JNCC Report

No. XXX

Seabird monitoring on Skomer Island in 2013

Birgitta Büche, Edward Stubbings and Lewis Yates

The Wildlife Trust of South and West Wales The Welsh Wildlife Centre Cilgerran, Cardigan SA43 2TB

D Boyle and CM Perrins

Edward Grey Institute Department of Zoology South Parks Road Oxford OX1 3PS

November 2013

This document should be cited as: Büche, BI, Stubbings, EM, Boyle, D. and Perrins, CM, Yates, L. (2013) Seabird monitoring on Skomer Island in 2013.JNCC Report No.XXX Table of Contents

Summary 6 1 Introduction 8 1.1 Introduction to survival estimates 9 2 General methods 9 2.1 Whole island counts 9 2.2 Study plot counts of Common Guillemots Uria aalge and Razorbills Alca torda 10 2.3 Breeding success 11 2.4 Weather 11 3 Northern Fulmar Fulmaris glacialis 12 3.1 Breeding numbers - whole island counts 12 3.2 Breeding success 13 3.2.1 Methods 13 .2.2 Results 13 3.3 Timing of breeding 15 4 European Storm-petrel Hydrobates pelagicus 16 4.1 Breeding numbers 16 5 Manx Shearwater Puffinus puffinus 18 5.1 Study plot census 18 5.2 Breeding success 21 5.3 Adult survival 22 6 Great Cormorant Phalacrocorax carbo 24 6.1 Breeding numbers 24 6.2 Breeding success 24 7 European Shag Phalacrocorax aristotelis 25 7.1 Breeding numbers 25 7.2 Breeding success 25 8 Lesser Black-backed Gull Larus fuscus 26 8.1 Methods for estimating breeding numbers 26 8.2 Breeding numbers – results 28 8.3 Breeding success 31 8.4 Adult survival 33 9 Herring Gull Larus argentatus 34 9.1 Breeding numbers 34 9.2 Breeding success 35 9.3 Adult survival 36 10 Great Black-backed Gull Larus marinus 37 10.1 Breeding numbers 37 10.2 Breeding success 38 10.3 Diet study 38 11 Black-legged Kittiwake Rissa tridactyla 41 11.1 Breeding numbers 41 11.2 Breeding success 42 11.2.1 Methods 42 11.2.2 Results 42 11.3 Timing of breeding 44 11.4 Adult survival 44 12 Common Guillemot Uria aalge 46

2 Seabird monitoring on Skomer Island in 2013

12.1 Breeding numbers - whole island counts 46 12.2 Breeding numbers – study plot counts 46 12.3 Breeding success 49 12.3.1 Methods 49 12.3.2 Results 49 12.4 Timing of breeding 50 12.5 Adult and juvenile survival 51 13 Razorbill Alca torda 52 13.1 Breeding numbers - whole island counts 52 13.2 Breeding numbers - study plot counts 52 13.3 Breeding success 54 13.4 Timing of breeding 56 13.5 Adult survival 57 14 Atlantic Puffin Fratercula arctica 58 14.1 Breeding numbers 58 14.2 Puffin burrow occupancy and breeding success on Skomer in 2013 59 14.3 Feeding rates 61 14.4 Adult survival 62 16 References 63 17 Appendices 65 Appendix 1 Mean seabird counts by section 65 Appendix 2 Dates of visits to Black legged kittiwake sub-colonies in 2013 67 Appendix 3 Guillemot and Razorbill Study Plots 68 Appendix 4 Maximum spring Atlantic Puffin counts on Skomer Island, including Middleholm, 2013 69 Appendix 5 Mean annual estimated survival rates of seabirds on Skomer Island 70 Appendix 6 Ringing Totals for 2013 71 Appendix 7 Skomer Island Guillemot Study 2013 72 Appendix 8 Gull Diet Survey Comparison of Methods 74

3 List of figures

Figure 1 Northern Fulmar breeding numbers on Skomer Island 1963-2013 13 Figure 2 Northern Fulmar breeding success on Skomer Island 1986-1987 and 1992-2013 15 Figure 3 Trend of annual survival estimate of Manx Shearwater from 1997-2013 23 Figure 4 Great cormorant breeding numbers on Skomer Island 1960-2013 24 Figure 5 European Shag breeding numbers on Skomer Island 1960-2013 25 Figure 6 Lesser Black-backed Gull breeding numbers on Skomer Island 1961-2013 26 Figure 7 % of empty Lesser Black-backed Gull nests 1991-2013 30 Figure 8 Estimated No. of fledgling Lesser Black-backed Gulls per AON 1981-2013 32 Figure 9 Trend of annual survival estimate of Lesser Black-backed Gull 1978-2013 33 Figure 10 Herring Gull: Number of AONs on Skomer Island 1961-2012 34 Figure 11 Breeding success of coast-nesting Herring Gulls on Skomer Island, 1962-2012 35 Figure 12 Trend of annual survival estimate of Herring Gull 1978 to 2013 36 Figure 13 Great Black-backed Gull breeding numbers on Skomer Island 1960-2013 37 Figure 14 Great Black-backed Gull breeding success on Skomer Island 1996-2013 38 Figure 15 Great Blacked-backed Gull diet remains from 25 nests on Skomer Island 2013 39 Figure 16 Frequency of occurrence of food items within 5m of 25 Great Black-backed Gull nests in 2013 40 Figure 17 Mean number of shearwater carcasses found at Greater Black-backed Gull nests in 1959, 1965, 1973, 1992, 2008, 2009, 2010, 2011, 2012 and 2013 40 Figure 18 Black-legged Kittiwake breeding numbers on Skomer Island 1960-2013 41 Figure 19 Black-legged Kittiwake breeding success on Skomer Island 1989-2013 43 Figure 20 Trend of annual survival estimate of Black-legged Kittiwake from 1978 up to and including survival in 2013 45 Figure 21 Common Guillemot breeding numbers on Skomer Island 1963-2013 48 Figure 22 Common Guillemot breeding success on Skomer Island 1989-2013 50 Figure 23 Razorbill breeding numbers on Skomer Island 1963-2013 54 Figure 24 Razorbill breeding success on Skomer Island 1993-2013 56 Figure 25 Trend of adult survival of Razorbill up to and including survival from 1979-2013 57 Figure 26 Maximum spring counts of Atlantic Puffin on Skomer Island 1989-2013 59 Figure 27 Number of feeds per hour for Puffins on Skomer Island in 2013 61 Figure 28 Trend of annual survival estimate of Atlantic Puffin 1972 up to and including survival from 2013 62

4 Seabird monitoring on Skomer Island in 2013

List of Tables

Table 1 Northern Fulmar whole island count details 2004-2013 12 Table 2 Northern Fulmar breeding success on Skomer Island 2013 14 Table 3 Northern Fulmar breeding success on Skomer Island 2006-2013 14 Table 4 Northern fulmar timing of breeding on Skomer Island 2006-2013 15 Table 5 Number of burrows recorded from Manx Shearwater study plots 1998–2012 19 Table 6 Number of responses from Manx Shearwater study plots 1998–2012 20 Table 7 Manx Shearwater breeding success in The Isthmus study plot in 2013 21 Table 8 Breeding success of Manx Shearwaters 1995-2013 22 Table 9 Record of Lesser Black-backed Gull Cane counting area 27 Table 10 Lesser Black-backed Gull actual nest count, total in cane counted areas and cane:eye count ratio 28 Table 11 Mean eye count totals of Lesser Black-backed Gull (AOT) in sub colonies 29 Table 12 Percentage of empty Lesser Black-backed Gull nests counted in May 2013 30 Table 13 Lesser Black-backed Gull empty nest ratios 1998–2013 30 Table 14 Ringed/resighting estimates for Lesser Black-back Backed Gull fledglings 31 Table 15 Productivity figures for Lesser Black-backed Gull in 2013 31 Table 16 Black-legged Kittiwake whole island count details Skomer 2007-2013 41 Table 17 Black-legged Kittiwake breeding success (per AON) on Skomer Island- 2013 42 Table 18 Black-legged Kittiwake breeding success (per AON) on Skomer Island 1989-2013 43 Table 19 Black-legged Kittiwake fledging success in relation to the number of chicks hatched per nest on Skomer Island 2012 44 Table 20 Black-legged Kittiwake - timing of breeding on Skomer Island 2007-2013 44 Table 21 Common Guillemot whole island counts on Skomer Island, 2004-2013 46 Table 22 Common Guillemot study plot totals on Skomer Island 2009-2013 47 Table 23 Common Guillemot breeding success (per active and regular sites) on Skomer Island 1989- 2013 49 Table 24 Common Guillemot breeding success (per active site and per active and regular site) on Skomer Island 2013 50 Table 25 Common Guillemot timing of breeding 2007-2013 50 Table 26 Razorbill whole island count details on Skomer Island, 2006-2013 52 Table 27 Razorbill study plot totals on Skomer Island 2009-2013 53 Table 28 Razorbill breeding success on Skomer Island 1993-2013 55 Table 29 Razorbill breeding success (per active site and per active and regular site) on Skomer Island 2013 55 Table 30 Razorbill timing of breeding 2006–2013 56 Table 31 Maximum spring counts of individual Puffins on Skomer Island in 2013 58 Table 32 Maximum spring counts of Atlantic puffin on Skomer Island and Middleholm 1989 -2013 58 Table 33 Burrow occupancy and breeding success of Puffins on Skomer Island in 2013 59 Table 34 Feeding rates of Puffins on Skomer Island in 2013 61

5 Summary

This document reports upon the 2013 breeding season for seabirds on Skomer Island, drawing together the work of the wardens and various research institutions. This includes whole island population counts, study plot counts and estimates of breeding success and adult survival.

The table below summarises population counts for ten species in 2013 and makes comparison with the previous year as well as giving a five year percentage change. A continued decline is obvious in the Kittiwake population and it should be noted that the birds were so late in nesting this year that a count could not be performed on the first round of all island counts, giving only one set of counts for this species in 2013. The huge increase in the Puffin population (max. spring count of Individuals) may be due to changes in (and between) the timing of breeding and all island counts.

Whole island seabird population counts for 2013 compared with 2012 and giving a five year percentage change

Species and count units Totals for Totals for 2012 % Change 5 Year % Change 2013 from 2012 Fulmar (AOS) 503 453 +11.04 -4.55 Cormorant (AON) 7 4 +75.00 -22.22 Shag (AON) 5 5 - +66.66 Lesser Black Backed Gull (AON) 8132 8643 -5.91 -20.42 Herring Gull (AON) 221(?) 401 - - Greater Black Backed Gull (AON) 84 96 -12.50 -20.75 Black-legged Kittiwake (AON) 1045 1594 -34.44 -48.93 Guillemot (IND) 20862 22508 -7.31 +6.92 Razorbill (IND) 6663 4971 +34.04 +26.63 Puffin (IND) 19280 11497 +67.70 +42.73

Count units used in this report AOS-Apparently Occupied Site AON-Apparently Occupied Nest AOT-Apparently Occupied Territory IND-Individual

Fulmar numbers were up slightly from 453 in 2012 to 503 in 2013. Productivity, however, was poor, the figure (0.29) being under the values of the last three years and below the average for the last five years (0.34) and even lower (0.16) if the plot at Tom’s House, with a single successful pair, is excluded.

Unfortunately the shearwater study plot census that has been carried out by a team from Oxford University since 1998 was cancelled at short notice in 2013 so there is no population data for this year. After an extremely poor year in 2012, possibly caused by burrow flooding and food availability, productivity in 2013 was back up to just below the average for the last five years (0.63) at 0.60.

Lesser Black-backed Gull numbers continue to fall with a 5.91% decrease on the previous year and a -20.42% five year change. This is reflected in a similar decline on Skokholm which saw a 32.9% decrease on 2012. Productivity was 0.076 chicks per AON, a significant decrease on last year’s figure (0.5). Low productivity, combined with a low adult survival rate, combine to show a steady decline in the breeding population.

6 Seabird monitoring on Skomer Island in 2013

Kittiwake numbers (whole island) were 34.44% down on 2012, the birds were late nesting and a lack of synchrony was apparent. Productivity was 0.33 chicks per AON, not dissimilar to last year but still very low.

Guillemot numbers (whole island) were -7.31% down on last year and -2.5% down within the study plots. Breeding success was the same as last year with 0.63 chicks fledged per active and regularly occupied site. This is close to the five year mean (0.65) and overall mean (0.68 (1989-2013)).

Razorbills numbers (whole island) were up by 34% and 49% across the study plots on 2012. Productivity was 0.38 chicks fledged per active and regularly occupied site, considerably up on the previous year and back to ‘normal’ levels of breeding success. The low population count in 2012 was possibly linked to the low productivity in that year and caused by a high level of nest site abandonment by the time of counting.

Maximum spring numbers of individual Puffins was up by a massive 67.70% and is discussed in section 14.1. Burrow occupancy at the study plot on the neck was slightly up and productivity was slightly down but the figures remained fairly stable.

