A set of 137 Ordovician graptolite species were used to examine the associations among geographic range, sampling, biofacies and species longevity. Model-choice using general linear models combined with partial least-squares regression analysis found seven distinct predictive variables.

The dominant factors were overall commonness, biofacies, geographic range and sampling in decreasing order of variance explained. However, the data-set is biased toward particularly well-sampled and widespread taxa.

Region (represented as a set of discrete geographic areas) was a strong factor in extinction risk, whereas latitudinal range and endemicity were poor predictors. Results suggest that other factors besides just geographic range and biofacies need to be considered when understanding extinction dynamics.