The combined effect of nutrients and temperature
was similarly tested by four-way ANOVA, with treatments
(nutrient and temperature) as fixed effects, tank as a random factor
nested in the nutrient treatment, and coral colony as a random factor
nested in each tank. The tank effect was dropped from the ANOVA analyses
because it was not significant for all the measured parameters. The
factor of coral colony was significant for the algal density and Chl a again
and thus, the combined effect of nutrients and temperature was tested
by two-way ANOVA for each coral colony.