Polar Biol (2004) 27: 257–265 DOI 10.1007/s00300-004-0595-3
O R I GI N A L P A P E R
Fre´de´rique Olivier Æ Andrew V. Lee Æ Eric J. Woehler
Distribution and abundance of snow petrels Pagodroma nivea in the Windmill Islands, East Antarctica
Received: 12 May 2003 / Accepted: 4 January 2004 / Published online: 17 February 2004 Springer-Verlag 2004
Abstract In light of a paucity of available data on snowpetrel breeding areas and population sizes in the Australian Antarctic Territory, we undertook a systematic regional survey in the Windmill Islands during the 2002/ 2003 summer. Snow petrels were found breeding at most ice-free localities at varying densities. A total of 4,034 active snow-petrel nests were located; 83% of these active nests were located in the southern half of the survey area. Based on the surveys, we provide distribution maps of nesting areas at a regional scale and estimate the total breeding population at approximately 12,200 pairs.
Introduction The petrel group (Order Procellariiformes) is the most diverse, widespread and abundant flying-seabird group in the Southern Ocean (Catry et al. 2003). Therefore, they may be an important ecological indicator group, as their behaviour and population dynamics may reflect changes in the Antarctic ecosystem (Montevecchi 1993; Monaghan 1996). Although numerous studies were conducted in the last two decades, many petrel species and populations remain poorly understood. Few population data sets exist for many petrel populations in the Antarctic. Many are anecdotal and provide little basis for identifying changes in the distribution and abundance of breeding populations. The snow petrels (Pagodroma nivea) are endemic to the Antarctic region. The entire life-cycle of the species occurs in the Southern Ocean waters surrounding the continent, and its breeding grounds are located in icefree areas on the Antarctic continent. While the snow F. Olivier (&) Æ A. V. Lee Æ E. J. Woehler Institute of Antarctic and Southern Ocean Studies, University of Tasmania, Private Bag 77, 7001 Hobart, Tasmania, Australia E-mail:
[email protected]
petrel is one of the most abundant flying seabirds around Antarctica (Barbraud et al. 2000), the distribution of breeding areas for this species at regional and local scales remains poorly described in the existing literature. A comprehensive review of the existing literature and unpublished records revealed breeding reported at 298 localities in a circumpolar distribution (Croxall et al. 1995). For the Australian Antarctic Territory in East Antarctica, descriptive studies by Brown (1966), Johnstone et al. (1973), Bonner and Lewis Smith (1985) and Woehler and Johnstone (1991) report on the species’ main breeding locations in the vicinities of Davis and Mawson. In the Casey area (Windmill Islands: 66S, 110E, Table 1), snow-petrel nests were first reported in 1963 by Orton, who mapped various bird colonies. Snow petrels were subsequently reported breeding in other areas (mapped by Murray and Luders in 1972 and Jones in 1974) but no detailed surveys were undertaken to provide a total estimate of the population for the Windmill Islands region. Detailed studies on snow petrels were conducted in the 1980s and 1990s on Ardery and Odbert Islands (Bonner and Lewis Smith 1985; van Franeker et al. 1990) and provided information on distribution and abundance for these islands only. Preliminary surveys conducted by the authors in 2001/2002 confirmed snow petrels were present in high concentrations in several areas of the Windmill Islands in addition to Ardery and Odbert Islands (van Franeker et al. 1990). Subsequently, detailed surveys were conducted in 2002/2003 to accurately locate and census these colonies. Here, we present the results of these systematic surveys; we provide preliminary estimates of the snow-petrel breeding population size in the Windmill Islands and more accurate distribution maps that present baseline data useful for future population assessments and monitoring.
Materials and methods The Windmill Islands region comprises 4 large peninsulas and more than 30 islands (Fig. 1). Snow petrels nest in cavities under rocks,
258 Table 1 Summary of existing data on snow-petrel distribution and abundance in the Windmill Islands Location
Status
Date
Source
Ardery I.
300 nests
1972
200–300 pairs >1,000 pairs
1985 1984–1987
300 nests
1972
150 pairs >100 pairs
1985 1984–1987
Ford I.
