February 7, 2002 Vol . 23, No. 3
By Candice Martins
High school students who question the relevance of math lessons should take note of Laura Cowens wildlife statistics research.
A PhD candidate in statistics and actuarial science, Cowen (above) is developing statistical models for salmon conservation programs, helping fisheries officers and conservationists obtain accurate information on endangered stocks.
Among Cowens current projects is a model that assesses the survival rate of juvenile salmon (smolt) migrating along the Columbia River. The model analyses data from a sample of smolts monitored with radio tags and estimates how many fish in total successfully pass through one of the rivers hydro dams and reach the ocean.
Radio tags are ideal for monitoring smolt, says Cowen. They are more accurate and smaller than conventional tags and fewer fish need to be tagged, which is important when working with an endangered population.
However, accurately reflecting the radio tags battery life in a statistical model is difficult. If there is no radio signal, you dont know if the fish died, or if the battery died, or if you just didnt pick up the signal, Cowen explains.
As a consequence, the model could underestimate smolt survival rates if a high number of fish complete their migration with dead radios.
This scenario is likely when the Columbias water levels are low and the smolt take longer to travel downstream. As migration time increases, so does the models inaccuracy.
Cowen is reducing the models inaccuracy by making adjustments that will account for the radios battery life, regardless of migration times or water levels.
The Surrey native, who is one of only a few wildlife research statisticians in Canada, says the idea for her PhD work originated from her undergraduate biology studies at SFU. I had all this data from field studies and no idea how to analyse it, she laughs. I decided to learn.
© Simon Fraser University, Media and Public Relations