In this study, molecular and morphological variability of guava plants (Psidium guajava L.) ‘Media China’ derived from seeds and subjected to low availability of Fe and Zn was evaluated by RAPD’s markers (Random Amplified Polymorphisms in the DNA). Associations among seventeen morphological characters and 49 RAPD’s loci and response of the plants to the nutrimental stress were established by cluster analysis using the NTSYS, DARwin and c2. The analysis of RAPD’s patterns obtained from a sample of 82 plants showed the existence of genetic variability. From a total of 49 loci evaluated, 96 % were polymorphous, with an average diversity rate of 0.744. The comparative analysis of the efficiency of RAPD’s markers derived from 82 plants versus morphological markers obtained from 72 plants to define and differentiate groups of genotypes with and without symptoms of deficiency of Fe and Zn, showed that the RAPD’s markers are more efficient for that purpose. The groups of plants without symptoms were associated with the presence of RAPD’s loci A20-4, A20-2 and A20-1, with a molecular weight of 0.85, 1.4 and 1.6 kb and a frequency of 86.5, 78 and 89 %, respectively, while the common characteristic for the plants with symptoms was the presence of a band A03-3, of 1.3 kb and a frequency of 74 %. Due to the association of more than one locus with the characteristic absence of symptoms, it can be inferred that this characteristic is polygenic and the combination of the three mentioned loci is necessary to define the groups without symptoms of Fe and Zn deficiency. When calculating the correlation, a non-significant association, r = 0.05743, was found. When the analysis was performed integrating the two types of markers in one single matrix and using the coefficient c2, no separation between the groups with and without symptoms of deficiency was found, and the average distance for RAPD’s was 8.09, with a range of 3.8, while for morphological traits the average distance was 0.15 and the range was 0.4. This confirms that efficiency of RAPD’s markers to detect genetic variability is greater than morphological markers. It also suggests that changes that occur at genome level are not always related to changes in the plant phenotype, and it is possible to establish that even when the correlation detected between the two types of markers is low, simultaneous evaluation is advisable because they are complementary: RAPD’s markers are good describers of the variability at genome level, while morphological variability permits a better explanation of the selection process of the material and has the advantage of better accessibility.

....