Reported breeding Reported breeding Reported breeding 100 nests
1974 1989–1990 2001 1972
Herring I.
>200 nests
1972
Holl I.
Reported breeding Scattered nests
1984–1985 1972
Reported breeding, a few pairs Colonies located Reported breeding Reported breeding
1972
Clark and Bailey Peninsulas Bailey Peninsula
Reported breeding ca. 50 nests
1974 1972
Casey Casey Casey Casey
ca.12 pairs ca. 70 pairs ca. 5 pairs 10 pairs
1968 1989–1990 1989–1990 1989–1990
Murray and Luders (unpublished ANARE map 1972) Bonner and Lewis-Smith (1985) Orton (1963); Cowan (1981); van Franeker et al. (1990) Murray and Luders (unpublished ANARE map 1972) Bonner and Lewis-Smith (1985) Orton (1963); Cowan (1981); van Franeker et al. (1990) Jones (unpublished ANARE map 1974) Woehler, unpublished data Woehler and Olivier, unpublished data Murray and Luders (unpublished ANARE map 1972) Murray and Luders (unpublished ANARE map 1972) van Franeker, pers. comm. Murray and Luders (unpublished ANARE map 1972) Murray and Luders (unpublished ANARE map 1972) Jones (unpublished ANARE map 1974) van Franeker, pers. comm. Murray and Luders (unpublished ANARE map 1972); Jones (unpublished ANARE map 1974) Jones (unpublished ANARE map 1974) Murray and Luders (unpublished ANARE map 1972) McKenzie 1968, unpublished map Murray and Luders (1990) Murray and Luders (1990) Murray and Luders (1990)
Odbert I.
Frazier Is.
Peterson I. and Browning Peninsula
Pigeon, Cronk, O’Connor, Werlein and Cloyd Islands
station station station station
area, area, area, area,
Old Casey Reeves Hill Whitney Point Budnick Hill
and consequently only ice-free areas were surveyed. All islands and peninsulas in the region contain some ice-free areas (generally more than 50% in midsummer, Table 2), which constitute potential breeding habitat for snow petrels. Any rocks under ice after late December are unsuitable breeding habitat as cavities would become available too late for nesting. Surveys were conducted from October 2002 to February 2003. Variations in habitat availability were taken into account in relation to the breeding season. The survey area was divided into a 200 m·200 m grid based on the distribution of ice-free areas identified on aerial photography from January 1994 (Fig. 1). For this survey, it was assumed that these ice-free areas would remain reasonably consistent from year to year; this was confirmed during the 2002/2003 surveys. Due to their remoteness (more than 15 km from Casey Station), outer islands such as the Frazier Islands and Donovan Islands were not surveyed. From the grid, a total of 1,050 grid squares comprising more than 1 ha of ice-free land (i.e. a quarter of the grid-square area) was extracted. This selection was conducted in order to avoid travelling to grid squares that contained no potential nesting habitat for snow petrels. The 1,050 grid squares were identified with a unique number, and a subset of 200 grid squares to be surveyed was selected with a standard worksheet random function. If one of the selected grid squares was covered in snow or judged inappropriate for surveys due to steep or inaccessible cliffs, or contained a lake, it was replaced by a grid square adjacent to the original grid square. Such grid squares were moved by no more than 200 m within the grid, so that the replacement grid square contained part of the original grid square or similar habitat (topography). Ten replacements occurred. The number of 200 randomly selected sites was planned to largely exceed the number of sites effectively surveyed by the end of the season. Thanks to good working conditions, 196 of the randomly selected grid squares were surveyed (Olivier 2003).