7 1 Introduction

Seabirds are a significant component of the marine environment and Britain has internationally important populations of several species. A recent census (Perrins et al. 2011) of the Manx Shearwater population on Skomer estimated 316,070 breeding pairs. This affords Britain’s (and Skomer’s) seabird populations even greater importance and probably makes Britain’s Manx Shearwater population(s) a higher proportion of the world population than is the case for any other bird species breeding in the Britain and Ireland. Skomer is the believed to hold the largest Manx Shearwater colony in the world. Other seabird species that breed on Skomer in important numbers include Fulmar, Lesser Black-backed Gull, Kittiwake, Common Guillemot, Razorbill and Puffin. A national Seabird Monitoring Programme, co-ordinated by the Joint Nature Conservation Committee (JNCC), includes a small number of "key site" seabird colonies where detailed monitoring of breeding success, annual survival rates and population trends is carried out. These sites are geographically spread to give as full coverage of British waters as possible.

Skomer Island is the most suitable site for this work in south-west Britain. It is a National Nature Reserve managed by The Wildlife Trust of South and West Wales (WTSWW) under a lease from Natural Resources Wales (NRW). Not only is Skomer the most important seabird colony in southern Britain, but the waters around the island have been designated a Marine Nature Reserve. Seabird monitoring fits within a broader framework of monitoring marine and terrestrial organisms on and around the island.

There is an impressive data set for seabirds on Skomer. This is especially important for species such as seabirds with long periods of immaturity and high adult survival rates. The Wildlife Trust has been monitoring seabirds on the island since the early 1960s. Additional detailed studies of particular species, annual adult survival rates, breeding success and other aspects of seabird ecology have been carried out for many years by other bodies.

In 2013, the whole island counts and study plot counts of Common Guillemot and Razorbill, the whole island counts of Northern Fulmar and all breeding gulls (including Kittiwake) and breeding success rates of Fulmar, Herring Gull, Great Black-backed Gull, Kittiwake and Common Guillemot were funded by JNCC. This work is carried out by the island Wardens and a contract Field Worker with additional help in some areas by the island Assistant Warden and volunteers. Lewis Yates was the Field Worker in 2013.

Lesser Black-backed Gulls cane count areas were altered in 2013 to build up a picture of correction factors across all colonies on the island. To be of use this must be continued into 2014. Census methods for estimating breeding numbers of Lesser Black-backed Gulls should be coming under further review in the near future.

A review of the Great Black-backed Gull diet study was carried out in 2012 and can be read in Appendix 7. The new tried and tested method was used in 2013 and can be read about in sec 10.3.

This report includes other seabird monitoring studies undertaken on Skomer by the Edward Grey Institute and Sheffield University.

8 Seabird monitoring on Skomer Island in 2013

1.1 Introduction to survival estimates

The survival rate analyses presented here have been made in the same way as in the other years since 1978, in that they have been made from a computer calculation of Maximum Likelihood Measurements (using the programme MARK) and only one measure of survival for each year is presented. As with similar methods, two years of observations are needed to obtain the estimate for a given year, i.e. the 1999 estimate can only be obtained after observations in 2000 and 2001. Hence the survival figure for the last year (2012-13) is not comparable with the others and is not presented. The estimates for other more recent years are likely to change (hopefully not much) with the addition of further years of data. Years for which the survival rates are not given are ones in which estimates cannot be made accurately; this is normally because they are very high and the confidence limits exceed 100.

The six graphs showing estimated survival rates of the species concerned are presented under each species account. A table listing the actual figures are given together in Appendix 4. Each graph has a line at 90% survival for ease of reading. For those species where a significant trend is apparent, the trend-line is also presented.

Survival rates were measured for a number of species on the island. Observations were made by David Boyle (EGI). We are grateful to Matt Wood for helping with the survival analyses in this report.

2 General methods 2.1 Whole island counts

The Lesser Black-backed Gull Larus fuscus colonies were counted by eye from established vantage points (between the 8th and 10th of May). An attempt was then made to ground truth a sample of colonies (between the 13th and 15th of May) to produce a correction factor.

Whole island counts of the cliff nesting species were carried out in June (1st-29th June) and two complete counts were made.

In mid-June 1999, black-and-white photographs were taken of all study count sections and these are filed on the island. In 2013 new photographs of some of the sections were taken in order to update the existing ones, as vegetation and the cliffs themselves have changed over the years.

Count units (explained under summary) and methods follow those recommended by Walsh et al (1995) but note that the Lesser Black-backed Gull census methodology has been developed on the island (see Sutcliffe 1993).

Graphs showing whole island populations since the 1960s are presented for each species. Note that in past years different counting units and methods have been used for some species, although those in recent years have been standardised. General trends can nonetheless be identified with some confidence.

9 2.2 Study plot counts of Common Guillemots Uria aalge and Razorbills Alca torda

Counts were made during the first two weeks of June of the same study plots used in previous years, using methods outlined in Walsh et al. (1995). In mid-June 1999, black-and-white photographs were taken of all study plot sitesand these are filed on the island. In 2013 new photographs of some of the study plots were taken in order to update the existing ones, as vegetation and the cliffs themselves have changed over the years.

10 Seabird monitoring on Skomer Island in 2013

2.3 Breeding success

Methodology follows that of Walsh et al. (1995). Brief details are given separately in each species account. Black-and-white photographs of the breeding success plots were taken in mid June 1999 and are filed on the island. In 2013 new photographs of some of breeding success plots were taken in order to update the existing ones, as vegetation and the cliffs themselves have change over the years.

2.4 Weather

Generally, a cold spring and rather dry and settled summer, in stark contrast to last year’s wet summer.

March – The coldest March for 50 years. Many nights with temperatures under 0°C. Wind mostly from east. Rain on eleven days and snow flurries on two. April – Warmer than March but still many days with strong northerly wind. Rain on eleven days and thick mist and fog on two. May – A rather cold month with max. temperatures around 16°C and min. temperatures around 5°C. Rain on 13 days. Gale force 8 from the south west on the 9th. June – Mostly settled and generally fine. Wind force 5 or 6 on nine days. Rain on eight days. July – Warm and dry with the most settled weather patterns for three decades. August – A rather mild, dry month with only five or six days with rain.

Most birds were approximately two weeks late in breeding. This was most obvious in Puffins, whose breeding season had become progressively earlier in recent years. Shags and Guillemots also reportedly had their latest breeding seasons since records began on timing of breeding. The delay in breeding in 2013 may have been caused by lower than usual temperatures in spring.

11 3 Northern Fulmar Fulmaris glacialis 3.1 Breeding numbers - whole island counts

Two all island counts for Fulmars were conducted during June. The mean of the two counts was 503 Apparently Occupied Sites (AOS) with a range of 477 and 530. Study plot counts and whole island counts suggest a decline in population since 1990. However, since 1998 there has been a gradual increase in productivity, suggesting the population may have stabilised with regards to food availability and competition for both nest sites and food.

On the 26th of August 2012 a serious rock fall occurred at the Isthmus near the wardens house and within the North Haven study plot. Some Fulmars were killed in the incident and it was feared that some nest sites were lost. However the number of sites occupied in the North Haven study plot was actually up from 38 to 43 in 2013.

Table 1 Northern Fulmar whole island count details 2004-2013

Year Total % 5 Year % Change Change 2004 730 +15.1 +5.6 2005 726 -0.5 -0.5 2006 595 -18.0 -6.3 2007 611 +2.7 -3.6 2008 565 -7.5 -22.6 2009 527 -6.7 -27.4 2010 530 +0.6 -10.92 2011 474 -10.57 -22.42 2012 453 -4.43 -19.82 2013 503 +11.04 -4.55

12 Seabird monitoring on Skomer Island in 2013

Figure 1 Northern Fulmar breeding numbers on Skomer Island 1963-2013 800

700 Whole Island Count 600 (mean) Study Plot Count 500

400

No of AOS of No 300

200

100

0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012

3.2 Breeding success

3.2.1 Methods

Three visits were made to the seven fulmar study plots on 28th May, 31st May and 5th June to observe site occupancy. Sites were considered occupied if a bird appeared to be incubating or an egg was seen on two consecutive visits. A further visit was made on the 9th August to determine the presence or absence of large chicks on the sites. All large chicks were assumed to have fledged.

3.2.2 Results

147 AOSs were identified in late May. The overall breeding success was 0.29, slightly lower than the last three year’s productivity values and remaining below the average for the period 1986 to 2011 (0.48). However if we exclude the Tom’s House site, which included a single successful pair, this gap increases with the mean dropping to 0.16 (Table 2). Matthews Wick remains the most successful site and general patterns of productivity seem to have persisted across the study plots.

13 Table 2 Northern Fulmar breeding success on Skomer Island 2013

No. site No. sites Chicks Breeding monitored occupied fledged success Tom’s House 11 1 1 1.00 Basin (West) 55 33 9 0.27 Basin (East) 32 17 7 0.41 North Haven 98 43 11 0.26 South Haven 32 17 4 0.24 Castle Bay 30 15 5 0.33 Matthew’s Wick 40 20 11 0.55 Mean 0.29 SD 0.27 SE 0.09

NB: if the small sample of Tom’s House is excluded the mean becomes 0.16 (SD 0.17, SE 0.06)

Table 3 Northern Fulmar breeding success on Skomer Island 2006-2013

2007 2008 2009 2010 2011 2012 2013

Tom’s House 0.00 0.00 0.25 0.43 0.00 0.00 1.00 Basin (West) 0.38 0.21 0.32 0.38 0.38 0.31 0.27 Basin (East) 0.36 0.25 0.33 0.64 0.43 0.43 0.41 North Haven 0.26 0.37 0.34 0.39 0.25 0.42 0.26 South Haven 0.5 0.32 0.11 0.24 0.23 0.33 0.24 Castle Bay 0.43 0.38 0.25 0.33 0.44 0.35 0.33 Matthew’s Wick 0.5 0.29 0.32 0.45 0.56 0.48 0.55 Mean 0.35 0.26 0.27 0.41 0.33 0.38 0.29 SE 0.07 0.05 0.03 0.05 0.06 0.06 0.09

14 Seabird monitoring on Skomer Island in 2013

Figure 2 Northern Fulmar breeding success on Skomer Island 1986-87 and 1992-2013

0.80

0.70

0.60

0.50

0.40

0.30

Chicks 'fledged' perAOS 0.20

0.10

0.00 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

3.3 Timing of breeding

Time of breeding was recorded in 2013 and is detailed in Table 4, however the first egg was seen relatively late on so is not thought to be indicative of this year’s timings. The first chick was seen much earlier than in previous years but this may not be indicative of general timings.

Table 4 Northern fulmar timing of breeding on Skomer Island 2006-2013

2006 2007 2008 2009 2010 2012 2013

First egg 1st June Not recorded 23rd May 20th May 22nd May 20th May 3rd June First chick 1st July Not recorded 14th July 8th July 6th July 13th July 10th June

15 4 European Storm-petrel Hydrobates pelagicus 4.1 Breeding numbers

Attempts at Storm Petrel survey work were carried out in three locations in 2013 and it is perhaps easiest to split this up into subheadings for each site as the work carried out was so different at each site.

North Haven An attempt was made to census the storm petrels at North Haven using the standard method (Fowler 2001, Brown 2006), but logistical constraints meant that only four visits could be made to the colony. Calls in response to playback were noted from nine burrows (sample size not recorded) which is likely to be an under-estimate of the number of breeding pairs here.

Mew Stone Visit(s) made by Matt Wood with an infrared camera to experiment on getting footage of birds. This early stage experimental work was to ascertain the value of infrared footage for long term population monitoring of Storm Petrels on Skomer Island. Visit(s) weremade by Lewis Yates, Will Whittington and Nia Stephens in an attempt to record calls from burrows.

Tom’s House Ringing adults for survival using standardised net arrangement and without using tape lure. Dates and results will be reported on in a future report.

A paper recently published by Soanes (2012) compared the response rates of different and manipulated recordings. Then the paper compared different estimation methods for censusing (“Standard” and the “du Feu”).

Estimate Number of AOS (95% confidence interval) Method Standard de Fue Method Brown (2003) (Year of Method (2008) (2008) census) Whole 101 64 203 Island (91-130) (55-79) (178-239)

The main aim of the study was to test the monitoring method and before using this figure as a population estimate there must be some investigation done into the extent of “searching” for burrows to ensure that this was a true whole island estimate.

16 Seabird monitoring on Skomer Island in 2013

References:

Brown, J.G. (2005) Storm Petrel census on Skomer Island in 2003. CCW Contract Science Report no 671.

Soanes, LM., Thomas, R.J. & Bolton, M. (2012) Evaluation of field and analytical methods for estimating the population size of burrow-nesting seabirds from playback surveys. Bird Study, 59(3), 353-357

17 5 Manx Shearwater Puffinus puffinus

5.1 Study plot census

Unfortunately the shearwater study plot census that has been carried out by a team from Oxford University since 1998 was cancelled at short notice in 2013 so there is no population data for this year. The MSc course that used to do the census has now finished. The census will, however, hopefully be picked up again next year by two students helped by island staff and volunteers.

Previous year’s data (1998-2012) are shown here (Tables 5 and 6) for completeness, but note there is no data for 2013.

