Evaluation of de la Mare’s composite standardised index for generating a simple time-series summary of many long-term datasets on Antarctic predators: consequences of missing values and criteria for inclusion of predator parameters

In 1997, de la Mare presented a statistic (labelled the Composite Standardised Index at the Workshop on Area 48) to WG-EMM for generating a simple time-series index of many long-term datasets on Antarctic predators. The value of the index in a given year is dependent on the specific parameters present in that year. In this respect, the robustness of the index is dependent on that subset of parameters being representative of all the parameters for that year. Clearly, this will depend on how correlated the parameters are to each other. This paper reports on a Monte Carlo simulation study that helps specify general criteria for the inclusion of predator parameters in the index with regard to the potential effects of missing values as well as the degree of correlation between parameters. These trials found that, for inclusion in the index, parameters should be positively correlated with all other parameters and preferably with correlations greater than 0.3. If this criterion is met then these results indicate that a time series of 3 years for the full matrix of 8 parameters used in this study is sufficient for determining the correlation coefficients. These criteria need to be accepted with caution. Further work is required to take assess the effects of sampling variability on the performance of the index. Also, this work needs to account for the characteristics of the current CEMP database and to assess which of the available parameters might be included in the index. A general challenge is to identify how to incorporate parameters that are highly negatively correlated to the other parameters in analyses of the effects of prey availability and the environment on predators. The latitude given to accepting some parameters in the index needs to be evaluated in the wider context of being able to make correct decisions based on the index despite the uncertainties or variability in the behaviour of those parameters.