Log-linear models are used extensively in the analysis of cross-tabulated data. Starting with the saturated model, which describes all the complex interrelationships in a multidimensional table and then testing for the significance of higher-order interactions, the simplest model providing a satisfactory explanation of the data can be arrived at. This approach was adopted to identify any discernible quantitative pattern in the data on the average number of fledgelings per nest of osprey in the northeast US for 1969–83. These models are then used to obtain an index of marine ecosystem degradation.