You have mentioned that your plants pertain to single genotype (or say a variety). The variations you have noticed are environmental as the genotype is the same. Ideally, since you are measuring the same genotype you might get low coefficient of variation. (For determining the number of replications Power analysis is often used.) So the remedy would be to check environmental variations (soil, shading, moisture etc.) to determine reason for high coefficient of variation.

Environmental conditions usually are not as uniform as we think. As you do not have genetic variance, all the variance is environmental. Different traits may show different environmental variance. For example, number of heads do not show any environmental variance, but fatness depends on food disponibility, so it may reflect a huge environmental variance.

Well, if I understanding correctly, you are asking about coefficientes of variation, ok?

It depends on the differences from the measures among the reps that you made in your field trial.

Some traits there are no variation among the reps, example plant high, and others there are a big difference as production.

Altough you are using a single genotype, the environment cause those differences among the reps, one with small effect and others with a big effect causing the differences in the coefficient of variation observed.

If you are a breeder you are looking for traits with a high heritability - low coefficients of variation. Typically yield, being the integral of very many processes, each controlled by many genes, will have a much higher coefficient of variation than a process that is much more simply controlled, physiologically and genetically, such as major dwarfing gene effects on plant height in cereals and single-gene disease resistance.

Because of these complex interactions, it is usually better to work with component traits than with the end product trait; for example grain size rather than yield per plant, as the processes that control those component traits will be more simply controlled/inherited.

As others have stated, your large observed variation is environmental, in its broadest sense - aerial, edaphic (soil) or carry-over from the previous generation, such as seed size and vitality. The more factors you have influencing a particular trait or characteristic, the more opportunity you have for variation. Therefore, as a researcher, your challenge is to keep as many of these factors as possible constant. This is why physiologists like to use controlled environment cabinets, in the (mistaken) belief that they can keep most factors affecting plant growth constant.

So, the more success you have in identifying the (environmental) factors that will influence your trait of interest, the greater will be your opportunity for controlling (or at least accounting for) that potential source of variation. It is the researcher's skill in accounting for these factors that determines the quality of the research - the smaller the coefficient of variation, the easier it is for you to be able to show treatment or genotype differences.

If you are studying the effect of a particular treatment, then maybe selecting plants for uniformity at the start of the treatment will help, ensuring that similar plants have been assigned to both the control and treatment. If availability of plants to assign to treatments is limited, then rank plants (visually for size, say), and assign plants of equal rank between control and treatment.

Without knowing a lot more about the species, the traits, the environments and objectives of the research it is not possible to give any more specific advice.

In my opinion though you have studied intra genotype in uniform environment, but it seems either you have not handled your materials carefully or you have got some mixture seeds.

May be the intra genotype are RILs or NILs which are usually different by a single gene loci

You also may be observing the expression of GxE. Your E may be changed among individuas in the genotype, even in a subtle way and so GxE may be expreesed in the phenotypes of some traits.

I agree with Dr Steve Q to use controlled environmental chambers to minimize variations/ external factors. The need to have uniform planting materials at the beginning is very important to ensure the treatments effect could be identified later if there is any. Within similar genotypes also sometimes we can see variations in certain cases but i would like to suggest to focus on the traits of interest rather than to look at so many traits at one time

Many points have been discussed above related to your question. I had same problem even my population was not so large and under controlled conditions. My CV was also higher. I cought the problem by measuring the exact temperature near the plants, inside the chamber and on the screen of meter controlling automatically. then I sorted out the required temperature. It might be helpful in your case.

Individual parameters of significance need to be weighted according to their significance in the problem at hand. Where ever the data is significantly large, the same may be treated statistically using Poisson distribution. The standard deviation or error 'sigma' represents only 67 % of the most probable range of measurement. If one wishes to care greater variational range, one goes to 3, 5, 7 sigma levels, thereby covering acceptance of individual measurement range to higher and higher percentage levels, say up to 95 or 98 % level of probability distribution!

Narendra Nath

If you are working whith a single genotype of plants, the implications large or small intra-genotypic variations in a trait can be:

1- Traits affected by environment

2- Variations are environmental, not genetic

3- Heritability will be depend of large or small variations

Brilliant answers from most of the colleagues. I did really enjoy going through each one of them. Nevertheless, from a point of view of a statistian for example, even intra-plant variation could be taken into account. A physiologist most probably would have examinded biochemical processes and their possible interactions whereas a plant breeder could have looked at CV or other ways to examine this variation, based on experimental design and of course the corresponding breeding model. When I was conducting an experiment on phytotoxic effects of a photosynthetic inhibitor herbicide on a range of winter wheat cultivars (10 of them) of contrasting growth habits (e.g. dwarf/semi dwarf vs traditional tall cultivars) I 've noticed that my tall cultivars (varieties) were responding much much better two-three days after herbicide application compared to modern and newly introduced dwarf and swmi-dwarf cvs. Controls (untreated-unsprayed plants) were used under same experimental/glasshouse conditions. I decided to investigate more my macroscopically obtained (e.g. plant height, dry weight number of leaves etc) results so I got inside the plant using a chlorophyll fluorescence technique. I found that tall varieties were able to breakdown the chemical more quickly than the dwarf/semi cvs (a pure genetic consequence). Inevitably, I had a look at the parental lines of my tall resistant cvs, in which a parent from three generations back, if I remember well, had shown similar responses (a gene had also identified for this kind of response in D-complex protein of the Z-scheme during electron exchange phase of photosynthesis in PSII ). Amazing case which showed me that environmental factors, as many could say in this occassion, held the least significant role (intar-species variation due to an inherited trait were the major player). I guess I was lucky to examine biochemical/physilogical paths instead of checking effects of the environment (which we know they do play a significant role in herbicide efficacy).

I agree with our colleagues (Steve and others) that your large observed variation is probably environmental in nature because it is very difficult to eliminate microvariation in field experiments. You could reduce it with slightly more complex experimental design. Also in many species maternal effects from previous generation could have large impact on different seedling traits, as Steve said- seed size or vitality. Means, you have to have information about environmental conditions that mothers encountered, at least during seed formation.

Many thanks for all your input. These have been insightful indeed. The large CVs observed in our system were indeed due to environmental effect and most probably experimental vagaries. The experimental design was deliberate as we wanted to quantify the magnitude of the environmental effect. What we did observe was that the environment had varying effects on different traits and this was linked to the extent of plasticity in these traits.