18 Table 5 Number of burrows recorded from Manx Shearwater study plots 1998-2012

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 max min A 51 70 87 94 98 145 87 35 105 62 91 61 87 69 100 145 35 B 75 102 193 240 98 91 78 81 74 108 49 91 53 74 176 240 49 C 299 255 259 202 193 332 287 262 309 387 346 236 246 385 358 387 193 D 200 235 296 244 320 313 98 210 253 303 204 206 201 238 316 320 98 E 63 65 66 67 61 58 48 37 49 38 48 32 46 40 42 67 32 F 14 17 12 11 17 20 15 18 15 13 13 12 17 17 15 20 11 G 11 16 15 14 22 21 14 22 29 19 34 25 19 28 21 34 11 H 98 97 120 120 140 126 88 118 85 167 84 87 89 141 110 167 84 I 271 293 199 321 260 309 236 389 230 331 246 465 278 437 442 465 199 J 339 311 455 401 360 359 305 224 219 337 407 315 275 351 438 455 219 L 473 506 596 560 593 661 527 693 445 709 472 604 422 560 716 7016 422 M 234 231 240 188 175 218 167 141 168 154 152 191 157 213 212 240 141 N 207 249 261 288 248 261 221 252 282 214 235 215 221 222 226 288 207 O 93 99 140 152 110 142 278 119 125 156 139 84 185 148 246 278 84 P 151 205 234 204 228 270 124 283 264 257 254 303 256 329 319 319 124 Q 84 82 77 95 85 71 112 132 108 119 85 111 77 106 104 132 71 R 190 235 329 236 214 314 278 276 279 197 158 167 189 287 214 329 158 S 97 187 127 237 213 274 241 244 286 344 260 311 248 209 260 344 97 TOTAL 2950 3255 3706 3674 3435 3985 3204 3536 3325 3915 3277 3516 3066 3854 4315

19 Table 6 Number of responses from Manx Shearwater study plots 1998–2012

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Max Min A 12 15 17 12 20 15 16 12 28 10 23 20 9 13 17 28 9 B 19 35 18 19 32 28 32 15 21 30 12 15 9 18 19 35 9 C 56 45 27 35 36 45 52 41 53 66 69 82 30 66 41 82 27 D 81 65 61 51 71 55 52 64 64 73 61 57 31 80 97 97 31 E 17 14 17 15 14 7 9 9 10 5 8 3 5 5 5 17 3 F 3 3 2 5 5 6 4 7 8 6 6 3 4 3 9 17 2 G 2 6 4 3 9 7 5 8 9 2 9 12 6 7 9 19 2 H 23 17 10 15 16 10 14 16 13 17 14 22 12 18 32 32 10 I 72 88 74 117 75 67 102 134 111 116 83 169 110 135 144 169 67 J 77 75 107 67 54 66 81 73 42 70 72 80 46 95 93 107 42 L 147 132 186 131 142 164 185 244 150 157 156 222 123 159 179 244 123 M 85 80 67 62 79 94 71 75 66 73 65 81 33 95 89 94 33 N 51 67 39 49 52 44 40 63 75 23 37 70 41 82 62 75 23 O 27 29 38 34 30 36 84 34 40 29 25 38 30 51 45 84 25 P 30 60 57 67 78 77 32 67 95 72 117 93 80 107 127 127 30 Q 34 26 17 17 29 26 32 32 32 31 20 65 20 25 28 65 17 R 48 44 65 39 56 83 91 92 72 65 62 53 65 79 65 92 39 S 37 67 45 51 63 75 63 65 55 73 69 96 87 75 56 96 37 TOTAL 821 868 851 789 861 905 965 1052 944 918 908 1181 767 1113 1117 Notes: 1) There is no plot K 2) Same tape used as in previous years

20 5.2 Breeding success

After an extremely poor year in 2012, possibly caused by burrow flooding and poor food availability, productivity in 2013 was back up to just below the average for the last five years (0.63) at 0.60.

Manx Shearwater breeding success in The Isthmus study plot in 2013 is detailed in table 7; table 8 shows the breeding success since 1995.

Table 7 Manx Shearwater breeding success in The Isthmus study plot in 2013

Total Number of eggs laid 87 Number of eggs known or assumed to have failed1 23 Number of eggs known or assumed to have hatched2 64 Number of chicks known or assumed to have died3 12 Number of chicks surviving to ringing age 52

Hatching success4 74% Fledging success5 81% Number of fledged young per egg laid 0.6 Notes: 1 Thirteen eggs are known to have failed, having been found abandoned or broken or having disappeared before they could have possibly hatched. A further ten pairs are assumed to have failed at the egg stage, the burrows being completely empty when checked in July, with no sign of eggshell or down in the nest chamber suggestive of the egg having been kicked out of the chamber. 2 Fifty-eight chicks were found; in six burrows only hatched eggshell and/or down was found suggesting the chicks had actually hatched and the burrow had not failed at the egg stage. 3 Six chicks were only ever recorded as small amounts of down (mentioned above), four chicks were found dead and two were recorded as small chicks but had disappeared when it came time to ring them and are assumed to have died. 4 Hatching success = % of eggs known or assumed to have hatched 5 Fledging success = % of chicks surviving to a large size

21 Table 8 Breeding success of Manx Shearwaters 1995-2013

No. No. %Hatched %Eggs (laid) No. Laid Hatched Fledged Fledged Fledged 1995 82 58 56 0.97 0.68 1996 75 57 55 0.96 0.73 1997 82 58 52 0.90 0.63 1998 71 61 54 0.89 0.76 1999 84 58 56 0.97 0.67 2000 83 59 51 0.86 0.61 2001 98 62 42 0.68 0.43 2002 108 73 62 0.85 0.57 2003 104 78 73 0.94 0.70 2004 97 61 51 0.84 0.53 2005 122 82 68 0.83 0.56 2006 110 64 52 0.81 0.47 2007 108 71 55 0.77 0.51 2008 99 73 67 0.92 0.68 2009 114 73 67 0.92 0.59 2010 115 80 70 0.88 0.70 2011 99 68 61 0.90 0.69 2012 91 50 37 0.40 0.55 2013 87 64 52 0.81 0.60

5.3 Adult survival

The shearwater survival estimates are based on birds that are marked in burrows on the Isthmus. All but a few of the nests are reached every year and the majority of the birds breeding in them are caught. In recent seasons, night searches for adults in the vicinity have turned up a few "missing" birds - birds that had survived, but were not breeding in the study burrows; presumably they were living nearby. Appendix 4 gives the estimated adult survival rates of Manx Shearwaters. As reported previously, these remain low, both in comparison with more detailed studies carried out earlier on Skokholm and with what one might expect for a bird with such a low reproductive rate, but there are no discernible long-term changes.

Figure 3 shows that survival for Manx Shearwater averages 0.87 with no significant time trends.

22 Seabird monitoring on Skomer Island in 2013

Figure 3 Trend of annual survival estimate of Manx Shearwater from 1997-2011

23 6 Great Cormorant Phalacrocorax carbo 6.1 Breeding numbers

A steady decline since 2008, resulting in no nests being observed in 2011, has been reversed and a steady increase is now apparent with four AON in 2012 and seven AON in 2013. All of these were on the southern face of the Mew Stone.

Figure 4 Great cormorant breeding numbers on Skomer Island 1960-2013 30

25

20

15 No. of AON 10

5

0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012

6.2 Breeding success

It is not known how many chicks fledged from Skomer in 2013. The nests were on the southern face of the Mew Stone and it was not possible to go out on a boat and count the fledglings.

24 Seabird monitoring on Skomer Island in 2013

7 European Shag Phalacrocorax aristotelis 7.1 Breeding numbers

Five pairs were seen nesting on the north coast of Skomer near the Garland Stone, the same number of pairs as recorded in 2012.

Figure 5 European Shag breeding numbers on Skomer Island 1960-2013 25

20

15

No. of AONs 10

5

0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012

7.2 Breeding success

Seven chicks fledged from five nests opposite the Garland Stone giving a productivity figure of 1.4 chicks per AON.

Shags breeding on Middleholm were monitored by the South Pembrokeshire Ringing Group and two visits were made on the 29th of June and 20th of July to monitor nests and ring as many chicks as possible. Not only was it a very late and protracted breeding season for Shags on Middleholm but also the worst on record for total numbers and breeding success since ringing began there in the early 1980’s. Thirty-six nests produced a maximum of 44 chicks giving a breeding success of 1.22 chicks per AON.

25 8 Lesser Black-backed Gull Larus fuscus 8.1 Methods for estimating breeding numbers

The Lesser Black-backed Gull Larus fuscus colonies were counted by eye from established vantage points between the 8th and 10th of May. Mike Wallen, a volunteer, has been doing the counts for many years and, to keep the counts consistent, this was continued in 2013. It was however a rather late breeding season for the gulls and some sections were counted in adverse weather conditions, so several sections were recounted in late May (29th and 30th) at the recommendation of Mike.

An attempt was then made to ground truth a sample of colonies (between the 13th and 15th of May) to produce a correction factor. Nests, including empty nests, in selected sub-colonies were systematically searched for, counted and marked with canes. The difference between the counts from viewpoints and the actual nests counted by the cane method produced a mean adjustment factor that was applied to the mean of the counts from viewpoints for the whole island. The method assumes that each pair built one nest. Applying a single correction factor to the whole range of sub-colonies, which vary in habitat type and breeding density, may lead to errors, although it is difficult to overcome this.

Since 2012, at the request of JNCC, the areas selected for cane counting have been changed where possible. The aim is to build up a rolling picture of the correction factors for specific sections over the course of several years. Only four sub-colonies were cane counted in 2013 (1,7,Y and Z) one of which (Y) was a standard area (Table 9), and thus not a new area. The standard analysis method was applied to all cane counting areas in 2013. The benefits of this are that the same areas are not disturbed each year and a correction factor for each colony on the island will be obtained over time.

Census methods for estimating breeding numbers of Lesser Black-backed Gulls should be coming under further review in the near future.

Figure 6 Lesser Black-backed Gull breeding numbers on Skomer Island 1961-2013

25000 Eye count

20000 Eye count + correction factor 15000

10000 No. of AON

5000

0 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012

26 Seabird monitoring on Skomer Island in 2013

Table 9 Record of Lesser Black-backed Gull Cane counting area

standard 2012 2013 areas 1 South Old Wall 2 Marble Rocks 3 Abyssinia + 24 4 Anvil Rock 5 Bull Hole 6 Pyramid Rock 7 North Plain 8 Sheer Face West 9 Sheer Face East 10 The Hill 11 Double Cliff 12 North slopes 13 North Valley Rise 14 Green Plain 15 South Neck - Thorn Rock 16 W/S Field 17 Saunders Fist 18 Harold Stone 19 Wick Cliff 20 Tom's House-Skomer Head 21 colony now joined with X 22 Garland Stone 23 North West Neck 24 East of West Pond – see 3 25 Toms House to Wick 26 Mew Stone A Lantern B Neck East C Neck main ridge D South Castle E Neck South West Coast F South Haven G South Stream Cliff H Welsh Way I High Cliff J South Wick Ridge K Wick L Welsh Way Ridge M Wick Ridge North N Wick Ridge North O Moorey Meadow P South Stream Q Bramble R Lower Shearing Hays S New Park T Shearing Hays U Captain Kites V Wick Basin W The Basin X / 21 (see 21) Y Field 11 Z Basin-South Pond

27 8.2 Breeding numbers – results

A mean total of 3836 Apparently Occupied Territories (AOT) were counted by eye from various standardised viewpoints around the island (Table 10). This count is then corrected for by nest searches in a selected sample of sub-colonies.

The number of nests, including empty (but active) nests, in selected sub-colonies were systematically counted physically using canes (Table 10) The mean ratio of the actual number of nests to “eye-counts” was 2.32. When multiplied by the eye-count this gives a population estimate of 8132 breeding pairs (Figure 6), which is 5.91% down on 2012. There is therefore a decrease of 20.42% over a five year period and a 59.74% decrease since 1993 when the population was at its peak.

This decline is mirrored on the neighbouring island of Skokholm which showed a 32.9% decrease on the previous year. Moreover, after a period of relative stability in the early 2000s the national trend appears to be one of decline again. Poor adult survival is implicated as one of the drivers of this decline but it is not known how many of these ‘missing’ birds die over the winter and how many simply move to another colony.

Tables 12 and 13 as well as figure 7 show an apparent increase of empty nests on previous records for the fourth year in a row.

Table 10 Lesser Black-backed Gull actual nest count, total in cane counted areas and cane:eye count ratio

Cane 2nd Cane Eye Sub-colony Count Count ** TOTAL count Ratio 1 South Old Wall 237 237 66 3.59 7 North Plain 747 747 333 2.24 Y Field 11 259 259 181 1.43 Z Basin-South 339 339 168 2.02 Pond Mean 2.32 SD 0.91

*Since 2011 a slightly different method of cane counting was adopted in order to reduce disturbance due to fragile ground caused by a dry summer and high rabbit density. This method, used on Skokholm Island, involved just one pass of each area by a group which was slightly closer together but moving at a slower pace and communicating with each other to ensure nests were not missed or double counted. The benefit being the terrain is only being passed over once, thus reducing burrow collapse. This also allows the gulls to return to their nests more quickly.

