins. Bird Study 20, 207-220. Horning, M. & Hill, R. D. (2005). Designing an archival satellite transmitter for life-long deployments on oceanic vertebrates: the life history transmitter. IEEE Journal of Oceanic Engineering, 30, 807-817. Hyrenbach, K.D., Forney, K.A. & Dayton, P.K. 2000. Marine protection areas and ocean basin management. Aquatic Conservation: Marine and Freshwater Ecosystems 10, 437-458. Huber, H. R., Jeffries, S. J., Brown, R. F., Delong, R. L., & Vanblaricom, G. (2001). Correcting aerial survey counts of harbor seals (Phoca vitulina richardsi) in Washington and Oregon. Marine Mammal Science, 17, 276- 293. Jaquemet, S., Le Corre, M., Marsac, F., Potier, M. & Weimerskirch, H. 2005. Foraging habitats of the seabird community of Europa Island (Mozambique Channel). Marine Biology 147, 573-582. Lander, M. E., Haulena, M., Gullan, F. M. D. & Harvey, J. T. (2005). Implantation of subcutaneous radio transmitters in the harbour seal (Phoca vitulina). Marine Mammal Science, 21, 154-161.

197 Montevecchi, W.A., Birt, V.L. & Cairns, D.K. 1988. Dietary Changes of Seabirds Associated with Local Fisheries Failures. Biological Oceanography 5, 153-161. Simpkins, M. A., Withrow, D. E., Cesarone, J. C. & Boveng, P. L. (2003). Stability in the proportion of harbor seals hauled out under locally ideal conditions. Marine Mammal Science, 19, 792-805. Tasker, M.L., Hope Jones, P., Dixon, T. & Blake, B.F. 1984. Counting seabirds at sea from ships: A review of the methods employed and a suggestion for a standardised approach. The Auk 101. Zijlstra, M. & Van Erden, M.R. 1995. Pellet production and the use of Otoliths in determining the diet of cormorants Phalacrocorax carbo sinensis: trials with captive birds. Ardea 83, 123-131.

198