Introduction
Many albatrosses and large petrels are threatened by mortality on fishing gear (BirdLife International 2004, 2008). In the Southern Ocean and adjacent temperate oceans, the large petrels of the genus Procellaria dominate the bycatch of most fisheries (e.g. Barnes et al. Reference Barnes, Ryan and Boix-Hinzen1997, Nel et al. Reference Nel, Ryan and Watkins2002a, Delord et al. Reference Delord, Gasco, Weimerskirch and Barbraud2005, Waugh et al. Reference Waugh, MacKenzie and Fletcher2008, Petersen et al. Reference Petersen, Honig, Ryan and Underhill2009) and all five species are listed as Threatened or Near Threatened (BirdLife International 2008). Mitigating the impacts of fishing gear on these birds is complicated by their proficiency as divers (Robertson et al. Reference Robertson, McNeill, Smith, Wienecke, Candy and Olivier2006). Among Procellaria petrels, two species have widespread breeding ranges, whereas three are confined to one or a few sites. The spectacled petrel or ringeye Procellaria conspicillata Gould only breeds on Inaccessible Island (37°18′S, 12°41′W), one of three main islands in the Tristan da Cunha archipelago in the South Atlantic Ocean (Ryan Reference Ryan1998, Ryan & Moloney Reference Ryan and Moloney2000, Ryan et al. Reference Ryan, Dorse and Hilton2006). It is well protected at its breeding island, which lacks introduced predators and is a nature reserve and natural world heritage site (Ryan & Glass Reference Ryan and Glass2001, Ryan Reference Ryan2007), but spectacled petrels regularly are killed on longlines, mainly off the east coast of South America (Bugoni et al. Reference Bugoni, Mancini, Monteiro, Nascimento and Neves2008a, Reference Bugoni, Neves, Leite, Carvalho, Sales, Furness, Stein, Peppes, Giffoni and Monteiro2008b, Jiménez et al. Reference Jiménez, Abreu, Pons, Ortiz and Domingo2010), but also occasionally off South Africa (Ryan et al. Reference Ryan, Dorse and Hilton2006). Catch rates generally are low (Bugoni et al. Reference Bugoni, Mancini, Monteiro, Nascimento and Neves2008a, Jiménez et al. Reference Jiménez, Abreu, Pons, Ortiz and Domingo2010), despite spectacled petrels being the most abundant bird attending fishing vessels off Brazil (Bugoni et al. Reference Bugoni, Mancini, Monteiro, Nascimento and Neves2008a), but they are the species killed most frequently on handlines off Brazil and are also killed by vessels trolling for tunas (Bugoni et al. Reference Bugoni, Neves, Leite, Carvalho, Sales, Furness, Stein, Peppes, Giffoni and Monteiro2008b). The species was listed as Critically Endangered (BirdLife International 2000, 2004) until successive surveys using the same techniques in 1999 and 2004 showed that the population had increased at roughly 7% per year (Ryan et al. Reference Ryan, Dorse and Hilton2006). Indeed the population appears to have increased at this rate since the 1930s, following its near extinction due to predation by introduced pigs (Ryan et al. Reference Ryan, Dorse and Hilton2006). Fortunately the pigs died out, probably between 1900 and 1930 largely due to hunting pressure (Fraser et al. Reference Fraser, Ryan and Watkins1988, Ryan & Glass Reference Ryan and Glass2001), before they were able to extirpate the petrels.
The spectacled petrel remains listed as Vulnerable given its small breeding range confined to a single island (BirdLife International 2008). Proposed conservation measures include ongoing monitoring of the species’ population (BirdLife International 2008), which is necessary to confirm that the population continues to recover. In this paper we report a third successive five-yearly survey of spectacled petrels at Inaccessible Island in 2009.