28 Seabird monitoring on Skomer Island in 2013

Table 11 Mean eye count totals of Lesser Black-backed Gull (AOT) in sub colonies

Sub-colony 2013 1 South Old Wall 66 2 Marble Rocks 36 3 Abyssinia + 24 81 4 Anvil Rock 83 5 Bull Hole 61 6 Pyramid Rock 33 7 North Plain 333 8,9,10 Sheer Face 146 11 Double Cliff 24 12 North Slopes 49 13 North Valley Rise 299 14 Green Plain 423 16 W/S Field 11 18 Harold Stone 2 19 Wick Cliff 3 20 Tom's House-Skomer Head 4 21 Colony now joined with X 90 22 Garland Stone 11 23 Neck North West 58 25 Toms House to Wick 0 B Neck East 59 C Neck Main Ridge 101 D South Castle 104 E Neck South West Coast 29 F South Haven 74 G South Stream Cliff 58 H Welsh Way 46 I High Cliff 62 J South Wick Ridge 54 L Welsh Way Ridge 80 M North Wick Ridges 356 O Moorey Meadow 129 P South Stream 74 Q Bramble 23 R Lower Shearing Hays 177 S New Park 96 T Shearing Hays 46 U Captain Kites 69 V Wick Basin 0 W The Basin 64 Y Field 11 181 Z Basin-South Pond 168 Extra coastal 0 TOTAL 3836

29 Table 12 Percentage of empty Lesser Black-backed Gull nests counted in May 2013

Empty Sub-colony TOTAL total % empty 1 South Old Wall 237 119 50.21 7 North Plain 747 248 33.20 Y Field 11 259 104 40.15 Basin-South Z Pond 339 137 40.41 Mean 40.99

Table 13 Lesser Black-backed Gull empty nest ratios 1998–2013

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2001 % Empty 19 26 39 28 40 49 23 14 20 24 16 31 nests

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

36 19 19 19 28 19 10.8 22.5 25.3 28 41

Figure 7 % of empty Lesser Black-backed Gull nests 1991–2013

60

50

40

30 % emptynests 20

10

0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

30 Seabird monitoring on Skomer Island in 2013

8.3 Breeding success

In 2013 the Lesser Black-backed Gulls’ had another year of very low breeding success. We make an estimate of the total numbers by using a simple capture:recapture technique. To do this we ring some of the large fledglings and then count the ringed:unringed ratio when most of the chicks have fledged. Our normal target is to ring at least 300 large chicks. In a few years it has been difficult to find this number. In 2013 it was clear early on that productivity would be low because it was very difficult to find large chicks; only 59 were ringed. The ringed/resighting estimates based on these are shown in Table 14 and the productivity in Table 15. As before, the persistent low productivity, combined in many years with a low adult survival rate, should combine to show a steady decline in the breeding population.

Table 14 Ringed/resighting estimates for Lesser Black-back Backed Gull fledglings

DATE No. RINGED No. UNRINGED TOTAL No. Est. No. OF FLEDGLINGS FLEDGLINGS FLEDGLINGS FLEDGLINGS SEEN SEEN SEEN 29/07 10 88 98 578 01/08 14 98 112 472 03/08 12 112 124 610 06/08 8 86 94 693 09/08 10 106 116 684 Mean 11 97 108 579

Est. No. of fledglings = total No. fledglings seen x No. ringed (59) / No. ringed fledglings seen

Table 15 Productivity figures for Lesser Black-backed Gull in 2013

Est No. of Whole Island Not including The fledglings Productivity Neck (Total AON = 8132) (Total AON = 7600) Max Productivity 693 0.085 0.091 Min. Productivity 472 0.058 0.062 Mean Productivity 579 0.071 0.076

* The figure of 8132 is provisional and not as robust as in previous years

Productivity = Est. No fledglings seen / Total Number of Apparently Occupied Nests

31 Figure 8 Estimated No. of fledgling Lesser Black-backed Gulls per AON 1981-2013

1.40

1.20

1.00

0.80

0.60

0.40

0.20 Estimated of fledglings No. per AON 0.00

32 Seabird monitoring on Skomer Island in 2013

8.4 Adult survival

Appendix 4 gives the estimated survival rates. These birds are all from the study area in compartment 8. Previously, it has been noted that there has been a decline in the breeding population in recent years, presumably due to the very poor breeding success Overall survival has averaged 0.88, but there has been a highly significant decline with time with the result that survival currently is much the same as Herring Gull.

Figure 9 Trend of annual survival estimate of Lesser Black-backed Gull 1978-2011

33 9 Herring Gull Larus argentatus 9.1 Breeding numbers

No accurate population figures are available for 2013. Two hundred and twenty one Apparently Occupied Nests (AON) were picked up on the Lesser Black-backed Gull census and whole island counts but this is almost certainly a gross under estimate of the population and this figure is not added to the data sets presented here. There will be a full census carried out in 2014.

Skomer’s Herring Gulls fell into heavy decline in the 1980’s but have stabilised at a lower level since then. Data up to 2012 is shown for completeness.

Figure 10 Herring Gull: Number of AONs on Skomer Island 1961-2012 3500

3000

2500

2000

1500 No. of AONs 1000

500

0 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011

34 Seabird monitoring on Skomer Island in 2013

9.2 Breeding success

There is currently no breeding success data for 2013 available for this species. The study site at Tom’s House was visited in May to identify Apparently Occupied Nests and then later to check for large chicks but the data is currently not available. Data up to 2012 is shown for completeness. The study site at Tom’s House will be monitored for breeding success in 2014.

Figure 11 Breeding success of coast-nesting herring gulls on Skomer Island, 1962-2012 1.4

1.2

1

0.8

0.6

0.4 No. of large chicks perAON 0.2

0 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012

35 9.3 Adult survival

Appendix 4 gives the estimated survival rates. These were originally based mainly on birds nesting along the north coast, but because the population dropped so markedly, we had to open up a second study plot in the area from the Amos to Skomer Head. The samples are still smaller than desirable.

As reported before, the figures for this species start off with more or less normal survival rates for such a gull, but after the first two years, the survival rates plummeted in 1981-1983, were almost up to what might be expected for a normal Herring Gull population in four of the five years 1984-1988.Overall survival rate averages at 0.83, but there is no long-term trend. Survival in the last two years (2009-10 [0.69], 2010-11[0.76]), and 2011-2012 [0.74], has been particularly poor.

Figure 12 Trend of annual survival estimate of Herring Gull 1978-2011

36 Seabird monitoring on Skomer Island in 2013

10 Great Black-backed Gull Larus marinus 10.1 Breeding numbers

In 2013 the breeding population of Great Black-backed Gulls (GBBGU) on Skomer was 84 Apparently Occupied Nests (AONs). This is a drop in numbers on 2012 and the second lowest count since 2003. It is also lower than the ten year average.

The decline since the 1960s has been attributed largely to control measures in the 1960s and 1970s that were implemented as a result of the species perceived predatory impact on other seabirds. An outbreak of botulism in the early 1980s also contributed to the decline (Sutcliffe 1997).

The national trend has shown a slow decline since 1999. Thus, although the Skomer population has fluctuated in recent years, the current situation suggests that the population may have found a natural level given competition for food (see Figure 13).

Figure 13 Great Black-backed Gull breeding numbers on Skomer Island 1960-2013

300 Whole island count 250

200

150 R² = 0.9085 No. of AONs 100

50

0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012

37 10.2 Breeding success

Monitoring of the breeding success of the GBBGU has been included in the JNCC contract since 1999.

Twenty five AONs were identified between 29th of April and 20th of May. A total of 39 large chicks were recorded between 1st June and 25th July, giving a productivity of 1.56. This is a significant increase in breeding success compared to the last few years (Fig 14). It is also the second highest productivity figure since records began in 1996 and the highest since 2004.

Figure 14 Great Black-backed Gull breeding success on Skomer Island 1996-2013

1.80

1.60

1.40

1.20

1.00

0.80

0.60

0.40

Chicks fledged per AON studied Productivity 0.20

0.00

10.3 Diet study

A trial study to monitor the diet of GBBGUs was initiated in 2008 and continued in 2013. The prey remains around a sample of 25 nests were recorded. This sample represented nests from differing habitats and from areas of differing shearwater densities. The survey was carried out after chicks fledged (early August).

38 Seabird monitoring on Skomer Island in 2013

Figure 15 Great Blacked-backed Gull diet remains from 25 nests on Skomer Island 2013

GBBGU Prey Percentages 2013 Veg. Pellet, 3.34 Crustacean, 3.22 Invert Pellet, 0.00 Manx Shearwater, 19.55 Feather Pellet, 21.53 Fish (inc pellets), 5.69 Refuse, 5.32

Rabbit, 9.65

Fur Pellet, 17.33 Bones (other), Bird (other), 5.94 8.42

Following a review of the diet study methodology in 2012 (Kipling, 2012) a modified method was used to standardise results in 2013. A radius of five metres was shown to contain the majority of food items so the search was restricted to this area, with a note taken of Manx Shearwater carcasses outside this range for the comparison of shearwater carcasses per nest between years. This change should have a minimal impact on the results of a previously vague methodology and should improve reliability of the data but the change should be considered when comparing the whole range of historical data.

Roughly 20% of the prey items recorded were Manx Shearwaters (Fig 15), compared to 13% in 2012.

Manx Shearwater remains were recorded at 96% of the nests studied (Fig 16). Fur pellets were the most prevalent prey items, being found at 100% of the nests. Refuse was found at 76% of the nests, compared to 72% in 2012. Other birds were found at 60% of nests and included a majority of Puffin and one Lesser Black Backed Gull (excluding chicks which did not survive to fledging).

In 2013a total of 266 Shearwater carcasses were found at the sample of 25 nest sites, giving a mean of 10.64 carcasses per nest (Fig 17). This is the highest rate recorded during this study, with previous highs being just below ten carcasses per nest.

39 Figure 16 Frequency of occurrence of food items within 5m of 25 Great Black-backed Gull nests in 2013 100 90 80 70 60 50 40 30 20 Freqencyof occurrence at nest site 10 0 Rabbit Refuse Fur Pellet Veg Pellet Crustacean Bird (other) Invert Pellet Bones (other) Feather Pellet Fish (inc pellets) Manx Shearwater

Figure 17 Mean number of Shearwater carcasses found at Greater Black-backed Gull nests in 1959, 1965, 1973, 1992, 2008, 2009, 2010, 2011, 2012 and 2013

12

10

8

6 GBBGU nest

4

2 No. of ManxShearwater carcasses per

0 1959 1965 1973 1992 2008 2009 2010 2011 2012 2013

40 Seabird monitoring on Skomer Island in 2013

11 Black-legged Kittiwake Rissa tridactyla 11.1 Breeding numbers

Kittiwake numbers continue to decline sharply and are at their lowest historical level. A mean of 1045 Apparently Occupied Nests (AONs) were counted in 2013. This is a decline of 34.44% on the previous year and a -48.93% five year change. Only one whole island count was possible as most birds were late to start nest building. A lack of synchrony was also apparent. This meant that although some birds laid eggs at roughly the normal time, many birds laid much later, making the breeding season protracted and unsynchronised. Moreover, at the time of the first round of whole island counts (early June) it was obvious that the majority of breeding birds were yet to lay, or even complete a nest, and that counting the breeding Kittiwakes would have to wait until the second round of counts in mid to late June.

Figure 18 Black-legged Kittiwake breeding numbers on Skomer Island 1960-2013

3000

2500

2000

1500 No. of AONs 1000

500

0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012

Table 16 Black-legged Kittiwake whole island count details Skomer 2007-2013

Total % change 5 year % change 2007 1942 -6 -16.1 2008 2282 +17.5 +45.4 2009 2046 -10.3 -10.3 2010 1992 -6.06 -7.01 2011 1837 -4.02 -5.41 2012 1594 -13.23 -30.15 2013 1045 -34.44 -48.93

41 11.2 Breeding success

11.2.1 Methods

The breeding success of 394 kittiwake AONs was monitored at the same three sub-colonies studied since 1989 (but note some areas within the sub-colonies have been dropped since then) using the same methods as in previous years. Photographs of the cliffs were used and each nest marked on a transparent overlay. Visits were made to each sub-colony to monitor progress from nest construction to fledging. Due to the disrupted breeding patterns of the Kittiwakes this year, eight visits were made in total, covering the unusually extended and unsynchronised 2013 season (see appendix 1 for details). All chicks that were large (in class ‘d’ and ‘e’ in Walsh et al 1995) on the final visit were assumed to have fledged. Standard recording sheets from the Seabird Monitoring Handbook (Walsh et al 1995) were used for data collection.

11.2.2 Results

In 2013, 500 nests were started in the study areas, which is a reduction of 132 nests from 2012 which in itself was reduced by 86 nests from 2011. Study sites Wick 8 A+B contained no nests (there was one nest in these plots in 2011, none in 2012). Breeding success per nest was roughly level with last season but still below historical levels, and this masks the significant drop in absolute numbers

The 394 AONs produced a minimum of 294 chicks. Because of the difficulty of recording small chicks in some of the plots this is likely to be an underestimate. One hundred and sixty chicks survived to “large” status this year (compared to 194 in 2012) and were counted as fledged.

The lowest productivity was at South Stream with no chicks fledging, with High Cliff and The Wick both doing similarly at around one chick for every two apparently occupied nests. This is a reversal of recent seasons where South Stream had done relatively well.

In 2013, 79% of AONs went on to apparently incubate eggs (89% in 2012), with 73% of these producing chicks (41% in 2012). 21% of pairs did not complete nests (‘trace’ nests only) compared to 6% in 2012.

