Retrospective Studies

A key problem is to establish the pattern of natural variations, both in physical forcing and in ecosystem response, over as long a time period as possible. That can only be done by looking backward over the period when suitable physical and biological data are available. The approach is known as retrospective analysis and is a necessary precursor for hypothesis formulation, modeling, process studies, and design of monitoring systems. Such studies were instrumental in identifying the so-called regime shift in the late 1970s, both in the nature of physical changes and responses in primary and secondary productivity and in recruitment and abundance of certain fish stocks.

Comprehensive retrospective analyses require reasonably long time series (e.g., up to 100 years) of data on key ecological variables. Unfortunately, such data on key variables rarely exist, so surrogate data are often used. For example, sea surface temperature data are common and have been used as proxies for scarce data on mixed layer temperature. Similarly, sea level data have been used as proxies for direct measures of circulation. Biological data are typically rarer than physical observations--often the longest time series of biological data result from total catch by commercial fisheries. Such fishery data provide only crude estimates of abundance, but that may be all that is available. Data from archaeological and paleoceanographic studies (such as isotope ratios and fish scale deposits) typically cover longer time horizons and represent a potentially useful alternative for reconstructing environmental conditions and biological production.

Long-term historical time series of biological data exist for two of the key fish species identified in this program: walleye pollock and sockeye salmon. Biological samples for the Bristol Bay sockeye salmon represent one of the longest time series collected for any population of Pacific salmon. The migratory pathways and summer feeding grounds of Bristol Bay sockeye salmon are well documented. The walleye pollock stock has been studied intensively for the last 20 years and fisheries data for some key indices (i.e., recruitment, and abundance) extend back 30 years.

Common approaches in retrospective analysis include various forms of correlation analysis and pattern matching. In all cases, methods of establishing relationships between physical and ecosystem variations must be examined critically before cause and effect are inferred. Developing critical reviews and providing guidance on methods of retrospective analyses are an ongoing activity of several PICES scientific working groups.

Two reviews of the Bering Sea ecosystem were recently conducted by PICES and the Natural Research Council. These reviews summarize the current state of knowledge with regards to the Eastern Bering Sea ecosystem and recommend analyses of past abundance levels of key species and their spatial distribution. In particular, changes in spatial distribution and species compositions require review so that some idea of the variability is documented. Summaries on the historical feeding habits of the key species will also be important for studies of species interactions and trophic interactions.


homepage contents previous section next section