Linear Correlation/Regression Analysis of Forestry Models 



Discussion 



We found statistically significant relation- 

 ships among several paired variables. Sequoia 

 CRA was positively correlated with increased 

 McNeil core measurements (Figure J-2) and 

 substrate score (Figure J-3) at the 28 fisheries 

 study watersheds and with total nitrogen (Fig- 

 ure J-4) and algae growth (Figure J-5) at ten 

 water quality study sites. There was also a 

 statistically significant regression between 

 HjOY water yield increase and these same four 

 dependent variables. (See Figures J-6 to J-9.) 

 Because of data availability, we were only able 

 to match 6 study sites with the WATSED water 

 yield increase model. WATSED WYI and 

 WATSED 75% peak flow duration were sig- 

 nificantly correlated with total nitrogen con- 

 centration (Figures J-10 and J-11) and maxi- 

 mum algae densities (Figures J-12 and J-13). 

 No significant relationship was observed be- 

 tween die WATSED WYI or WATSED 75% 

 peak flow duration values and any sediment 

 measurements. We also found that the percent 

 of harvest within a watershed was positively 

 correlated with total nutrient concentrations 

 (total phosphorus and total nitrogen) and with 

 algae growth. (See Figures J-14 to J-16). 



The Montana WaterQuality Bureau's stream 

 reach assessment was done on 26 of the study 

 watersheds. We found no statistically signifi- 

 cant correlations between the stream reach in- 

 dex and any other independent or. dependent 

 variables. 



Because Channel Stability Rating and fish 

 habitat measurements have common parame- 

 ters, it was hypothesized that there would be a 

 correlation between the variables. However, 

 there were no statistically significant correla- 

 tions between the CSR and any of the three fish 

 habitat measurements. 



CSR values for those study sites that had 

 historic CSR values were compared to the new 



CSR values to measure the change over time. 

 This change in CSR was hypothesized to be a 

 result of water yield increase. We did not find 

 any statistically significant relationships be- 

 tween the change in CSR and any of the quan- 

 titatively measured dependent variables. How- 

 ever, there were statistically significant rela- 

 tionships between the change in CSR and the 

 water yield models and risk assessment (H^OY 

 water yield increase. Sequoia CRA, WATSED 

 water yield increase, and WATSED water yield 

 increase — 75 percent peak flow duration). (See 

 Figures J-17 to J-20.) 



We observed a highly significant relation- 

 ship between WATSED predicted sediment 

 and the measured sediment yields from the 

 Flathead National Forest water quality moni- 

 toring. In order to do the correlation of sampled 

 suspended sediment to WATSED predicted 

 sediment, the predicted natural background 

 values and the management-caused sediment 

 increases were added together. The correlation 

 between the measured suspended sediment val- 

 ues and the predicted values was highly signifi- 

 cant with an r' = 0.919 and p < 0.(X)1 — an 

 extremely high level of significance. (See Fig- 

 ure J-21.) 



We have the following observations of the 

 relationships that were tested on the sample 

 watersheds. First, we observed a statistically 

 significant relationship between increased tim- 

 ber harvest (as calculated by Sequoia and H^OY) 

 and increased substrate sediment within known 

 bull trout spawning areas. Although only a fifth 

 of the variation in the substrate sediment values 

 and substrate scores was explained by timber 

 harvest disturbance (Sequoia and H^OY), the 

 regressions were positively correlated with sig- 

 nificance levels better than 95%. However, to 

 use the watershed disturbance models to predict 

 timber harvest impacts to fish habitat would be 

 undesirable because of the high level of vari- 

 ance. Therefore, we recommend that more re- 



Flathead Basin Cooperative Program Final Report 



Page 133 