Table 17 Black-legged Kittiwake breeding success (per AON) on Skomer Island- 2013

Large Nests Incubating Nests Total (fledged) Breeding started AON'S Pairs w/chicks chicks chicks Success S.Stream 127 70 26 4 4 0 0.00 High Cliff 92 76 68 58 68 39 0.51 The Wick 281 248 218 165 222 121 0.48 Total 500 394 312 227 294 160 Mean 0.33 SD 0.23 SE 0.13

42 Seabird monitoring on Skomer Island in 2013

Table 18 Black-legged Kittiwake breeding success (per AON) on Skomer Island 1989-2013

Year Mean breeding success Standard Error

1989 0.70 0.04 1990 0.60 0.07 1991 0.86 0.07 1992 0.47 0.12 1993 0.65 0.08 1994 0.90 0.14 1995 0.94 0.11 1996 0.45 0.06 1997 0.68 0.06 1998 0.79 0.09 1999 0.95 0.06 2000 0.78 0.08 2001 0.21 0.08 2002 0.61 0.07 2003 0.60 0.06 2004 0.53 0.08 2005 0.47 0.08 2006 1.01 0.16 2007 0.30 0.07 2008 0.39 0.13 2009 0.55 0.09 2010 0.65 0.06 2011 0.52 0.10 2012 0.32 0.06 2013 0.33 0.13 Mean (SE) 0.61 0.09

Figure 19 Black-legged Kittiwake breeding success on Skomer Island 1989-2013

1.2

1

0.8

0.6

0.4 Chicks Fledged perAON 0.2

0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Due to the very poor productivity and lack of synchrony between pairs, very few clutches were observed directly and as such the relationship between clutch size and breeding success could not be examined as in previous years.

43 In previous years, the relationship between breeding success and number of eggs and chicks hatched was examined (see Table 19 for 2012). However, the number of nests where egg numbers could be observed with certainty was too small this year and breeding was so uncoordinated that the data was not analysed. This study will hopefully be picked up again in 2014.

Table 19 Black-legged Kittiwake fledging success in relation to the number of chicks hatched per nest on Skomer Island 2012

No. of chicks in No. of nests Total no. of chicks No. chicks fledged Success rate per nest nest 1 11 11 2 0.18 2012 2 11 22 4 0.36 3* - - - - * No nests were observed with three eggs in 2012

11.3 Timing of breeding

Nest building was first seen on the 10th May, however there were several bouts of bad weather early in the season which set many pairs back and this season was very poorly synchronised. Therefore the 2013 dates in Table 21 are a very rough guide of Kittiwake chick progression.

The first egg was seen on the 28th of May with the first chick sighted on the 23rd of June. Despite a slightly late and prolonged 2013 season due to adverse weather, there does not appear to be any major change in the timing of breeding over the last 7 years.

Table 20 Black-legged Kittiwake - timing of breeding on Skomer Island 2007-2013

2007 2008 2009 2010 2011 2012 2013

Nest building start 7th May 8th May 30th April 30th April 7th May 7th May 10th May First egg 19th May 24th May 11th May 21st May 13th May 20th May 28th May First chick 16th June 20th June 11th June 8th June 10th June 11th June 23rd June

11.4 Adult survival

Appendix 4 gives the estimated survival rates of Kittiwakes. These are based on colour-ringed birds at two sites, Tom's House and South Cliff.

Survival averages 0.85. There is a slight but not nearly significant, decline with time.

44 Seabird monitoring on Skomer Island in 2013

Figure 20 Trend of annual survival estimate of Black-legged Kittiwake from 1978 up to and including survival in 2011

45 12 Common Guillemot Uria aalge 12.1 Breeding numbers - whole island counts

Two full whole island counts were conducted in 2013, helped enormously by the, mostly, settled weather at the start of June, especially with sea based counts. It is also worth noting the change of wardening staff in 2013 which can cause slight differences in counts from year to year. To mitigate for this, we always had two or three people counting each section, especially when sea based. A mean of 20862 Individuals (IND) were counted, with a range of 19422 – 22302. This is 7.31% lower than 2012 (Table 21, Figure 21).

Table 21 Common Guillemot whole island counts on Skomer Island, 2004-2013

Year Land % change Sea count % change Total % change 5-year count count % change 2008 11579 -23.6 5509 +56.5 17088 -2.60 +20.45 2009 14339 +23.8 5173 -6.10 19512 +14.19 -1.01 2010 15643 +9.09 4319 -16.51 19962 +2.31 +17.58 2011 15064 -3.70 6624 +53.37 21688 +8.65 +23.62 2012 16557 +3.78 5951 -10.17 22508 +3.78 +31.72 2013 - - - - 20862 -7.31 +6.92

12.2 Breeding numbers - study plot counts

The study plots are thought to be representative of the whole colony (Wilson 1992) and may reflect any population change more accurately than the whole island counts, as repeated counts take account of variations in attendance that is thought to occur within colonies. For details of counts refer to Appendix 2.

The number of common guillemots within the study plots as a whole has changed little over the last four years, with a small increase of 4.3% on 2009 numbers (Table 22). Overall there does still seem to be a slow increase taking place with Skomer guillemots as a whole, but there are still significant differences between study plots.

During 2011/12 a slightly different methodology (involving varying timings and observers) was used due to time restrictions placed upon staff. In 2013 the traditional method of standardised timings of counts and a single observer to reduce variation between repeat counts led to data with a much lower level of variance (Standard deviation of 282.2 rather than 416.5 and Standard Error of 106.7 rather than 157.4). Counts have traditionally been compared year on year as well as with the season five years previous; this year a comparison with the 2010 season’s data has been included (being the last season with the standard methodology in place). This gave a statistically insignificant (p>0.05) increase of individuals of 5.3% over all of the study plots.

46 Seabird monitoring on Skomer Island in 2013

Table 22 Common Guillemot study plot totals on Skomer Island 2009-2013

5yr Study Plot Year Mean SD SE Signif %change %change

Bull Hole 2009 3503.1 126.4 40.0 * * +19.9 +27.7 2010 3493.9 287.2 90.8 NS -0.26 +23.7 2011 3569.1 348.1 110.1 NS +2.2 +21.4 2012 4201.3 214.1 80.9 * * +17.7 +43.8 2013 3553.9 144.6 54.6 * * -15.4 +1.5 2013 v 2010 NS +1.7

High Cliff 2009 2057.8 105.3 33.3 * * +36.3 +37.4 2010 2024.1 158.6 50.2 NS -1.6 +34.8 2011 2006.5 124.1 39.2 NS -0.9 +31.6 2012 1801.7 346.6 131.0 * -10.2 +19.3 2013 2161.4 106.0 40.1 * +20.0 +5.0 2013 v 2010 * +6.8

S.Stream 2009 897.4 70.7 22.4 * * +38.9 +36.7 2010 882.4 98.3 31.1 NS -1.7 +23.7 2011 804.1 47.3 15.0 * -8.8 +19.3 2012 908.1 100.6 38.0 * +12.9 +40.6 2013 1021.3 41.5 15.7 * +12.5 +13.8 2013 v 2010 * * +15.74

All 2009 6458.3 282.3 89.3 * * +27.2 +31.9 2010 6400.4 446.2 141.1 NS -0.9 +27.5 2011 6360.5 419.2 148.2 NS -0.62 +23.7 2012 6911.1 416.5 157.4 * +8.66 +36.1 2013 6736.6 282.2 106.7 NS -2.5 +4.3 2013 v 2010 NS +5.3

Note: Significance between years established using the t-test for comparing the means of two small samples (two- tailed test, df=16). N S Not significant, * Statistically significant (P<0.05), * * Statistically highly significant (P<0.01). See Appendix 9 for count details.

47 Figure 21 Common Guillemot breeding numbers on Skomer Island 1963-2013

25000

20000 Whole island count (mean)

Study plot count 15000

10000 N0. of individuals

5000

0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012

48 Seabird monitoring on Skomer Island in 2013

12.3 Breeding success

12.3.1 Methods

The number of active and regularly occupied sites was established at study plots and their histories were followed, using the methodology outlined in Walsh et al. (1995). Visits were made from mid-April but due to bad weather postponing egg laying records started from 15th May with last visits occurring on the 21st July. Sites received visits with a similar frequency, of one visit every one or two days, but the number of visits ranged from 26 to 44 as the small and exposed Transverse Ledge site had a significantly shorter active period.

12.3.2 Results

2013 saw a mean productivity of 0.63 fledged birds per active and regularly occupied site, which is the same figure as in 2012 (Tables 23 and 24, Fig 22), and is slightly lower than the overall mean of 0.68 (1989 – 2013). Three hundred and twenty eight active and regular sites were recorded this year, ten more than last year. However, only 86% of these sites were active, which leads to a relatively higher productivity quota for active sites only of 0.72.

The median fledge date was 10th July, 13 days later than in 2012, and 53% of chicks “fledged” in the five visit dates centred on this day (7th to the 12th July).

Table 23 Common Guillemot breeding success (per active and regular sites) on Skomer Island 1989-2013

Mean Productivity Year No. Sites Large Chicks across sites SE 1989 120 96 0.80 0.05 1990 112 80 0.69 0.05 1991 117 89 0.76 0.05 1992 169 121 0.72 0.04 1993 198 141 0.72 0.05 1994 187 131 0.72 0.03 1995 198 151 0.75 0.04 1996 210 161 0.77 0.02 1997 226 174 0.77 0.33 1998 201 154 0.77 0.04 1999 242 147 0.65 0.05 2000 227 143 0.65 0.08 2001 259 160 0.65 0.08 2002 259 170 0.68 0.03 2003 268 179 0.71 0.05 2004 292 184 0.63 0.01 2005 297 200 0.70 0.03 2006 287 142 0.47 0.07 2007 258 164 0.63 0.02 2008 269 164 0.62 0.06 2009 254 185 0.73 0.05 2010 315 211 0.69 0.04 2011 292 149 0.55 0.06 2012 318 185 0.63 0.08 2013 328 212 0.63 0.05 Mean (1989-2013) 0.68 0.06

49 Table 24 Common Guillemot breeding success (per active site and per active and regular site) on Skomer Island 2013

No. active + Large Productivity Productivity (a regular sites No. active sites chicks (a+r) only) Wick 1G 69 56 39 0.57 0.70 Wick 2G 95 80 59 0.62 0.74 Wick Corner 103 92 74 0.72 0.80 Bull Hole 40 37 30 0.75 0.81 Transverse Ledge 21 18 10 0.48 0.56 Mean 0.63 0.72 SD 0.11 0.10 SE 0.05 0.05

Figure 22 Common Guillemot breeding success on Skomer Island 1989-2013

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

Fledglings per active and regular nest nest siteregular and active per Fledglings 0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

12.4 Timing of breeding

The first egg was noted on 7th May at the Amos (two weeks later than in 2012), the first chicks on the 9th June and the first ‘jumpling’ on the 1st July at the Wick and Transverse Ledge. The last ‘jumpers’ in study plots were 21st July, but fledgling sized chicks were present on the Wick until at least the 11th August.

Table 25 Common Guillemot timing of breeding 2007-2013

2007 2008 2009 2010 2011 2012 2013 First egg 5th May 11th May 25th April 29th April 21st April 23rd April 7th May First chick 8th June 14th June 26th May 31st May 26th May 27th May 9th June First 30th June 25th June 11th June 23rd June 15th June 18th June 1st July ‘jumpling’

50 Seabird monitoring on Skomer Island in 2013

12.5 Adult and juvenile survival

This and other Common Guillemot studies were undertaken by University of Sheffield in 2013. This report is attached in Appendix 6.

51 13 Razorbill Alca torda 13.1 Breeding numbers - whole island counts

Due to difficulties in censusing this species (being less concentrated than Common Guillemots and often breeding in hidden sites amongst boulders and in burrows), the pattern of Razorbill numbers on Skomer is probably not a true reflection of the true population trends (Figure 23).

Two whole island counts were carried out in June 2013 producing a mean of 6663 Individuals (IND). The range was 6550 – 6777. This is a 34.04% increase on the previous year and a 26.63% increase over a five year period.

The large increase in numbers on the previous year is probably due to low productivity relating to high nest site abandonment by the time of counting in 2012. However, it should be born in mind that the most visible pairs are in the sub-optimal sites and are therefore most likely to fail.

Table 26 Razorbill whole island count details on Skomer Island, 2006-2013

Year Land % change Sea count % change Total % change % 5-yr count count change 2006 2955 -22.5 1606 -17.6 4561 -20.8 -10.5 2007 3588 +21.4 1259 -21.6 4847 +6.3 +14.3 2008 2336 - 34.9 2637 + 109.5 4973 + 2.6 + 2.6 2009 2970 +27.1 2292 -13.1 5262 +5.8 - 8.6 2010 2835 -4.55 2556 +11.6 5391 +2.5 +18.2 2011 2141 -24.48 2977 16.47 5118 -5.06 5.59 2012 2428 13.40 2543 -14.58 4971 -2.87 -0.04 2013 - - - - 6663 +34.04 +26.63

13.2 Breeding numbers - study plot counts

The Razorbill study plot counts are not thought to be as representative of the whole island population as those of Guillemots (Wilson 1992) although changes in the plot counts between years is still useful information.