Methods
We visited Inaccessible Island from 5 October–1 December 2009. The total number of spectacled petrel burrows was estimated during the early incubation period between 29 October and 4 November. The survey was led by PGR, who also led the 1999 and 2004 surveys, ensuring consistency in survey techniques. We surveyed all areas where spectacled petrels are known to breed (Ryan & Moloney Reference Ryan and Moloney2000, Ryan et al. Reference Ryan, Dorse and Hilton2006). Counts were divided into the 20 count areas proposed by Ryan (Reference Ryan2006), but results are reported for the 14 zones used in both previous surveys (Fig. 1). Spectacled petrel colonies are readily detected because the birds regularly arrive and depart from their burrows during daylight (Ryan et al. Reference Ryan, Dorse and Hilton2006). Most burrows are easily distinguished from those of other petrels at Inaccessible Island by their large size, often with an entrance pool or moat (Rowan et al. Reference Rowan, Elliott and Rowan1951, Hagen Reference Hagen1952, Ryan & Moloney Reference Ryan and Moloney2000). Most burrows are concentrated in distinctive ‘ringeye bogs’, marshy areas dominated by Scirpus sulcatus Thouars that are created by the birds’ burrowing activity. However, some breed along stream banks or under dense vegetation where burrows are easily overlooked unless the adults call (Ryan et al. Reference Ryan, Dorse and Hilton2006).
Fig. 1 The distribution of spectacled petrel burrows at Inaccessible Island. Red represents areas where marked range expansions occurred between 2004 and 2009. Count areas reported in Table I are labelled A–N. Dashed lines are c. 100 m contours.
After completing the burrow survey, we checked observer accuracy on 6 November 2009 by independently estimating the number of burrows at five colonies in different count zones to assess the relationship between estimated and actual numbers of burrows (see Ryan & Moloney Reference Ryan and Moloney2000, Ryan et al. Reference Ryan, Dorse and Hilton2006 for details). Nest occupancy was assessed at the same sites using the criteria used in previous surveys (Ryan & Moloney Reference Ryan and Moloney2000, Ryan et al. Reference Ryan, Dorse and Hilton2006). Spectacled petrel calls were played down the burrow entrance for 10–30 sec, ceasing if a response was obtained. The recording used was of a pair of birds giving a combination of rattles, wheezes and groans (Ryan Reference Ryan1998), and so should elicit responses from a high proportion of incubating birds (a combination of wheezy and rattle calls elicited responses among 89% of incubating white-chinned petrels, Procellaria aequinoctialis L., Berrow Reference Berrow2000). If there was no response (24% of burrows), an arm was inserted down the burrow as far as could be reached (c. 1 m), and the burrow scored as occupied (adult and/or egg felt, or adult called), empty, or uncertain (where nothing was detected, but the burrow was too long to be checked completely). Comparisons of numbers of birds responding in different periods were made using chi-squared goodness of fit tests with Yates’ correction for continuity.
Results
Spectacled petrel burrows were found in all areas where they were previously recorded to breed, but their range had expanded, especially around ‘Denstone Hill’ and Round Hill, isolated colonies on high points on the eastern plateau (Table I, Fig. 1). Smaller range expansions also were recorded around the periphery of core breeding areas on the western plateau. The total breeding range was slightly >2 km2. A total of 13 765 burrows was counted, 55% more than in 2004 (Table I). Numbers of burrows increased in all sectors. The greatest proportional increases (two- to threefold) occurred at Denstone Hill, Round Hill and the south-eastern slopes of Long Ridge (Table I), all at the eastern edge of the species’ breeding range (Fig. 1).
Table I Counts of spectacled petrel burrows at Inaccessible Island (see Fig. 1 for count areas).