1974 1984–1985 1972–1974
The search of a grid square was considered complete when the observer was satisfied that all nests were located within the grid square. The same observer conducted all exhaustive grid-square searches throughout the season. Search time ranged from half an hour to 5 h depending on site topography, geomorphology and nest density. Depending on weather conditions, two to five sites were searched in a day. Individual nests were mapped using Garmin 26 GPS receivers and Arcpad software on pocket PC computers. Nests were defined as an area with a clearly recognizable nest cavity. Active nests were located using visual clues only in order to prevent introducing a bias from hearing birds calling from their nest, and to minimize the contribution of varying activity levels throughout the breeding season. Snow conditions and occupation indicators were collected at each nest. Old nests [occupied during previous season(s)] were recognizable, but distinguished from nests occupied within the year, and were not included in analyses. Current nests were clearly identified with the presence of a bird or signs of recent activity, such as stomach oil spat on surrounding rocks, guano, footprints, presence of an egg or chick, etc. Only nests occupied at some time during the breeding season were considered to belong to a potential breeding pair and these were used to estimate the regional breeding population. Three methods were used to estimate the regional breeding population of snow petrels. The first estimate of population size was obtained by assuming that the total number of snow petrels present at a given locality (island or peninsula as shown in Table 2) c
Fig. 1 Map of the Windmill Islands showing ice-free areas and the 200 m·200 m grid from which survey sites were randomly selected
259
3153.3
618.5
8065.4
54.3 190.2 187.0 418.0 69.9 71.1 40.8 185.8 327.0 367.4 1159.8 124.4 3195.7
160.0 1038.7 736.8 66.0 2088.8 193.4 158.2 88.3 208.1 0.0 131.4 4869.7
Total land area (incl. permanent ice) (ha)
39
71 68 29 60 67 63 63 57 65 73 44 57 60
7 40
33 31 24 67 15 49 56 65 52
Proportion of ice-free land in area (%)
17.5
26.9 27.5 23.4 25.7 21.7 0 19.0 17.7 25.0 24.3 21.9 14.7 20.7
0 20.4 10.0 19.6 15.2 21.0 26.4 17.2 14.3 9.3 0 14.0
Proportion of available (ice-free) habitat searched (%)
b
Comprises Boffa, Birkenhauer, Teigan, Wilson and associated islands. Nests from the study colony of Reeves Hill and Casey station surroundings are included in this number.
38.7 129.6 54.7 251.0 47.1 44.8 25.8 106.4 214.1 267.3 516.1 71.3 1766.9
10.4 35.7 12.8 64.6 10.2 0.0 4.9 18.8 53.6 65.0 112.9 10.5 399.4
a
52.2 321.7 179.0 44.1 304.5 94.2 88.2 57.5 108.8 126.7 9.5 1386.4
0.0 65.8 18.0 8.6 46.4 19.8 23.3 9.9 15.5 11.8 0.0 219.1
Swain Group Clark Peninsula Bailey Peninsula Shirley I. Mitchell Peninsula Beall I. Pigeon I. Warrington I. Hollin/Midgley Is. Moraines Nunataks Total Northern Windmill Is. Ardery I. Odbert I. Robinson Ridge Holl I. O’Connor I. Werlein I. Cloyd I. Ford I. Herring I. Peterson I. Browning Peninsula Browning Is.a Total Southern Windmill Is. Grand totals (or average)
Total area of ice-free habitat (ha)
Area searched (ha)
Locality
196
3 10 4 20 3 0 2 6 17 19 34 3 121
0 23 7 4 16 6 7 3 5 4 0 75
No. of grid squares searched in locality
18456
622 1092 2137 2815 4575 760 15014
118 193 535 685 1001 112 3429 4034
752 824 26 1084 327
3442
259 329 61 1103 452 470 354 414 0
Estimated number of active nests in locality
202 227 6 279 71
605
53 33 12 168b 95 124 61 59 0
No. of active nests found in grid squares
5.82
4.9 24.1 7.9 9.0 7.4 9.3 8.6
16.8 5.7 0.4 3.5 5.9
2.4
0.6 1.2 0.8 2.6 4.0 4.4 5.1 3.0 0
Density of active nests vin grid squares (no. nests/ha)
Table 2 Summary of the areas searched, number of nests located and associated estimates of snow-petrel population size at each locality in the Windmill Islands from the 2002/2003 season
260
261 is proportional to the number of birds found in the grid squares at this locality. The coefficient of proportionality used is the fraction of the total area of ice-free habitat that was searched as grid squares at each locality. To minimize the errors for this calculation, care was taken to maintain proportionality between the number of sites searched and the total area of available ice-free habitat at each locality. The regional estimate is the sum of the locality estimates. Alternatively, it is possible to make this calculation with the overall proportion of ice-free areas searched and the total number of nests found in grid squares for the entire Windmill Islands region. The second estimate was derived from the mean snow-petrel densities per locality based on 200*200 m grid-squares searches. Locality densities were averaged with an arithmetic mean and, assuming that the randomly selected grid squares are a representative sample of the available habitat, 95% confidence limits were calculated on the proportion of ice-free area surveyed to total icefree area available. The third method was identical to the second, but used a geometric mean with a 95% confidence limit; this produced a more conservative estimate (see Discussion).
