Statistical significance of the linear regressions was determined by analysis of covariance (ANCOVA) using the Statistics toolbox in MATLAB (v 7.9 [R2009b], The MathWorks, Natick, Mass., U.S.A.). It was shown that a statistically significant difference ex isted between all of the regression slopes (P 0.01). At 95% statistical significance, there were no significant differences be- tween any of the 8 intercepts. However, differences were observed between the slopes as reported in Figure 6.