As part of my PhD work, I've conducted an inoculation experiment concerned with marine phytoplankton community productivity (dependent variable, as 'no. of cells') vs nutrient availability. I have two factors:

  1. Nutrient Treatment (4 levels - including control)
  2. Time (sampled at 5 points)

The puzzling aspect to this design is that within the phytoplankton community, there are 4 main taxonomic groups (each with 'no. of cell' data), which I need to account for (i.e. I would like to know which of these groups significantly differed both within and between treatments over the entire sampling regime).

So, I understand a post-hoc test (e.g. TukeyHSD) will be needed to run multiple comparisons test in order to elucidate group differences, but first need to know which test to initially run (i.e. Two-way repeated measures ANOVA, or multiple ANOVA), and in what structure? Or, is it simply not possible to simultaneously analyse for between phytoplankton group differences?


You could run a MANOVA, with nutrient treatment, time and their interaction as independent variables. You could try time as a categorical variable if you really want individual post-hoc comparisons, or as a covariate. It sounds like you are familiar with MANOVA, but in case not (and for other readers who aren't), here's a brief intro: http://userwww.sfsu.edu/efc/classes/biol710/manova/manovanewest.htm

However, it is likely that your dependent variables are correlated, because increasing nutrient could increase all the abundance of all species, or because of ecological interactions between species. How many ponds do you have? I might be tempted to play around with doing something like a PCA or NMDS on the species data and see whether they cluster by treatment and/or time (this would also show you how the abundances of different species correlate).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.