12.07.2015 Views

Fraser River sockeye salmon: data synthesis and cumulative impacts

Fraser River sockeye salmon: data synthesis and cumulative impacts

Fraser River sockeye salmon: data synthesis and cumulative impacts

SHOW MORE
SHOW LESS
  • No tags were found...

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

This project is unusual in its scope. While the response variable, ln(R/S), is available for 19stocks across B.C. <strong>and</strong> approximately 50 years of <strong>data</strong> are available for each stock, the number ofpotential covariates is very large. A total of 126 quantitative <strong>and</strong> 5 qualitative <strong>data</strong> sets wereprovided to us from the other technical reports (Table A3.4-9). We then calculated an additional32 <strong>data</strong> sets (i.e., derived variables based on the <strong>data</strong> provided) that were more appropriate forour analyses.It is possible for a single <strong>data</strong> set to be linked to (i.e., hypothesized to impact) multiple life stagesof <strong>Fraser</strong> <strong>River</strong> <strong>sockeye</strong>. In addition, there are up to 4 different age types (i.e., 4sub2, 5sub2,4sub1, <strong>and</strong> 3sub1). These links result in a total of 1058 possible covariates to include in theanalysis. However, not all covariates are available for all years <strong>and</strong> stocks. Models can only becompared when the models are fit using the same <strong>data</strong>. The implication of this is that we cannotcompare all models of interest on the full <strong>data</strong> set but instead must identify time periods withcomplete <strong>data</strong> for different subsets of the covariates.For example, there is a small subset of the covariates (e.g., sea surface salinity for the Strait ofGeorgia) that have <strong>data</strong> extending back to 1950, but there are other covariates that only have <strong>data</strong>starting in 1996 (e.g., chlorophyll a). If we wish to compare models with these two covariates(i.e. salinity <strong>and</strong> chlorophyll), we would have to either reduce the <strong>data</strong> set to those years with<strong>data</strong> for both covariates (i.e., limit the model to 1996-present <strong>and</strong> sacrifice the earlier <strong>data</strong> forsalinity), or exclude covariates with limited years of <strong>data</strong> (i.e., limit the model to only salinity<strong>and</strong> ignore chlorophyll a, but extend the analysis back to 1950). Choosing any particular set ofcovariates forces you to truncate longer time series to the length of the shortest <strong>data</strong> set.Choosing any particular time period forces you to limit your analyses to those covariates with aperiod of record that is sufficiently long.We chose to evaluate different time-periods independently because each time period presents adifferent trade-off between the length of the <strong>data</strong> <strong>and</strong> the number of covariates that can beincluded. Within each time-period we generated different model sets. A model set represents aset of covariates that have complete <strong>data</strong> over the chosen time-period. Within each model set,different models (i.e. combinations of variables) can be tested to determine their ability toexplain the observed variability in the dependent variable, <strong>sockeye</strong> productivity in this case.Expressed another way, a ‘model set’ is simply a suite of c<strong>and</strong>idate models within a given timeperiodthat are organized to address a particular question. For example, one question of interest iswhether the set of factors affecting a particular life stage are more important than others.26

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!