we're trying to get some input about a simple model we've constructed to predict fish abundance....groups of a particular species of fish were collected during certain water quality conditions throughout multiple days during a study year.....is the model below an abuse of the proc mixed technique? we "blocked" for date and location with the random statement b/c we feared that fish were disproportional in distribution but felt that the water quality parameters were a legitimate way to predict which environmental components structed their abundance…..suggestions

proc mixed data = new; title "earlu juv sunfishes both years "; class real_date location julian_day; model fish = do temp_c daylight riverheight ph flow conductivity / solution htype=1 htype=3 outpred=X1; random real_date location / solution; run;

proc univariate data = X1 normal plot; var resid; run;quit;