It is always good to have replicates, both biological and technical to increase the confidence of your results.

Prabodh Kmar, Thnk you for the answer. In case of normal plants (they are clones and don't express considerable genetic variations) both biological and technical replicates are ok. But in case of weedy plants the Technical replicates are ok but biological replicates are still confusing as every individual is genetically different from other within same population and behave differently to different environs.

Shahid, therefore biological replicates will normalize such variations and give you probably a better average or conclusions.

Shahid, in any outbred population, plants or otherwise, biological replicates are essential, not optional.  The point is that if you wish to say something about the population as a whole, then you must adequately sample the range of inherent biological variability amongst the individuals in that population.

Typically, your single largest source of variation in a population is the variability amongst individuals within that population.  If you fail to properly sample the population, then you will never be able to distinguish true population trends from the individual variability.

If you did not include biological replicates, then your study would only apply to the single individual you actually sampled, and it might in fact be very different from the population level affect you were originally interested in.

You can think of it as a single to noise problem  The noise arises from the high levels of biological variation amongst the individuals who comprise the populaiton.  If you now want to identify some affect or trend across the population as a whole, then you first need to be able to quantify or characterize the individual variation, so you can now test for population variability against that background of inherent variation.

Furthermore, with RNA-seq, I would say that technical replicates are really not necessary.  There was a lot of interest in technical variability early on with NGS techniques, but it seems clear that technical variability is typically extremely low, and especially so when considered relative to biological variability amongst individuals in a population.