Regarding the conclusions about the activity: here should you consider that other labs that want to reproduce your results have to use different plants (biological replicates). An adequate measure of your uncertainty must therefor be based on the number of bilogical replicates only. A simple way is to average all the data per plant (across all different levels of technical replicates) to get 8 values from which you then calculate the quantity of interest together with some measurement of uncertainty (standard error, confidence interval, p-value, ...).

Regarding conclusions about your experimental approach: here is is of interest at which level how much variability is introduced. This information is relevant to design better future experiments (like: will I get equally good data without triplicate measurent?, or would it be better to increase the number of leaves per plant than to increase the number of measurement replicates per leave to get better data?, etc).To get such information you should use a mixed model for analysis, where the random effect factors are your different levels of technical replication.

I totally agree with Prof

I think that you may need to use three way mixed anova for this purpose