50 
The evidence is strong (p < 0.0001) that the coefficient for the turbidity term is not 
consistent among tributary systems. Thus some splitting of the tributaries into groups 
should be explored. 
3a. Is chla an important predictor? Yes, but contribution is less than turbidity. 
To address this question, the ANCOVA model was expanded to include terms for 
chlorophyll (as measured by DATAFLOW) and tributary*chlorophyll (Table D-3). 
Table D-3. ANCOVA table showing test for chlorophyll and consistency of chlorophyll 
effect over tributaries. 
Source 
DF 
Type III SS 
Mean Square 
F Value 
Pr > F 
tributary 
16 
19.290 
1.205 
2.39 
0.0016 
rl_5turb 
1 
197.379 
197.379 
391.14 
<.0001 
rl_5turb*tributary 
16 
25.554 
1.597 
3.17 
<.0001 
xCHLA 
1 
14.771 
14.771 
29.27 
<.0001 
xCH LA* tributary 
16 
33.057 
2.066 
4.09 
<.0001 
Both the chlorophyll term (p<0.0001) and the chlorophyll* Tributary term 
(p<0.0001) are statistically significant. However, the mean square for turbidity 
(msIII(turb) = 197.2318735) is much greater than the meansquare for chlorophyll 
(msIII(chla) = 14.7708689). From this we infer that while chlorophyll is an impor¬ 
tant predictor (p<0.0001) it is much less important than turbidity. 
3b. Is chla effect same for all tributaries? No 
The interaction statistic for chlorophyll and tributary is significant (p<0.0001) and 
this implies that the association of chlorophyll and K d is not uniform over tributaries. 
3c. Is it better to use chla or logchla? Chla 
Using the above model, the overall r 2 (chla) = 0.739471 and the overall r : (logchla) = 
0.721274. Thus is appears that the untransformed chla gives better prediction. 
4. Is Salinity a useful predictor? Yes 
Table D-4. ANCOVA table for model expanded to include salinity terms. 
Salinity appears to be an important predictor but its effect is not 
tributaries. 
Table D-4. ANCOVA table for model expanded to include salinity terms. 
consistent over 
Source 
DF 
Type III SS 
Mean Square 
F Value 
Pr > F 
tributary 
16 
16.086 
1.005 
2.03 
0.0093 
rl_5turb 
1 
162.130 
162.130 
327.82 
<.0001 
rl_5turb*tributary 
16 
14.711 
0.919 
1.86 
0.0206 
xCHLA 
1 
9.717 
9.717 
19.65 
<.0001 
xCHLA*tributary 
16 
21.609 
1.350 
2.73 
0.0003 
xInSALINITY 
1 
0.057 
0.057 
0.12 
0.7339 
x!nSALINIT*tributary 
16 
18.498 
1.156 
2.34 
0.0021 
appendix d 
Derivation of K<j Regressions 
