There are two fundamental ways of doing science: the experimental-predictive and the historical-descriptive. The experimental-predictive approach uses the techniques of controlled experiment, the reduction of natural complexity to a minimal set of general causes, and presupposes that all times can be treated alike and adequately simulated in the laboratory. The historical-descriptive approach uses a mode of analysis which is rooted in the comparative and observational richness of our data, is holistic in its treatment of systems and events, and assumes that the final result being studied is unique, i.e. dependent or contingent upon everything that came before. We suggest that one of the real difficulties we have in understanding ecosystem properties is our inability to deal with scale, and we show how historical science allows us to approach the issue of scale through the interpretation of pattern in time and space. We then use the techniques of the historical-descriptive approach to doing science in the context of our own and other research on climate change and biological production in the North-east Pacific Ocean. In particular, we examine rapid decadal-scale shifts in the abundance and distribution of two major components–salmon and zooplankton - of the large marine ecosystem of the North-east Pacific, and how they relate to similar shifts in North Pacific atmosphere and ocean climate. We conclude that they are all related, and that climate-driven regime shifts, such as those we have identified in the North-east Pacific, can cause major reorganizations of ecological relationships over vast oceanic regions.