Results Very few data on distribution and abundance of snow petrels in the Windmill Islands are available from previous surveys (Table 1). Only two regional surveys have attempted to map snow-petrel colonies and provide local estimates of breeding populations at ice-free localities throughout the Windmill Islands (Murray and Luders, unpublished work; Jones, unpublished work). During the 2002/2003 season, a total of one hundred and ninety-six 200 m·200 m grid squares were surveyed. In these grid squares, 4,034 snow-petrel nests were located. Snow-petrel nests were found throughout the Windmill Islands, consistent with earlier studies (Table 1). There were noticeable differences in the distribution of nests: 669 nests (17% of the nests found in surveys) in 75 grid squares in the northern Windmill versus 3,365 nests located in 121 grid squares (83% of nests) in the southern Windmill Islands (Fig. 2). The numbers of snow-petrel nests found during surveys at each locality are summarized in Fig. 2 and Table 2. On average, 17.5% of the ice-free areas were surveyed (range 5–54%). The total estimate of the snow-petrel population size was extrapolated from the number of nests found at each locality (Table 2). Snow-petrel nests were found in 160 (81.6%) of the 196 grid squares surveyed and were distributed throughout the entire Windmill Islands region (Table 2). Snow-petrel nests were absent in 36 grid squares (18.4%). A mean of 20.5 nests per grid square was recorded in the 160 grid squares where snow petrels were present. Nest densities were highly variable, ranging from 1 to 238 per grid square (very high density). Fewer than 10 nests were located in 30.6% of the grid squares where snow-petrel activity was recorded. Empty grid squares were found also throughout the entire region but were more frequent in northern localities (Fig. 1). Densities were generally lower in the northern Windmill Islands (Fig. 3). The number of snow-petrel nests found per grid square in the northern areas ranged between 0 and 45 (with an average density of 2.6 nests/ha). Nest
density varied across the entire range of densities in the southern Windmill Islands, with 12% of the grid squares holding between 50 and 238 active nests (with an average density of 8 nests/ha).
Discussion Limitations of the study This study is the first to estimate the breeding population of snow petrels at a regional scale within the Australian Antarctic Territory based on empirical data. This regional estimate is more accurate than previously published work derived from limited areas within the Windmill Islands (Table 1). Reliable breeding-population estimates were only available for Ardery and Odbert Islands (a total of approximately 1,100 breeding pairs: van Franeker et al. 1990). Furthermore, snowpetrel estimates vary considerably among authors even within a limited area such as Ardery and Odbert Islands (Table 1), due largely to observer variability. It is also due to the lack of a standard survey methodology and set criteria defining what constitutes an active nest and how to account for breeding pairs from these criteria. Although we attempted to ensure that data collection was as exhaustive and consistent as possible, we encountered obstacles relating both to the spatial and temporal scales of the survey. First, a number of localities were not surveyed (nunataks, Werlein Island and the Swain Group). At other localities, the survey coverage was relatively limited (moraine: only 9.3% of the habitat was searched). When calculating the average proportion of available habitat searched (17.5%), we included these areas as ‘‘0% search effort’’ in order to be able to estimate the overall population size without knowing the nest densities within these areas. However, the total area of localities that were not surveyed represents only a small proportion of the available ice-free habitat in the Windmill Islands (approximately 7.2%). It was also known from earlier