1. Self-thinning, the reduction in density (N) as a consequence of the increase in the mean weight (W) of individuals, occurs widely in populations of plants when their growth is constrained at high densities, but has only recently been reported to occur within populations of wild mobile animals. It has been suggested that a self-thinning relationship in which the gradient (Delta log(W)/Delta log(N)) approximates to -1.33 describes concomitant changes in W and N within cohorts of a population of anadromous brown trout (sea trout) Salmo trutta throughout their lives as juveniles in fresh water. This apparent self-thinning occurs with no evidence of density-dependent growth or mortality (as measured by key-factor analysis) after the critical period. 2. Here, the relationship between weights and densities of trout has been examined further. A linear model incorporating log(N), year class and time of year explained 99.5% of the variation in log(W), giving a significantly better fit than incorporating only log(N) and year class. Variation in log(N) explained 85% of the variation in log(W); the addition of year class increased this value by 6% and the addition of monitoring period increased it by a further 8.5%. 3. The gradients for the relationships between log(W) and log(N) within years varied significantly with time of the year. pooled thinning trajectories for the first winter of the life of trout differed significantly from a gradient of -1.33, while those for the second summer were highly variable and sometimes positive (suggesting that immigration could exceed mortality). Thinning trajectories over the first summer of life were related inversely to the densities of trout at the start of the monitoring period in that summer (N-s), but this relation appeared to approach an asymptote at high values of N-s. The mean of the thinning trajectories for cohorts with high N-s (exceeding 300 fish 60 m(-2)) was not significantly different from -1.33. However, these slopes were shown to result primarily from the presence of high numbers of failed nonterritorial fish early in the summer rather than true thinning throughout the summer. 4. The apparent absence of a simple consistent thinning relationship after a critical period in the first couple of months of the lives of cohorts can be explained by three factors. First, the total availability of limiting resources, such as space and food, is not constant but changes with the mean size of the trout. Secondly, there is a winter cessation of growth. Thirdly, trout switch to use different habitats during their development. The implications for interpreting thinning relationships for salmonid fishes in general are discussed.
