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

Despite the higher frequency of kleptoparasitism and predation in 1996, puffin activity was not similarly heightened as we had predicted. In return-rate counts, mean number of puffins standing on the reference grid squares decreased by 23% in 1996 and number of puffins landing decreased by 40% (Table 2). There were over three times as many return rate counts in 1996 than in 1995 during which no puffins returned (Table 2). Fewer puffins returned with food in 1996, although proportion of returning puffins observed with food did not differ between years (Table 2). However, a greater number of unscored puffins (i.e. we did not know whether or not they were carrying fish) in 1995 may obscure a similar decrease: if all unidentified puffins in both years are assumed to be carrying food (food-laden puffins tend to dart directly into their burrows making it difficult to detect fish), proportion with food actually decreased by as much as 30% (Table 2). Relative to number of returning puffins observed with food in each year, significantly more puffins were approached by gulls in 1996 (Table 2). That equates to an eight fold increase in percentage of food-carrying puffins that were chased or attacked during return-rate counts (Table 2).

Factors affecting kleptoparasitism frequency and success.-Data were tallied for comparing rates of kleptoparasitism in two ways, temporally as hourly totals, and spatially as grid marker totals. In the first subset, we investigated effects of the actual density of returning puffins, predictability of returning puffins, weather conditions, and tidal height on the hourly rates of kleptoparasitism (attacks plus chases) using multiple regression. We measured puffin density as number of puffins returning with food as well as the total number of puffins returning during each observation hour. For "puffin predictability," we used hourly averages of number of puffins returning with fish in each season to generate diel patterns of puffin activity. The best-fit significant relationship between those variables was linear in 1995 (R^sup 2^ = 0.38, df = 1 and 14, P = 0.01) and cubic in 1996 (R^sup 2^ = 0.82, df = 1 and 12, P

Of the seven independent variables analysed, four contributed significantly to variation in the rates of kleptoparasitism activity (Table 4). Kleptoparasitism rates increased with number of puffins actually returning with fish as well as with number predicted to return by hourly averages. Kleptoparasitism also decreased during good weather although it showed no measurable increase during foul weather. Total number of puffins returning with food and tidal height also did not contribute significantly, though those small effects were in the predicted directions. Inclusion of the four significant factors (Table 4) yielded a regression model that explained only 20% of the variance in gull activity (R^sup 2^ = 0.192, F = 21.8, df = 3 and 346, P 0.46). Foul weather alone explained only 1% of the variance in kleptoparasitism success (R^sup 2^ = 0.012, F = 6.06, df = 1 and 406, P = 0.014).