visits (2001/2002) that some of the unsurveyed areas were virtually empty of snow petrels (nunataks and moraine, F. Olivier, unpublished data). Thus we believe that the error introduced in the calculations of total population estimate is likely to be minimal. Temporal variations in snow-petrel activity The behaviour of snow petrels at their nests varies throughout the season and thus affects nest detection by observers. During the initial nest reoccupation period, the birds are very active and call when approached. In the following phase of the pre-egg exodus, the colonies are virtually empty, but nests show signs of prior occupation. After egg laying, snow petrels no longer call at the nest. Active nests become therefore less obvious to
262
263 b
Fig. 2 Distribution of grid squares and total number of active snow-petrel nests found at each locality during 2002/2003 survey. The density gradient in the number of nests found per grid square is also represented
observers. The survey protocol used in this study reduced this potential bias by locating active nests visually rather than using auditory cues alone. Although nest detection is facilitated by bird activity in the early nestsettlement period, it may lead to over-estimating breeding populations as some nests may only be briefly occupied for roosting, or young birds may only be Fig. 3a–c Frequency distribution of the number of active snow-petrel nests located in grid squares for the Northern Windmill Islands (a), for the Southern Windmill Islands (b) and for the entire Windmill Islands region (c)
prospecting for a nest and do not lay eggs. From early in the season until mid-late November, the extensive snow cover may also confound nest recognition and counts due to multiple digging attempts at various entrances to a single nest. Careful observations and searches will prevent excessive counts at the cost of increased search effort. Based on the 2002/2003 season, the first 2 weeks of December were identified as the most appropriate time to conduct surveys of nesting snow petrels in the Windmill Islands. The large area in which surveys were conducted (approximately 80 km2, Table 2) did not permit a census of all localities within this period.
264
van Franeker et al. (1990) attempted to relate their estimates of Apparently Occupied Sites to the actual breeding population (birds that laid eggs) using a proportionality coefficient. In our study, nests with no recent signs of occupation during the 2002/2003 season were not used for estimating breeding numbers. Because a breeding pair may attend the nest, attempt to breed but not lay an egg, only nests with a bird or a nest with recent signs of activity (such as stomach oil or guano) were considered to be active. Further, it is known that some nonbreeding birds roost within the colonies and thus only a fraction of the snow-petrel population is breeding in a given year (Chastel et al. 1993). However, owing to the longer time frame over which surveys were conducted (5 months), we considered it more accurate to include all active nests in the calculation of population estimates
available locally on each of the islands or peninsulas surveyed. A detailed habitat analysis using digital elevation models and other geomorphologic parameters is currently underway to determine whether this assumption is realistic. The random selection of survey sites allowed sampling across all types of terrain and habitats, but grid squares located within high-density colonies may lead to a local overestimation of the total number of nests. As the surface of available habitat varies with habitat complexity among grid squares (hill slopes), further work is required to calculate the effective area that was searched, and interpret nest densities in light of habitat characteristics. Future investigations will also explore more thoroughly habitat preferences of snow petrels in order to refine the regional estimate of snowpetrel population size.
