Phenotypic evaluation

A. Recording phenotypic data

A key part of QTL analysis is obtaining accurate estimates of the traits of interest for each line in the mapping population. For most traits, this involves the use of randomized, replicated designs and measurement of multiple plants per line. Evaluation methods and growing conditions must be as uniform as possible across the whole population. Often a QTL population is evaluated in multiple locations or years, to determine whether the same or distinct QTLs influence a trait under different environmental conditions. When evaluating disease or insect resistance, it is essential that each line is exposed to the same level of pathogen or insect pressure. This is often done through artificial inoculation or infestation with the appropriate organism. Another method is to plant in a field with a history of disease or insect problems, and to use management practices that encourage development of the problem.

B. Phenotypic data analysis

The phenotypic data are usually perused and analyzed in a number of ways before they are combined with molecular marker data for QTL detection. These steps often include the following:

i. Frequency distribution

Graphing the number of individuals or lines of a population that fall into different phenotypic classes is a useful place to start in examining phenotypic data. Some points to consider:

  • Is the frequency distribution of the population normal or approximately so? Fig. 4 shows an example of a trait having an approximately normal distribution. Normality is one of the assumptions for analysis of variance, a statistical procedure often used for QTL analysis, although the procedure is considered robust to deviations from normality. If the trait is not normally distributed, transformation to obtain a more normal distribution might be considered; however, bear in mind that transformation may modify the data in such a way that genetic relationships may be obscured.
  • Is there evidence of transgressive segregation, i.e., are there members of the population having trait values higher or lower than the two parents, as shown in Fig. 4? If so, this indicates that alleles for the trait reside in both parents.

Fig. 4: Frequency distribution of biomass per plant in a wheat recombinant inbred line population.

ii. Analysis of variance (ANOVA) of the phenotypic data

  • Are there statistically significant differences among genotypes (lines or plants) for the traits of interest? If not, then QTL analysis will be of limited value, because of the small differences among the lines.
  • Is genotype x environment interaction significant, i.e., does the ranking of lines change markedly from one environment to another? If so, then QTL analyses should be performed separately for each environment, at least initially.

An explanation of the ANOVA technique is available on the web site 'Concepts and Applications of Inferential Statistics', Chapters 13, 14, and 15.

iii. Heritability estimates

The higher the heritability estimate for a trait, the greater the proportion of total variability that is due to genetic, rather than environmental, causes. Thus, higher heritability in a QTL study means that a greater percentage of phenotypic variance can be accounted for by the QTLs. (Heritability is described in more detail in Quantitative Traits, a separate lesson of the Library of Crop Technology.)

iv. Correlation analysis

If two traits are highly correlated, it may indicate that the same QTLs influence both traits (a condition known as pleiotropy). On the other hand, this result may indicate that linked QTLs, rather than the same genes, are associated with the traits.

An introduction to correlation analysis is contained on the web site 'Concepts and Applications of Inferential Statistics', Chapter 3.

Mean Values of Recombinant Inbred Lines











Plant height, cm








Days to heading








Tan spot, 1-5 scale








Yield, ton/hectare








v. Mean calculation

The end product of the phenotypic analysis is a set of mean values for each line for each trait, as shown below. If the population was evaluated in multiple locations or years, the data is usually maintained and analyzed separately for each environment.

This concludes the phenotypic evaluation section.