Is it possible to statistically analyse effect of independent variable on one of two interacting dependent variables?

#1
This is the first question I have posted here so I kindly ask for your patience.

My issue is that I'm not sure whether I can apply a statistical test to my experimental data.

My research question

Does the concentration of phosphate in the fertiliser affect the level of fungal colonisation (of plant roots)?

The experimental design:

- 1 independent variable (phosphate treatment) with two levels (2 different concentrations of phosphate fertiliser; 'high' and 'low')
- 1 continuous dependent variable (fungal colonisation [%])
- Fungal colonisation was assessed at several time points, however samples corresponding to each time point were independent of those in the other time points.

Therefore, my statistical analysis consisted of testing for differences between colonisation levels between 'high' and 'low' treatment groups (separately for each time point) using an independent t-test.

Results showed fungal colonisation levels were significantly lower in the 'high' phosphate treatment compared to the 'low' phosphate treatment, at most of the time points (see attached figure).

This is where I have some confusion:

In addition to measuring colonisation levels, I measured another continuous variable (plant biomass [g]) for each sample in the two treatment groups to determine whether the phosphate treatment had an effect on plant growth.

There are a few time points at which the 'high' phosphate-treated group shows higher biomass than the 'low' phosphate-treated group. However, these groups also show a significant difference in fungal colonisation levels, and fungal colonisation itself is thought to have an effect on biomass.

So it seems to me like the comparison of the effect of phosphate concentration on biomass would be confounded by different colonisation levels.

Therefore, given that it is not possible to separate the two, is it valid to use a t- test to compare the direct effect of different phosphate treatments on biomass, given that phosphate may also indirectly affect biomass by affecting colonisation levels?

Happy to elaborate if required. Many thanks for your help. Screen Shot 2019-08-12 at 13.04.19.png
 
#3
Thank you for your reply. No, according to scientific literature, the effect of phosphate is not indirect through stimulating growth, but through another yet unidentified mechanism.

The treatment was not randomised due to technical limitations related to the way phosphate fertiliser can be applied to plants.