Male siskins (C. spinus
) used in the experiments (n
= 29) were captured in northeast Spain in the autumn and winter of 2007–2008. Birds were separated in groups of four and kept in cages provided with eight feeders spaced along 1 m. This prevented competition for food within the group. To estimate overall patch size, we measured the length of the yellow stripe on the primary wing feather six (P6) (descendent) from the distal edge of primary covert four to the distal end of the yellow colouring, according to Senar & Escobar [5
]. This measurement is strongly correlated with the size of the whole yellow patch [5
]. Siskins also show a black badge that reflects their dominance within the social group. The size of this black badge is uncorrelated with the yellow wing strip length [6
]. We also measured the size of this black patch, following Senar et al
]. Four hours before the trial, birds were isolated and food-deprived to foster motivation. After fasting, a small feeder containing pine seeds was placed on an elevated platform in each test cage. Access to the pine seeds was partially blocked by toothpicks that were crisscrossed from side to side through lateral holes in the feeder. The toothpicks protruded 4 cm on each side. The birds could see the food, but they were unable to reach it without removing at least one of the sticks. They could do this by moving them sideways, until one end of the stick slid out of the side hole in which it was inserted.
An observer measured the time spent by each bird from the moment they landed on the platform or on the feeder trying to get the food, until the moment when they solved the problem, reaching the pine seeds. If the bird did not go on the platform within 3 min, the trial was considered ended. The maximum time set to solve the problem once birds reached the platform was 5 min. Each bird was observed for a maximum of five trials. If, for instance, a bird got the food source in the third trial, after 62 s, we computed 300 + 300 + 62 s as the total time needed to solve the problem.
We analysed, using generalized linear models (GLM), differences in the length of the yellow wing stripe, measured on primary six (n
= 29), between slow and fast problem solvers, including age (yearling versus adult) as a covariate. We defined slow solvers as birds that either did not solve the problem or needed more than one trial to solve the problem (greater than 300 s). Fast solvers were defined as birds that solved the problem within the first trial (less than 300 s). Time to solution was then related to the length of the yellow wing stripe, age, size of the black badge and size of the bird (as measured from tarsus length), using a Cox proportional hazards regression model (n
= 21). All analyses were implemented in the R statistical computing environment, v. 2.12.2 [8
]. Cox regressions were applied using the survival