Does Kleptoparasitism by Glaucous-winged Gulls limit the reproductive success of Tufted Puffins?

Auk, The, Oct 2001 by St Clair, Colleen Cassady, St Clair, Robert C, Williams, Tony D

METHODS

The study took place on Triangle Island, the outermost island of the Cape Scott archipelago, 46 km northwest of Vancouver Island, during summers of 1995 and 1996. Most of the breeding puffins on the island nest on the steep, hairgrass (Deschampsia caespitosa) covered slopes of Puffin Rock, a 90 m hummock connected to the rest of the island by a low-tide isthmus. Glaucous-winged Gulls nest along the ridge tops above and among puffin burrows where they can easily survey incoming puffins. Gulls typically kleptoparasitize puffins by waiting until a puffin approaches its burrow and then lunging or flying toward it and attempting to grasp it by the wing, tail or head, and then shaking it, sometimes violently, until it drops its fish. Gulls frequently patrol puffin burrows at other times and probe in and around burrow entrances, apparently looking for displaced eggs or chicks and dropped fish.

Frequency of kleptoparasitism and predation.-Observations of puffin and gull activity took place on one southeast-facing nesting slope of Puffin Rock visible from a blind -100 m away on an opposite-facing slope. To reference observations and burrow locations, we placed a grid over the accessible portion of the southeast slope over an area of approximately 40 x 100 m with flags at 10 m intervals. During the puffin nestling period (30 June-16 August, 1995, and 29 June-15 August 1996), we observed the slope for periods of 1-8 h per day. Those periods were spread approximately equally over the daylight hours (0600-2200 h). We recorded kleptoparasitism and predation by gulls by ad libitum, continuous sampling (Martin and Bateson 1986) of the gridded area, including the air above it. For each kleptoparasitism event, we recorded the following variables: date, time, number of gulls involved, type (attack or chase, see below), location of closest contact between the gull and puffin (air or ground), approximate two-dimensional grid location, and success or failure. We captured and marked three gulls in 1995; two others were identifiable by unique markings in both years, so in some instances we knew the identity of the kleptoparasite. An attack was defined as an event involving contact between the gulls) and puffins, and chases were those without contact. Because it was often difficult to tell when small fish had been dropped, we recorded events as successful only when we observed a gull eating pirated fish and that was easily determined by spotting scope or binoculars. In 1996, we also recorded the escape substrate of the puffin (air or burrow; the water was far below the nesting slope and not easily visible) and interference by other gulls with a primary kleptoparasitizing individual.

Factors influencing kleptoparasitism frequency and success.-To monitor abundance and foraging success of puffins, we counted the number of puffins returning with and without fish once during each hour of observation. For consistency, those data were collected during a 10 min period from two grid squares (10 x 20 m) that were easily observed from the blind. To provide a relative measure of kleptoparasitism frequency, we recorded the number of gulls present in the grid squares during the count and recorded any chases or attacks that occurred. Prior to those counts, we also counted the number of puffins standing on the nesting slope in three (1995) or two (1996) grid squares adjacent to return rate squares. We collected qualitative data on cloud cover (five categories) and obtained Environment Canada weather data from the automated buoy at nearby Sartine Island (14 km to the southeast) for three other variables: temperature, wind speed, and wind direction. We added an additional wind direction variable, "northwest wind direction," to recognize the prevailing weather patterns: 180 for high pressure systems from the northwest grading symmetrically on both sides of the compass to 0 for the stormy weather from the southeast. Finally, we determined tidal heights for each hour of observation using the program XTide (Flater 1996).