There were significant increases at all study plot sites in 2013 over 2012 figures (Table 27), several of which were highly significant (p<0.01). Over all four study plots this totalled an increase of 49.0% between 2012 and 2013, suggesting the more modest increase of 12.9% over the last five years is a more indicative result.

During 2011/12 a slightly different methodology (involving varying timings and observers) was used due to time restrictions placed upon staff. In 2013 the traditional method of standardised timings of counts and a single observer to reduce variation between repeat counts led to data with a much lower level of variance (Standard deviation of 68.6 rather than 168 and Standard Error of 25.9 rather than 63.5). Counts have traditionally been compared year on year as well as with the season five years previous; this year a comparison with the 2010 season’s data has been included (being the last season with the standard methodology in place). This gave a statistically significant (p<0.05) increase of 11% over all of the study plots.

52 Seabird monitoring on Skomer Island in 2013

Table 27 Razorbill study plot totals on Skomer Island 2009-2013

Study Plot Year Mean SD SE Signif %Change 5Yr %Change

Bull Hole 2009 390.1 32.8 10.4 * * +27.0 +34.6 2010 432.5 88.0 27.8 NS +10.9 +87.3 2011 304.3 45.4 14.3 * -29.7 -4.71 2012 316.6 77.8 29.4 NS +4 +3 2013 451.3 31.2 11.8 * * +42.5 +15.7 2013 v 2010 NS +4.3

High Cliff 2009 393.4 67.2 21.3 * * +42.4 +31.0 2010 380.2 63.4 20.0 NS -3.4 +102.8 2011 292.1 54.8 17.3 * -23.2 16.1 2012 309.9 68 25.7 NS +6 +12.2 2013 359.6 29.8 11.3 NS +16.0 -8.6 2013 v 2010 NS -5.4

S.Stream 2009 97.4 17.8 5.6 * * +32.0 -15.8 2010 111.4 26.3 8.3 NS +14.4 +23.8 2011 72.0 24.7 7.8 NS -35.4 -23.4 2012 78 46.7 17.6 NS +8.3 +5.7 2013 127.4 16.5 6.2 * +63.3 +30.8 2013 v 2010 NS +14.4

The Wick 2009 739.6 39.6 12.5 NS +1.6 +40.5 2010 723.8 33.1 10.5 NS -2.1 +40.5 2011 718.0 19.8 6.3 NS -0.8 +33.9 2012 568 29.9 11.3 * * -20.9 -21.9 2013 891.1 42.6 16.1 * * +56.9 +20.5 2013 v 2010 * * +23.1

All Plots 2009 1620.5 105.21 33.27 * * +17.0 +31.5 2010 1647.9 184.7 58.4 NS +1.7 +61.0 2011 1386 102 36 * -15.9 15.4 2012 1227.4 168 63.5 * -11.4 -11.4 2013 1829.4 68.6 25.9 * * +49.0 +12.9 2013 v 2010 * +11.0

Note: Significance between years established using the t-test for comparing the means of two small samples (two- tailed test, df=16). N S Not significant, * Statistically significant (P<0.05), * * Statistically highly significant (P<0.01). See Appendix 9 for count details.

53 Figure 23 Razorbill breeding numbers on Skomer Island 1963-2013

7000

6000 Whole island count (mean) 5000 Study plot count 4000

3000

No. of individuals 2000

1000

0 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013

13.3 Breeding success

Sites were visited were every one to two days. The date of the first visit was on 10th April, with the last visit on 18th July. All sites were visited at similar frequency between 39 and 45 times.

Productivity was given as the number of fledged or apparently fledged chicks (last seen at 15 or more days old) per active and regularly occupied site and per active site only (as defined by Walsh et al. 1995). Results are presented in Table 28, 29 and Figure 24. The mean productivity per active and regular site was 0.38 (0.47 per active only site), a large increase from the 2011 productivity figures (Table 28) and a return to recent “normal” levels of breeding success.

The least productive site was Bull hole at 0.41chicks per active or regular site (0.35 per active site only), and the most productive was Wick 3B at 0.54 chicks per active or regular site (0.46 per active site only).

The median fledge date in 2013 was 5th July, roughly a week later than in previous years, with 37% of chicks “fledging” within the five days centred on this date (3rd July to the 7th July inclusive).

54 Seabird monitoring on Skomer Island in 2013

Table 28 Razorbill breeding success on Skomer Island 1993-2013

Productivity per Productivity per active + active site regular site 1993 - 0.56 1994 - 0.55 1995 0.79 0.72 1996 0.71 0.64 1997 0.73 0.75 1998 0.71 0.66 1999 0.74 0.56 2000 0.54 0.48 2001 0.64 0.58 2002 0.37 0.36 2003 0.61 0.48 2004 0.56 0.50 2005 0.64 0.57 2006 0.33 0.30 2007 0.62 0.56 2008 0.32 0.22 2009 0.47 0.39 2010 0.51 0.40 2012 0.21 0.17 2013 0.47 0.38 Mean 0.55 0.49

Table 29 Razorbill breeding success (per active site and per active and regular site) on Skomer Island 2013

No. active + Large Productivity Productivity (a regular sites No. active sites chicks (a+r) only) High Cliff 115 95 43 0.37 0.45 Wick 1A 57 41 19 0.33 0.46 Wick 3A - - - - - Wick 3B 41 35 19 0.46 0.54 Bull Hole 81 69 19 0.35 0.41 Mean 0.38 0.47 SD 0.06 0.05 SE 0.03 0.03

55 Figure 24 Razorbill breeding success on Skomer Island 1993-2013

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

Fledglings per active and regular nest nest siteregular and active per Fledglings 0 1989 1990 1991 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2012 2013

13.4 Timing of breeding

The first egg was noted at the Basin on 8th May (fifteen days later than in 2012). The first chick was seen on 10th June at the Basin and the first ‘jumplings’ on 29th June at High Cliff (16 days later than 2012). The last chicks in the study plots “fledged” between the 14th – 18th of July, the last of the fledglings being at High Cliff.

Table 30 Razorbill timing of breeding 2006–2013

2006 2007 2008 2009 2010 2012 2013 First egg 4th May 1st May 6th May 26th April 24th April 23rd April 8th May First chick 4th June 2nd June 5th June 24th May 3rd June 27th May 10th June First 25th June 23rd June 28th June 13th June 21st June 13th June 29th June ‘jumpling’

56 Seabird monitoring on Skomer Island in 2013

13.5 Adult survival

Appendix 4 gives the estimated survival rates of Razorbill. Survival averages 0.904. The main point to make about these data is the one made in previous reports, namely that they indicate that the survival rates, averaged over several years, for this species were higher in the early 1970s than they have been since. However, there is no significant long term trend and there are no unusual figures in the last few years.

Razorbills are very difficult to census accurately and the changes in the numbers recorded do not match the variations in survival rates.

Figure 25 Trend of adult survival of Razorbill up to and including survival from 1979- 2011

57 14 Atlantic Puffin Fratercula arctica 14.1 Breeding numbers

Whole island counts of Puffins were made on two dates in April and one at the start of May. They were all made in the afternoon and in favourable weather conditions and are set out in table 30. The max count of 19,280 from the 16th of April seems the most accurate estimate of the total population, including Middleholm. The mean of all three counts was: including Middleholm: 16,437; excluding Middleholm: 16,234 but some birds were certainly in burrows by the date of the last count as can be seen by the first egg dates. No July counts were undertaken. Table 32 and Figure 26 show maximum spring counts of individuals since 1989.

Table 31 Maximum spring counts of individual Puffins on Skomer Island in 2013

Date No. individual puffins No. of individual inc. Middleholm Puffins excl. Middleholm 16/04/2013 19,280 19,120 25/04/2013 17,228 16,908 04/05/2013 12,803 12,673 Mean 16,437 16,234

Max. count: incl. MH: 19,280; excl. MH: 19,120 Mean of first two counts: incl. MH: 18,254; excl. MH: 18,014 Mean of all three counts: incl. MH: 16,437; excl. MH: 16,234 Range: incl. MH: 12,803 – 19,280; excl. MH: 12,673 – 19,120

MH=Middleholm

First egg found: 03/05/2013 First egg as calculated by back dating from the first birds seen with fish: Mid April Date of first birds seen with fish: 25/05/2013

Table 32 Maximum spring counts of Atlantic puffin on Skomer Island and Middleholm 1989-2013

Year No. individual puffins %change 5 year % change 2004 10688 +25.2 +0.7 2005 10717 +0.3 +36.5 2006 10876 +1.5 +5.5 2007 11821 +8.7 +38.5 2008 10487 -11.3 -1.9 2009 13508 +28.8 +26.0 2010 12577 -6.89 +15.64 2011* - - - 2012 11497 -8.59 +9.63 2013 19280 +67.70 +42.73 * 2011 – No puffin count was possible due to timings/weather/availability of counters.

58 Seabird monitoring on Skomer Island in 2013

Figure 26 Maximum spring counts of Atlantic Puffin on Skomer Island 1989-2013

25000

20000

15000

10000 No. of individuals

5000

0 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

14.2 Puffin burrow occupancy and breeding success on Skomer in 2013

Puffin burrow occupancy and breeding success in the South East Isthmus study plot for 2013 is shown in table 33.

Table 33 Burrow occupancy and breeding success of Puffins on Skomer Island in 2013

Burrow Total No. No. occupied % Occupied No. of chicks Productivity distance from burrows burrows based on 2 based on 2 (3+) the cliff edge (3+) feeds feeds (m) < 5 94 72 77% 57 (57) 0.8 (0.8) 5-10 57 46 81% 37 (37) 0.8 (0.8) >10 74 45 61% 33 (30) 0.7 (0.66) Total 225 163 72% 127 (125) 0.78 (0.76)

Burrow occupancy was established over eight evening watches, from 18:30 to dusk, between the 25th April and 20th May. Due to the exceptionally cold spring this year nest-building was very protracted and some birds were still mating until the end of May and actively nest-building into early June.

The first adult Puffins carrying fish were seen coming ashore on 25th May, eight days later than last year but still suggesting a lay date in mid-April. However it was not until the second week of June that many birds started to feed chicks, suggesting the majority of birds did not lay until May.

59 Breeding success was based on 24 hour feeding watches. In the past, two watches have always been carried out, one timed for mid-feeding and the second just before the first chicks fledge. However, getting the timing right is very hard – too late and early chicks can fledge before the second watch, too early and late chicks have not hatched by the first watch. Therefore, as recommended in last year’s report, feeding watches were carried out every two weeks from the first date adults were seen bringing in fish.

Due to the spread out breeding season this year four 24 hour watches had to be done:

10:00 – 22:30 on 8 June and 04:30 – 10:00 on 9 June: two weeks after the first adults were seen coming in with fish.

04:30 – 22:30 on 24 June: the oldest chicks should only have been 30 days old so none should have fledged and therefore should have been recorded being fed on both watches if they survived. Chicks hatched since the first watch were now being fed as well.

04:30 – 22:30 on 10 July: the very oldest chicks will have fledged but any chicks born since the first watch should only be an absolute maximum of 31 days old and therefore fed on the second and third watches.

10:00 – 22:30 on 22 July and 04:30 – 10:00 on 23 June: most chicks should have fledged and only the very latest chicks should still be being fed

For a burrow to be considered successful it had to have been fed during at least two watches. This gives a success rate of 78% (or 0.78 chicks fledged per burrow). Although this seems to be a potentially much more accurate method for monitoring Puffin productivity nothing is perfect and five known chicks did not get fed during a 24 hour period: two chicks were fed during the first and third watches but not the second; one chick was fed during the second watch but not the third but was ringed the next day and two chicks were seen coming out and exercising from burrows during feeding watches that had not been fed during that watch.

Neither method really takes Great Black-backed Gull predation into account either. This usually occurs when old chicks are coming out and exercising in the evenings. On Skokholm Puffin productivity was monitored this year by short daily watches throughout the chick rearing period and chicks were only assumed to have survived if they were known to have reached at least 31 days old. This gave a much lower success rate of 0.49 chicks fledged per occupied burrow.

60 Seabird monitoring on Skomer Island in 2013

14.3 Feeding rates

Details of feeding rates were recorded as follows:

Table 34 Feeding rates of Puffins on Skomer Island in 2013

8/9 June 24 June 10 July 22/23 July No. of burrows to which feeds 85 130 125 43 were recorded Total number of burrows fed 85 140 148 149 during watches Total No. recorded feeds 273 742 455 135 Mean No. feeds per burrow 3.2 5.7 3.6 3.1 (range) (1-10) (1-17) (1-10) (1-15)

These totals include single feeds brought in to five burrows not thought to be occupied

Figure 27 Number of feeds per hour for Puffins on Skomer Island in 2013

90

80

70

60

50 8/9 June 40 24-Jun 30 10-Jul 20 22/23 Jul

10

0 04:30-05:00 05:00-06:00 06:00-07:00 07:00-08:00 08:00-09:00 09:00-10:00 10:00-11:00 11:00-12:00 12:00-13:00 13:00-14:00 14:00-15:00 15:00-16:00 16:00-17:00 17:00-18:00 18:00-19:00 19:00-20:00 20:00-21:00 21:00-22:00 22:00-22:30

As usual there was a large early morning peak in feeding activity, followed by a smaller evening peak in all four watches.