The five count validation areas contained 146 burrows (n = 18–36, Table II). The two observers estimated 89% (PGR) and 94% (RAR) of the total number of burrows (average 91%), with estimates within each area ranging from 77–113% of the actual number of burrows. Combined count accuracy was 90.2 ± 5.4% (±s.e., n = ten estimates). Adjusting for counter bias suggests that the total number of burrows was 15 050 (95% confidence interval 13 500–16 600, Table II). This estimate fails to account for burrows not detected amongst dense vegetation. Assuming that at least 5% of nests are overlooked (Ryan et al. Reference Ryan, Dorse and Hilton2006), there are probably 15 800 (14 200–17 400) spectacled petrel burrows on Inaccessible Island.
Table II Population estimates for breeding spectacled petrels in three successive five-yearly surveys at Inaccessible Island.
*based on careful checks of subsets of the count area, see methods for details.
+assumes 5% missed in 2004 and 2009, based on data from repeat surveys of study areas in 2004, and 10% missed in 1999, given later census period (when adults are less vocal) and reduced observer experience.
Response to playback during one-off occupancy surveys was uniformly high in the five study areas (average 76%, range 69–85%), with an additional 5% (0–9%) of nests proving to be occupied when birds were detected when an arm was inserted into the burrow. Coupled with a few known breeding failures (broken eggs in burrow entrances), average occupancy was 81% (range 72–85%). Most other burrows were confirmed empty (14%), with only 5% of burrows too long to check their contents. The occupancy rate was slightly higher than the one-off estimate in 2004 (77%), but the difference was not significant (χ2 = 0.70, df = 1, P > 0.2). The increase in the proportion of burrows where there was a response to playback was more marked between 2009 (76%) and 2004 (68%, n = 311, χ2 = 2.81, df = 1, P < 0.1), which might be expected given that the check in 2009 occurred slightly earlier (6 November) than in 2004 (11 and 12 November). Response to playback decreases throughout the breeding season after egg laying in late October (Ryan et al. Reference Ryan, Dorse and Hilton2006).
If 81% of burrows were occupied, the population of spectacled petrels in 2009 was 12 800 pairs (Table II). However, multiple checks of marked burrows in 2004 indicated that occupancy was 91%, 14% greater than that recorded during one-off surveys (Ryan et al. Reference Ryan, Dorse and Hilton2006). Assuming a 91% occupancy rate in 2009 gives a total population of 14 400 breeding pairs (Table II) and an annual average growth rate from 2004–09 of just over 7% per year. It gives a total population of roughly 30 000 adult birds in 2009.
Discussion
The 2004 spectacled petrel survey suggested that the population was increasing at about 7% per year (Ryan et al. Reference Ryan, Dorse and Hilton2006). However, this growth rate may have been inflated in part by improved observer experience and greater search effort in 2004 (Ryan et al. Reference Ryan, Dorse and Hilton2006). A simple deterministic model suggested that numbers of spectacled petrels could increase at around 7% per year, but it was sensitive to the magnitude and timing of fishing mortality as well as the way in which fishing mortality changes in relation to the population size and density of spectacled petrels at sea (Ryan et al. Reference Ryan, Dorse and Hilton2006). Given these uncertainties, it was important to conduct another survey to confirm that the population was indeed increasing despite ongoing mortality on fishing gear.
The 2009 survey confirms that the population of spectacled petrels has continued to increase, and the 7% per year growth rate over the last five years is the same as that estimated for the period 1999–2004. Observer biases are unlikely to account for the increase from 2004–09, given that all three surveys were led by the same observer. The increase was apparent in the larger numbers of birds visiting the island in 2009 compared to 2004 and 1999. In the late afternoon it was not unusual to see up to 50 birds sitting outside their burrows on a single ringeye terrace, and to have thousands of birds soaring over the island's ridges. Estimates of the numbers of burrows increased throughout the species’ breeding range, but the fact that the increases were greatest at the periphery of its range suggest that favoured breeding sites may be limited in core areas.
