I have a study that explores the effects of stream flow (RC1) on the condition (RW) of a fish species (n=1,900 fish).
Fish were collected from 11 streams (STREAM) in 3 years (YEAR; 2012-2014). Flow variables are calculated each year for each stream. Watershed size (DA.std) was also included as a fixed effect to account for any effects of stream size. I am interested in how condition may differ across years as well, so I have included it as a fixed effect. I included STREAM as a random variable. RW is normally distributed. Initial examination of the dataset indicated that condition increased with RC1 in 2012 and 2013, but had no effect in 2014. See plots of mean condition values vs flow below.
Mean values of RW relative to RC1
I therefore anticipated that RC1 may have a positive significant effect on condition but no effect in 2014.
My current model looks like this: RW2 = lme(RW ~ RC1+DA.std+YEAR, random=~1|STREAM,data=dat)
Output below: Long story short, RC1 had the opposite effect on RW than I would have anticipated. I should note that when I subset out 2014, the relationship is the anticipated direction and is significant. What am I doing wrong? Thanks!