61 14.4 Adult survival

Appendix 4 gives the estimated survival rates of Puffins. As with Razorbills survival rates for the last few years show no unusual patterns. Average survival averages remains at 0.91, but there is an almost significant decline with time (p = 0.063, R2 = 7.0%).

Figure 28 Trend of annual survival estimate of Atlantic Puffin 1972 up to and including survival from 2011

62 Seabird monitoring on Skomer Island in 2013

16 References

Ashton, H (1993) The Breeding Population of the British Storm Petrel Hydrobatespelagicus on Skomer Island, Dyfed. Report to the Dyfed Wildlife Trust.

Birkhead, TR (2003) Skomer guillemot studies 1999-2001. Department of Animal and Plant Sciences, The University of Sheffield. CCW Contract Science Report 532.

Birkhead, TR and Hatchwell,BJ (2000) Skomer Island Razorbill Study 2000. Department of Animal & Plant Sciences, University of Sheffield.

Birkhead, TR and Hatchwell,BJ (2001) Skomer Island Razorbill Study 2001. Department of Animal & Plant Sciences, University of Sheffield.

Birkhead, TR, Hatchwell,BJ and Bowgen, K. (2010) Skomer Island Guillemot Study 2010. Department of Animal & Plant Sciences, University of Sheffield.

Boulinier, T and Danchin, E (1996) Population trends in Kittiwake Rissatridactyla colonies in relation to tick infestation. Ibis 138:326-334.

Cramp, S and Simmons, KEL (eds) (1983) Handbook of the birds of Europe, the Middle East and North Africa: the birds of the Western Palearctic. Volume III. Waders to gulls. Oxford University Press, Oxford.

Field, RH, Birkhead, TR &Hatchwell, BJ (1999) Skomer Island razorbill study 1999. Department of Animal and Plant Sciences, The University of Sheffield.

James, PC (1984a) The vocal and homing behaviour of the Manx shearwater Puffinuspuffinus with additional studies on other Procellariformes. Appendix B. PhD thesis, Oxford University.

James, PC (1984b) Sexual dimorphism in the voice of the British Storm Petrel Hydrobatespelagicus. Ibis, 126, 89-92.

Poole, J and Sutcliffe, SJ (1992) Seabird studies on Skomer Island in 1992. JNCC Report No. 139.

Poole, J & Sutcliffe, SJ (1993) Monitoring of seabirds on Skomer Island in 1993. Report to JNCC by DWT.

Poole, J, Smith, S, Perrins, CM, Birkhead, TR and Thompson, KR (2001) Seabird monitoring on Skomer Island in 1997 and 1998. JNCC Report, No.318.

Sutcliffe, SJ (1993) Population trends of the Larusgulls breeding on Skomer Island NNR 1960-92. Report by Dyfed Wildlife Trust for CCW.

Sutcliffe, SJ (1997) Populations of breeding Larus gulls on Welsh islands. In Rhind, PM, Blackstock, TH and Parr, SJ (eds.) Welsh islands: ecology, conservation and land use. CCW.

Taylor, CJ., Boyle, D. and Perrins, CM, Kipling, R. (2012) Seabird monitoring on Skomer Island in 2012.JNCC Report

63 Walsh, PM, Halley, DJ, Harris, MP, del Nevo, A, Sim, IMW, Tasker, ML (1995) Seabird monitoring handbook for Britain and Ireland. Peterborough, JNCC, RSPB, ITE, Seabird Group.

Wilson, SA (1992) Population trends of guillemots Uria aalge and razorbills Alca torda on Skomer Island, Wales 1963-1991. Report to CCW by Sheffield University.

64 Seabird monitoring on Skomer Island in 2013

17 Appendices Appendix 1 Mean seabird counts by section

Counted Fulmar - Guillemot Razorbill Kittiwake Section from Sea AOS - IND - IND - AON or Land? mean mean mean 1 Sea 38.5 0.0 606.0 441.0 2 Sea 0.0 10.0 241.0 162.0 3 Sea 9.5 85.0 196.0 92.5 4 Sea 25.0 2.0 267.0 114.5 5 Sea 15.0 29.0 445.0 84.0 6 Sea 12.5 9.0 189.5 90.0 7 Sea 0.0 0.0 14.5 28.0 Land and 8 sea 15.0 0.0 176.5 127.5 9 Land 12.0 0.0 91.0 72.0 10 Sea 1.0 0.0 0.0 3.0 11 Sea 27.0 0.0 167.5 211.0 12 Land 29.0 0.0 16.5 67.5 13 Land 0.0 0.0 0.0 57.0 14 Land 4.0 45.0 1013.0 125.0 15 Sea 6.5 22.0 7.5 27.5 16 Land 26.5 61.0 2126.0 352.0 17 Sea 0.0 0.0 0.0 17.5 18 Sea 0.0 0.0 0.0 2.0 19 Sea 0.0 0.0 314.5 232.5 20 Sea 1.5 0.0 0.0 42.5 21 Land 0.0 0.0 0.0 0.0 22 Land 56.5 503.0 4059.5 813.5 23 Land 0.0 0.0 0.0 2.0 24 Sea 0.0 0.0 67.0 67.0 25 Sea 0.0 0.0 20.0 22.0 26 Land 37.5 0.0 292.5 223.0 27 Land 0.0 102.0 2311.5 88.5 28 Land 1.0 0.0 11.5 104.0 29 Sea 0.0 0.0 0.0 0.0 30 Land 0.0 0.0 239.5 34.5 31 Land 2.0 0.0 258.0 66.5 32 Land 0.0 0.0 95.0 45.5 33 Land 1.5 33.0 687.0 35.5 34 Land 0.0 0.0 553.0 45.0 35 Sea 10.0 0.0 120.0 270.0 36 Sea 7.0 4.0 535.0 200.0 37 Land 14.5 90.0 3094.5 460.5 38 Sea 4.0 50.0 505.0 256.0 39 Sea 54.5 0.0 757.0 385.0 40 Sea 2.0 0.0 522.5 258.0 41 Sea 24.5 0.0 83.5 171.5 42 Sea 47.5 0.0 302.0 217.0 43 Sea 17.0 0.0 116.0 117.5 44 Sea 1.0 0.0 243.5 301.5

65 45 Sea 0.0 0.0 117.0 130.5 TOTAL 503.5 1045.0 20862.0 6663.5

66 Seabird monitoring on Skomer Island in 2013

Appendix 2 Dates of visits to Black legged kittiwake sub- colonies in 2013

The kittiwake population lacked much of its normalsynchrony in 2013. Therefore, visits could not be timed accurately to record certain age stages and more visits were required to cover the wide range of chick ages.

1st visit 26th May to identify nests sites 2nd visit 6th June to record incubating birds / new sites 3rd visit 23 June to record incubating birds / small chicks 4th visit 6th July to record large chicks / site attendance 5th visit 23rd July to record fledgers / site attendance 6th visit 2nd August to record fledgers / site attendance 7th visit 11th August to record fledgers / site attendance 8th visit 18th August to record fledgers / site attendance

67 Appendix 3 Guillemot and Razorbill Study Plots

Common Guillemot study plot totals in 2013 (No. Individuals)

South Stream High Cliff Bull Hole All Plots 01-Jun NNW4 983 2065 3457 6505 02-Jun NNW3 999 2132 3566 6697 03-Jun NE2 1006 2168 3492 6666 05-Jun ENE3 1050 2143 3500 6693 07-Jun ENE3 971 2059 3392 6422 09-Jun SE1 1067 2188 3642 6897 18-Jun NW2 1073 2375 3828 7276 Mean 1021.29 2161.43 3553.86 6736.57 SD 41.46 105.95 144.55 282.18 SE 15.67 40.05 54.64 106.65

Razorbill study plot totals in 2013 (No. Individuals)

South Stream High Cliff Bull Hole Wick All Plots 01-Jun NNW4 107 343 509 961 1920 02-Jun NNW3 115 357 453 907 1832 03-Jun NE2 137 358 427 897 1819 05-Jun ENE3 148 357 454 920 1879 07-Jun ENE3 110 311 435 850 1706 09-Jun SE1 133 388 414 857 1792 18-Jun NW2 142 403 467 846 1858 Mean 127.43 359.57 451.29 891.14 1829.43 SD 16.50 29.84 31.18 42.61 68.59 SE 6.24 11.28 11.79 16.11 25.92

68 Seabird monitoring on Skomer Island in 2013

Appendix 4 Maximum spring Atlantic Puffin counts on Skomer Island, including Middleholm, 2013

Count date No. individual Puffins 16/04/2013 19,280 25/04/2013 17,228 04/05/2013 12,803

69 Appendix 5 Mean annual estimated survival rates of seabirds on Skomer Island

Note: figures relate to estimated survival rate from the year in column one to the following year.

Skomer Seabird Survival Estimate to 2011

Lesser Black- Manx backed Herring Year Shearwater Gull Gull Kittiwake Razorbill Puffin 1970 0.9486 1971 0.9793 1972 0.9744 0.9589 1973 0.9660 0.9461 1974 0.9273 1975 0.9094 0.9554 1976 0.9933 0.9728 1977 0.8125 0.8672 0.9911 1978 0.7805 0.9830 0.8937 0.9441 0.9086 0.8241 1979 0.7931 0.9211 0.8086 0.8910 0.8760 0.8961 1980 0.6118 0.9214 0.9820 0.8861 0.9125 1981 0.8741 0.6075 0.8423 0.6607 0.8490 1982 0.8865 0.9596 0.7051 0.9182 0.8372 0.8656 1983 0.8506 0.8943 0.7237 0.7264 0.8985 0.9080 1984 0.9533 0.9170 0.8959 0.8297 0.8456 0.8469 1985 0.8923 0.7270 0.8046 0.8276 0.8727 1986 0.8764 0.8834 0.8721 0.9139 0.8970 0.9485 1987 0.9420 0.9437 0.9507 0.8977 0.8997 0.9349 1988 0.9144 0.9383 0.8968 0.9087 0.9715 1989 0.7216 0.9217 0.8797 0.9595 0.9201 0.9640 1990 0.9236 0.8422 0.8205 0.9003 0.8964 0.8855 1991 0.9775 0.8380 0.9508 0.9217 1992 0.8226 0.8812 0.8751 0.8574 0.9236 0.9804 1993 0.7272 0.9727 0.9030 1994 0.8989 0.8255 0.7996 0.7662 0.8668 0.8922 1995 0.8908 0.8586 0.7716 0.7745 0.9075 0.9195 1996 0.8823 0.8091 0.7212 0.7064 0.8944 0.8676 1997 0.8674 0.8025 0.7325 0.8813 0.9156 1998 0.9502 0.9059 0.8296 0.8660 0.9626 0.8741 1999 0.9815 0.8680 0.8466 0.7766 0.9021 0.9280 2000 0.9047 0.8220 0.8677 0.9073 0.9520 0.8808 2001 0.9073 0.7944 0.7871 0.9172 0.8801 0.9184 2002 0.8352 0.7548 0.8713 0.7949 0.8841 0.8874 2003 0.9181 0.7755 0.8043 0.8878 0.8478 0.8294 2004 0.9089 0.8984 0.8556 0.7560 0.9582 0.8922 2005 0.8217 0.8701 0.7814 0.8276 0.8800 0.8792 2006 0.9234 0.9019 0.8063 0.6718 0.7742 0.8681 2007 0.8594 0.8810 0.8913 0.7431 0.8641 0.8475 2008 0.8193 0.9201 0.8800 0.7467 0.9797 0.9389 2009 0.7709 0.8290 0.6949 0.9455 0.9047 0.9335 2010 0.7872 0.8304 0.7632 0.9062 0.9706 0.8378 2011 0.8192 0.8363 0.7419 0.7899 0.9091 0.9364

70 Seabird monitoring on Skomer Island in 2013

Appendix 6 Ringing Totals for 2013

Adult Pullus Total Manx Shearwater 183 416 599 Storm Petrel* Puffin 72 116 188 Guillemot 51 240 291 Razorbill 28 51 79 Kittiwake 5 5 Lesser 16 376 392 Black-backed Gull Herring Gull 5 5 All birds ringed as part of Research Projects. * Data from Matt Wood unavailable at time of writing.

71 Appendix 7 Skomer Island Guillemot Study 2013 (Birkhead, Hatchwell and Meade, 2010)

SKOMER ISLAND GUILLEMOT STUDY 2013*

T. R. BIRKHEAD, B. J. HATCHWELL, J. M. S. CRANE University of Sheffield, S10 2TN, UK

Summary

1. A total of 252 individuals ringed as adults was recorded on the Amos Guillemot colony. Annual survival, calculated as the number of birds resighted in 2013 that were ringed as adults in previous years was 91% (252/278). The improved estimate of the 2011-2012 annual adult survival was 88%.

2. A total of 741 individuals previously ringed as chicks on Skomer were resighted.

3. A total of 281 chicks were ringed on the Amos in 2013. In addition, 41 new adults were ringed and four had rings replaced due to loss or damage. Ring sequences used were R0001-R0300 and Y685-Y800.

4. The productivity of 117 breeding sites was monitored at the Amos between 7 April and 21 July, with 111 pairs laying eggs and 95 chicks fledging from these sites. Breeding success was 0.86 (95/111) chicks per pair. This figure is higher than that of 2012 (0.79) and 2010 (0.76), but comparable to that of 2011 (0.87) and 2009 (0.84).