Estimates of the breeding population of snow petrels in the Windmill Islands
Conclusion
Three estimates of the regional breeding population of snow petrels in the Windmill Islands were made. The first estimate was obtained by summing snow-petrel populations estimated at each locality. This total of 18,456 snow petrels incorporates slight variations in search effort among localities (number of grid squares searched at each locality) and locality-specific differences that influence the number of birds found locally (habitat). The alternative calculation that uses the overall proportion of ice-free land (17.5%, Table 2) incorporates localities that were not searched and gives a higher number: 23,112 breeding pairs. No confidence limits can be generated for these estimates. Therefore, it was judged more appropriate to calculate estimates of the snow-petrel population size from the nest densities (given in number of nests per hectare in Table 2) averaged among grid squares at each locality. Using the arithmetic mean of nest densities in grid squares across all localities, the second estimate of population size for the Windmill Islands is 18,361 breeding pairs (with 10,250 and 26,471 nests as lower and upper 95% confidence limits, respectively). It has to be noted that this estimate averages the density of snow petrels across the entire Windmill Islands region whereas our data clearly show that nest densities are higher in the southern Windmill Islands than in the northern Windmill Islands. As the frequency distribution of snow-petrel densities per locality is heavily skewed (Fig. 3), using the geometric mean of nest densities provides a more conservative estimate of the regional population: 12,218 breeding pairs (with 8,747 and 15,689 as lower and upper 95% confidence limits, respectively). Habitat These regional estimates of the snow-petrel population in the Windmill Islands assume that the sampled grid squares are representative of the nesting habitat
This study provides a preliminary regional estimate of snow-petrel population size in the Windmill Islands and more detailed data on the local distribution of snow petrels as a basis for future and more detailed research. Long-term population trends are better monitored in small and well-defined areas with repeated observations, such as the study colony of Reeves Hill near Casey Station or the colonies of Ile des Pe´trels in the vicinity of Dumont d’Urville (Micol and Jouventin 2001). However, regional data on the location and size of snow-petrel breeding populations around the Antarctic continent are generally lacking even in many easily accessible areas in the vicinity of bases (Croxall et al. 1995). Integrated with local studies, regional surveys are a starting point in the collection of reliable baseline information that will contribute to a better conservation and management of this Antarctic species and add to our ability to monitor future changes. Acknowledgements We thank George Jackson, Katrina Phillips and Samantha Lake for useful comments on the manuscript. Part of this study was supported by a grant from the Australian Geographic Society. The survey is one part of an ongoing program on long-term monitoring of bird population supported by the Australian Antarctic Science Advisory Committee (Project 1219). The Australian Antarctic Division supplied logistical support for the survey. We thank all Casey expeditioners from 2001/2002 and 2002/2003 summers for their assistance and support.
References Barbraud C, Weimerskirch H, Guinet C, Jouventin P (2000) Effect of sea-ice extent on adult survival of an Antarctic top predator: the snow petrel Pagodroma nivea. Oecologia 125:483–488 Bonner WN, Lewis Smith RI (1985) Conservation areas in the Antarctic. SCAR/ICSU, Cambridge Brown DA (1966) Breeding biology of the snow petrel Pagodroma nivea (Forster). ANARE Sci Rep B Zool 1:1–63 Catry P, Campos A, Segurado P, Silva M, Strange I (2003) Population census and nesting habitat selection of thin-billed prion Pachyptila belcheri on New Island, Falkland Islands. Polar Biol 26:202–207
265 Chastel O, Weimerskirch H, Jouventin P (1993) High annual variability in reproductive success and survival of an Antarctic seabird, the snow petrel Pagrodoma nivea. Oecologia 94:278–285 Cowan AN (1981) Size variation in the snow petrel (Pagodroma nivea). Notornis 28:169–188 Croxall JP, Steele WK, McInnes SJ, Prince P (1995) Breeding distribution of the snow petrel Pagodroma nivea. Mar Ornithol 23:69–99 Franeker JA van, Bell PJ, Montague TL (1990) Birds of Ardery and Odbert islands, Windmill Islands, Antarctica. Emu 90:74–80 Johnstone GW, Lugg DJ, Brown DA (1973) The biology of the Vestfold Hills, Antarctica. ANARE Sci Rep B Zool 123:1–62 Micol T, Jouventin P (2001) Long-term population trends in seven Antarctic seabirds at Pointe Ge´ologie (Terre Ade´lie): human impact compared with environmental change. Polar Biol 24:175–185
Monaghan P (1996) Relevance of the behaviour of seabirds to the conservation of marine environments. Oikos 77:227–237 Montevecchi WA (1993) Birds as indicators of change in marine prey stocks. Birds as monitors of environmental change. Chapman & Hall, London Olivier F (2003) Detailed information on 196 grid squares used for snow petrel surveys in the Windmill Islands during the 2002/ 2003 season. Australian Antarctic Data Centre—SnoWhite Metadata (http://www-aadc.aad.gov.au/metadata/) Orton MN (1963) A brief survey of the fauna of the Windmill Islands, Wilkes land, Antarctica. Emu 63:15–22 Woehler EJ, Johnstone GW (1991) Status and conservation of seabirds of the Australian Antarctic Territory. In: Croxall JP (ed) Seabird status and conservation: a supplement. ICBP Tech Publ 11:279–308
