I have a set of t-rflp fragment-length profiles from an investigation into microbial degradation of hydrocarbons as part of my undergrad dissertation. I want to study the effect of various variables on the microbial community of uncontaminated soils - specifically if the presence of surfactants and soil from oil-contaminated areas influences the microbial community under both aerobic and anaerobic conditions. Essentially the investigation has four t-rflp replicates for each of the following:


(No surfactant), (No Surfactant + Augmented), (Surfactant), (Surfactant + Augmented)


(No surfactant), (No Surfactant + Augmented), (Surfactant), (Surfactant + Augmented)

Problem is I've never done anything remotely similar to this before, and my supervisor is away for a few weeks. Is it appropriate to use dichotomous independent variables (+/- surfactant, +/- soil augmentation, Aerobic/Anaerobic conditions) for nMDS? I'm using the Paleontological statistics package (PAST) if it helps.

On a related note, is it appropriate to run statistical analysis on DCA axis coordinates of the samples (GLM ANOVA?)

  • $\begingroup$ What is "nMDS", & what are "DCA axis coordinates"? In general, there is no problem with using dichotomous IVs in statistics. It sounds like what you want is a simple ANOVA. $\endgroup$ – gung - Reinstate Monica Nov 2 '12 at 22:53
  • $\begingroup$ Ah sorry, should have said about those: nMDS > Non-metric multidimensional scaling, DCA > detrended correspondence analysis $\endgroup$ – OIOIOIO Nov 2 '12 at 23:03
  • $\begingroup$ Thanks, that helps. Why do you want to do MDS? Usually, you have a matrix of distance measurements, & you determine if you can adequately represent them in a lower dimensional space. (I'm less familiar w/ non-metric, but ultimately it should be the same game.) What would be your distance? I was assuming you had data on some property of the t-rflp replicates & wanted to see if that property differed over the conditions (ie, ANOVA). $\endgroup$ – gung - Reinstate Monica Nov 2 '12 at 23:11
  • $\begingroup$ Each t-rflp replicate has nucleotide fragments ranging from 62 to 573 base pairs in length, each fragment length corresponding to a certain microbial taxon. The fragments vary by intensity (from 0.0 to 1.0 (total absence to total sample domination)), denoting the relative abundance of various microbes in each sample. The intensity is normalized across the samples so all fragment intensities in each sample add up to 1. I'm looking to visualize the independent variable similar to userwww.sfsu.edu/efc/classes/biol710/ordination/… $\endgroup$ – OIOIOIO Nov 2 '12 at 23:33

Your Answer

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

Browse other questions tagged or ask your own question.