5. The first egg was laid at the Amos on 7 May. The median lay, hatch and fledge dates for the monitored pairs were 18 May, 20 June and 12 July respectively, making the 2013 breeding season approximately 11 days later than 2012 and 17 days later than 2011. Compared to previous years, the 2013 season was most similar in timing to 2008, with these two seasons being the latest since 1996. The delayed onset of breeding was probably caused by the exceptionally cold spring and cooler sea temperatures.

6. Clupeids (Clupeidae) constituted a smaller proportion (65.1%) of chick diet than seen in previous years, with sandeel (Ammodytidae) (27.6%) and gadid (Gadidae) (7.3%) being represented in higher proportions. This may suggest that clupeids were less easily available as a food source than in previous years. The mean feeding rate was 4.53 feeds per chick per day, a relatively high rate compared to most previous years.

72 Seabird monitoring on Skomer Island in 2013

7. Of 25 geolocators in deployment on adult guillemots, six were recovered. Data were successfully downloaded from five devices and six devices were redeployed. The device that failed to download has been sent to Biotrack for data extraction.

*summary of a report available from the senior author

73 Appendix 8

Gull Diet Survey Comparison of Methods

This report was issued to JNCC in November 2012. The JNCC have had the raw data and are doing some further analysis (2012).

Ali Quinney and Richard Kipling

Introduction For several years the WTSWW Field Assistant on Skomer Island has carried out a diet survey of the 25 Great Black-backed gull nests included in the breeding productivity survey. The diet survey has evolved over time, and as a result there are some vagaries in the methodology which may affect the accuracy of findings. Here, the results of the WT diet survey carried out in 2012 are compared with the results of a more rigorous method, in order to test the validity of the WT findings.

The WT Method The WT survey of GBBGU diet involves a visit to each nest shortly after the chicks have fledged. The purpose is to provide a straightforward estimate of the diet of GBBGU chicks, and to record the frequency, number and relative abundance of rabbit and Manx Shearwater remains. A simple visual survey is made of the area surrounding each nest, and each prey item is recorded. There are two issues with the WT methodology: 1) Some of the prey categories are to some extent subjective, and some prey items could be fitted into a number of them, depending on the experience of the recorder. For example, if a bone is found it may be classed simply as ‘bone’, identified as bird or mammal, or identified to species level. A number of bones found close together may be grouped together as a single entry, or treated separately, and again this is a matter of judgement for the surveyor. 2) The size of the area around the nest in which prey is included is left to the discretion of the recorder, and this introduces a further element of uncertainty and variation into the dataset.

The Comparison (Transect) Method The purpose of the comparison method was: 1) To assess if the proportions of recorded prey types differed from those found using the WT survey method. 2) To assess changes in the amount or type of prey items found at increasing distances from each nest, in order to provide recommendations on how far from a nest prey items should be recorded. A cross-shaped transect centred on each nest was used to observe prey items. Along the transect a one metre square quadrat was used, and prey items observed within each quadrat recorded. Prey categories defined in a previous intensive diet survey were used as the basis for the classification of prey. Items were divided into categories specific enough to minimise subjectivity in the classification of each prey type (Appendix 2). Prey were recorded at the centre of the transect (the nest site itself, distance zero). From the centre, quadrats then led out from the centre to a total distance of ten metres. The direction of

74 Seabird monitoring on Skomer Island in 2013

each arm of the cross-shaped transect was, either, north south east and west, or along and perpendicular to the rocky outcrop (for nests found on linear ridges). Nests were allocated to three categories: ridge (nest on linear feature), rocky outcrop (nest on non-linear feature), plateau (nest on flat area with no significant drops or edges close). Only at ridge nests were transect arms dictated by the direction of the geographical feature on which the nest was situated. The nest categories allowed data gathered using different orientation rules to be analysed separately, and enabled a comparison of ecological differences between nest sites to be undertaken (results in separate report).

Data Analysis Methods Analysis of Prey Changes with Distance from nest The amounts of different prey types recorded at each distance (0-10 m) from nests were calculated. Because each distance except zero metres (the centre) included observations from four quadrats (one on each transect arm), these data were divided by four to allow comparison with the amount of prey recorded in the central quadrat. All prey items were included, and results were summed across all nests. Linear regression was used to ascertain whether the number of prey items decreased significantly with increasing distance from the nest, and the cumulative percentage of total prey recorded was calculated for each increase in distance. The analysis was repeated for nests in the three different categories defined, so see if the location of a nest affected the spread of prey around it. Changes in the type of prey recorded at increasing distances from the nest were analysed by comparing the proportions of different prey types recorded at different distances. A Chi Squared Test was used, and data were merged into broad prey categories to ensure that the test was valid (no categories with expected values < 5). Methodological Comparison of Findings Data were collected using the Transect Method at 18 of the 25 nest sites at which the WT diet study was conducted. The WT survey was carried out at the same time as the Transect survey where possible. At the remaining sites the WT survey was carried out first, with care taken not to disturb prey items. In order to compare the two methods, prey categories used in the Transect survey were merged to match those in the WT survey. Prey amounts were summed over all 18 nests for both methods. The proportion of prey in each merged category was compared across methodologies using a Chi Squared test, in order to ascertain if the two methods returned significantly different prey proportions.

Results Prey Changes with Distance from Nest The amount of prey found was found to decrease with distance from nests with a leptokurtic distribution (Fig. 1a). Linear regressions were carried out on Log10 transformed data for all nests (Fig. 1b) and showed a significant relationship between distance and amount of prey for all nests, and for Ridge and Stony Outcrop type nests treated separately (Table 1). There was only one Plateau type nest, so no analyses were carried out on this category. More than 80 % of prey was found between zero and five metres from nests.

75 180 3 a) b) 160 2.5 140

120 2

100 1.5 R² = 0.8081, P < 0.001 80 Amount Amount of Prey

60 Log10Amount of prey 1

40 0.5 20

0 0 0 5 10 15 0 2 4 6 8 10 12 Distance from nest (metres) Distance from nest (metres)

Fig. 1 Amounts of prey found at different distances from nest sites: a) total amounts, b) Log10 transformed amounts, with the results of linear regression analysis.

2.5 2.5

2 2

1.5 1.5

1 1 Log10 transformed prey amounts prey transformed Log10 Log10 transformed prey amounts prey transformed Log10

0.5 R² = 0.6963, P < 0.001 0.5 R² = 0.6155, P < 0.001

0 0 0 5 10 15 0 5 10 15 Distance from nest (metres) Distance from nest (metres) Fig.2 Log10 transformations of amounts of prey found at different distances from nest sites, with results of linear regression analysis a) Ridge nests, b) Stony Outcrop nests.

76 Seabird monitoring on Skomer Island in 2013

The proportions of different types of prey found close to the nest (zero to five metres) were significantly different to the proportions found further from the nest (six to ten metres) (Fig. 2). The proportion of Manx Shearwaters was lower closer to the nest site and the proportion of rabbits higher. There was no significant difference between the proportions of different types of prey found between zero and five metres from the nest, and the proportions found over all distances (Fig. 3).

0.35

Χ2 = 83.95, P < 0.001 0.3 d.f. 10

0.25

0.2

0.15 Proportion of prey of Proportion 0.1

0.05

0

Prey Type 0 to 5 6 to 10 Fig. 2 Proportions of prey found close to nests (0-5 metres) and far from nests (6-10 metres). Results of Chi Squared analysis are shown.

77 0.35

2 Χ d.f. 10 = 9.68, P = 0.47 0.3

0.25

0.2

0.15 Proportion of prey of Proportion 0.1

0.05

0

Prey Type 0 to 5 OVERALL Fig 3 Proportions of prey found close to nests (0-5 metres) compared to overall proportions of prey found (0-10 metres). Results of Chi Squared analysis are shown.

Methodological Comparison of Findings The proportions of prey in different categories for each survey methodology are shown in Fig. 4, along with the results of Chi Squared analysis. There were significant differences in the proportions of different prey items recorded using each method. The largest differences were in the categories: rabbit, bones (other), fur pellet and fish pellet. Manx shearwater, refuse, bird (other), feather pellet and crustacean categories differed least between methodologies.

78 Seabird monitoring on Skomer Island in 2013

0.3

a) Manx Shearwater 0.25 fish (inc pellets) refuse 0.2 rabbit bird (other) 0.15 bones (other) fur pellet 0.1 feather pellet Proportion of prey of Proportion invert pellet 0.05 veg pellet crust 0 WT Method TRANSECT Method 0.3 b) 0.25 2 Χ (10 d.f.) = 199, P < 0.001 0.2

0.15

0.1 WT Method

Proportion of prey of Proportion TRANSECT Method 0.05

0

Fig. 4 The proportions of prey types found using the WT and Transect survey methods, a) arranged by method, and b) arranged by prey type. Panel b) includes the results of a Chi Squared test on the differences in prey proportions between methods.

Discussion The results of this study, and big differences in the proportions of prey found in 2012 relative to 2011 (sea Skomer Sea Bird Reports 2011 & 2012) show that the outcomes of the WT GBBGU diet survey are highly sensitive to changes in the definition of prey categories, and their interpretation on the ground. This sensitivity means that the findings of previous studies must be treated with caution. Problems with prey categorisation appear to be the main reason that the WT and Transect methodologies yielded different results, and overcoming such difficulties would increase confidence in the data from WT method surveys. Analysis of the distribution of prey around nests suggests that more than 80 % of prey remains are deposited within five metres of a nest site. Prey proportions up to five metres from nests are significantly different to prey proportions between six and ten metres (Fig. 4). This suggests that a survey area greater than five metres is necessary to capture prey proportions and amounts more accurately. However, as so little prey is found beyond five metres, excluding larger distances would not have significantly changed the prey proportions recorded using the transect method within a five metre area (Fig. 3). It appears that surveying up to five metres from a nest is sufficient to gain an accurate estimate of prey types. As Manx Shearwater and Rabbit remains are quick to survey, and their relative numbers are of ecological significance, future surveys could include a wider sweep (up to ten metres) for these prey remains only. Of course, extra rabbit and shearwater remains found in this wider area should not be included in comparisons with prey types sampled only within five metres of nests. Although many of the differences between the WT and Transect survey could be accounted for by differences in categorisation, some are likely to have arisen due to prey being

79 overlooked using the WT method. The use of quadrats is likely to increase survey accuracy, providing detailed analysis of specific areas; small pellets etc. are liable to be missed in a general sweep of an area. However, quadrats are time consuming so a decision needs to be taken as to whether the increased accuracy achieved is worth the added survey effort. One solution would be to give a set time for observations at each nest (fifteen minutes may be reasonable), which would to some extent standardise survey effort and ensure that estimates of prey amounts between years were comparable.

Summary and Recommendations 1) Prey categories should be clearly and unambiguously defined for future WT surveys. Eleven broad categories are suggested below (Appendix 1). 2) It should be ensured that surveyors do not try to identify remains in more detail than the categories suggest. If this occurs, similar remains are likely to be counted differently depending on the knowledge of the surveyor (for example, an experienced worker might identify a bone as rabbit remains, while a less experienced worker would simply record them as ‘bones – other’) 3) An area of five metres, centred on the nest, should be sufficient to gain a good estimate of prey remains for all nest types 4) If a quadrat-based method is not used (due to the survey effort required) a set time should be defined for the WT search method, within the zero to five metre perimeter. This should ensure that a similar survey effort is made each year, so that prey estimates are comparable. 5) Manx shearwater and rabbit remains may be estimated up to ten metres from a nest, in order to capture differences in the proportions of these species observed at increasing distances. Only observations of rabbits and shearwaters found within the five metre area should be included in comparisons with other prey types.

Appendix 1: Suggested Prey Categories Manx Shearwater Includes shearwater bones, wings, skulls, carcasses Fish (inc pellets) Fish bones, or entire pellets Refuse Pieces of plastic, cloth, glass etc Rabbit Bones with rabbit fur, carcasses, pellets with feet etc (if bones w/o fur etc count as Bones (other) Bird (other) All bird remains (gull, guillemot, razorbill etc). Can make note e.g. of puffin numbers separate from main analysis; for analysis include all species except Manx shearwaters in this category. Bones (other) All bones not obviously Manx shearwater or fish (do not attempt detailed ID as many bones taken from landfill sites) Fur pellet Pellets containing fur. Unless obvious rabbit foot or bone count in this category, not as rabbit Feather pellet All pellets containing feathers. Include pellets with egg shell Invert Pellet All pellets containing invertebrates Veg Pellet All pellets which are vegetation only (without feathers/fur/inverts Crustacean All pellets, crustacean remains, shells

80 Seabird monitoring on Skomer Island in 2013

Appendix 2: Prey Categories used in Transect survey Prey type Prey Item Manx shearwater Complete carcasse Partial carcasse Wings Pellet of feathers without inverts Pellet of feathers with inverts Head Assorted bones Rabbit Complete carcasse Partial carcasse Fur pellet without inverts Fur pellet with inverts Bones Refuse Plastic Paper Bones (not from island) Other Fish Pellet Bones Intertidal Crab remains Mussel/limpet Pellet: Vegetation solely veg with invert remains egg shell in pellet Other birds(record species as reqd) Other bird bone Puffin foot Puffin wing LBB Remains Guillemot/Razorbill carcass Puffin skull Bones Bird sp. Mammal sp.

81