The ISSR primers allowed the detection of genetic polymorphism in perennial cotton. High levels of polymorphism (45.8%) represented the genetic variation in genomic ISSR regions of the combined dataset (28 genotypes of G. hirsutum L. r. marie-galante and G. barbadense L.). Therefore, this technique proved itself to be effective in differentiating species using little resources and with the possibility of high reproducibility.
The efficiency of individual ISSR primers was variable. Ordinarily, the efficiency of a certain ISSR primer is correlated with the level of polymorphism that could be generated among the genotypes [21]. However, our results (mean values 0.304 for PIC and 45.8% for polymorphism) were similar to those reported in several studies that assessed the genetic diversity of Gossypium spp. using ISSR markers [22–28]. Thus, the electrophoretic profiles and ISSR marker efficiency in this study are within the expected range for Gossypium L.
All primers showed polymorphism, suggesting the efficacy of these ISSR markers for the assessment of genetic variation between the Gossypium L. species. In this way, all of the primers can be effectively used in the primary evaluation of perennial cotton germplasm. However, UBC-808, UBC-812, UBC-866, UBC-828, UBC-814, and UBC-823 primers revealed specific bands in G. hirsutum L. r. marie-galante (eight bands) and G. barbadense L. (ten bands). This finding can be useful for botanical identification, enabling authentication and qualitative identification of errors of classification in cotton germplasm collections.
The genetic polymorphism revealed by the ISSR marker system was effective in discriminating two perennial cotton species, as well as revealing intraspecific genetic relationships. Genetic analysis of population samples of G. hirsutum L. r. marie-galante and G. barbadense L. provided evidence of three levels of genetic differentiation and two levels of reproductive isolation between species.
The high interspecific genetic differentiation (GST = 0.598) was consistent with the other differentiation index, including genetic structuring in two groups (K = 2) and low genetic identity (0.657) between species. The absence of mixing (Q > 0.98) and the low gene flow (Nm = 0.338) suggest the existence of reproductive barriers between G. hirsutum L. r. marie-galante and G. barbadense L.
In this context, despite these two species being sexually compatible [29] the low gene flow can be attributed to geographic isolation. As a matter of fact, natural hybrids between G. hirsutum L. and G. barbadense L. do not occur in situ, or occur with low frequency due to the rare places where they appear in sympatry [30]. Another hypothesis can be attributed to the habitats where G. hirsutum L r. marie-galante occurs in the semi-arid region of Northeast Brazil which, generally, does not have adequate rainfall for the establishment of G. barbadense L. [31].
Intraspecific genetic similarity was high in both species but the genotypes of G. barbadense L. (0.938) showed greater genetic identity than the genotypes of G. hirsutum L. r. maria galante (0.899). These results suggest that the genetic diversity in these groups is low or moderate, in agreement with the results reported by Hinze [32–33] who also reported low genetic diversity in reference groups of these species. The high similarity between the semi-domesticated genotypes of the two species indicates that there is a potential risk of decline in genetic variation, requiring the inclusion of effective conservation measures. Nonetheless, the high divergence between semi-domesticated and improved genotypes suggests that hybridization between these types can result in more heterotic combinations.
The high divergence between semi-domesticated and improved genotypes shows the usefulness of including them in breeding programs to increase the variability within germplasm collections. The smallest similarity observed between the semi-domesticated genotypes and the improved Pima S7 genotype of the G. barbadense L. group was also reported by [6] and attributed to the wide differentiation between the gene pools of origin. The improved genotypes CNPA 5M and BRS 200 of the group G. hirsutum L. r marie-galante are different from the semi-domesticated genotypes of this group. This divergence stems from the selection of precocity and fibre colour characteristics which were applied during the creation of the improved genotypes CNPA 5M and BRS 200, respectively [34].
The divergence between semi-domesticated genotypes collected in the same province suggests that semi-domesticated populations of G. hirsutum L. r. marie-galante and populations of G. barbadense L. were originally formed by germplasm from different sources. However, the introduction of new sources of germplasm mediated by humans in these populations also seems plausible, as the exchange of seeds is a common practice among horticulturists and farmers in different Brazilian regions.