The accuracy of burrow estimates and burrow occupancy rates were similar to those obtained in 2004 (Ryan et al. Reference Ryan, Dorse and Hilton2006). Our trials confirm that a large proportion of Procellaria adults respond to playback during the early incubation period (Berrow Reference Berrow2000, Ryan et al. Reference Ryan, Dorse and Hilton2006), which is the best time to conduct surveys of these species. Berrow (Reference Berrow2000) found marked differences in response rates of white-chinned petrels in successive years at Bird Island, South Georgia, but this apparently was due to differences in occupancy linked to colder weather and heavy snow cover in one year. Such interannual differences are less likely to occur at temperate Inaccessible Island, but a larger sample of years is necessary to confirm that occupancy rates are consistently high in all years.
The growth of the spectacled petrel population apparently results from the recovery of the species after its population was impacted by an introduced predator (Ryan et al. Reference Ryan, Dorse and Hilton2006). The elimination of introduced predators, in this case pigs, is essential for the recovery of many endangered seabird species (e.g. Rayner et al. Reference Rayner, Parker and Imber2008). However, the rate of recovery is perhaps surprising given ongoing mortality of spectacled petrels on fishing gear mainly off the east coast of South America (Bugoni et al. Reference Bugoni, Mancini, Monteiro, Nascimento and Neves2008a, Reference Bugoni, Neves, Leite, Carvalho, Sales, Furness, Stein, Peppes, Giffoni and Monteiro2008b, Jiménez et al. Reference Jiménez, Abreu, Pons, Ortiz and Domingo2010). The marine distribution of spectacled petrels is concentrated along the shelf break and deeper offshore waters off Brazil and Uruguay (Camphuysen & Van der Meer Reference Camphuysen and van der Meer2000), where they overlap extensively with pelagic longline fisheries (Bugoni et al. Reference Bugoni, D'Alba and Furness2009). One reason for their relatively rapid growth may be a rather low susceptibility to capture on longlines. Despite being the most abundant seabird attending longline vessels off Brazil, spectacled petrel capture rates are low in comparison to white-chinned petrels (Bugoni et al. Reference Bugoni, Mancini, Monteiro, Nascimento and Neves2008a). The absence of juvenile spectacled petrels possibly also contributes to low capture rates in this region, given that juveniles may be more prone to being caught on longlines (Bugoni & Furness Reference Bugoni and Furness2009). Bycatch rates of spectacled petrels in Uruguayan longline fisheries also are low, and total seabird catch rates in this region have decreased over the last decade (Jiménez et al. Reference Jiménez, Abreu, Pons, Ortiz and Domingo2010). For other populations of Procellariformes, population growth and stability has been attributed to decreased fishing effort and associated bycatch rates (Nel et al. Reference Nel, Ryan, Crawford, Cooper and Huyser2002b, Rolland et al. Reference Rolland, Nevoux, Barbraud and Weimerskirch2009, Rivalan et al. Reference Rivalan, Barbraud, Inchausti and Weimerskirch2010) emphasising the importance of fisheries management in the recovery of endangered albatross and petrel populations. The preponderance of spectacled petrels in handline bycatch (and, to a lesser extent, trolling fisheries) off Brazil (Bugoni et al. Reference Bugoni, Neves, Leite, Carvalho, Sales, Furness, Stein, Peppes, Giffoni and Monteiro2008b) remains cause for concern.
Despite the larger population recorded in 2009, the spectacled petrel still qualifies as globally Vulnerable under criterion D2 (population very small or restricted), because the entire breeding population is confined to a single location, with an area <10 km2. Our results suggest that at least some procellariiforms are able to sustain strong growth rates despite limited mortality on fishing gear.
Acknowledgements
We thank the Administrator and Island Council of Tristan for permission to visit Inaccessible Island. Logistical support was supplied by the South African National Antarctic Programme (SANAP), and Tristan's Conservation Department; we especially thank Captain Clarence October and the crew of the MV Edinburgh and Norman Glass. Financial support was provided by SANAP, the European Commission's EDF-9 through the South Atlantic Invasive Species project, and the University of Cape Town. RAR was supported by the Killam Trust, Dalhousie University.
