Description of cashew cultivars and sampling

Twelve cashew cultivars were evaluated. Among them, ten were early dwarf (CCP 06, CCP 09, CCP 76, CCP 1001, EMBRAPA 50, EMBRAPA 51, BRS 189, BRS 226, BRS 253, BRS 265), one hybrid between an early dwarf and a common clone (BRS 275), and one common clone (BRS 274) (Table 1). All cultivars were from the Cashew Breeding Program of Embrapa Agroindústria Tropical, Fortaleza, Ceará state, Brazil. Figure 2 shows the geographic origin of cultivars.

Figure 2: Geographic origin of cashew tree for this study in Brazil. Pacajus (-4.17398, -38.4638) at the Ceará and Pio IX (-6.84773, -40.5836) at the Piauí. 

Leaf and kernel samples for each cultivar were collected from trees located on the experimental field of the Embrapa Agroindústria Tropical Cashew Germplasm Bank (BAG Cashew), in Pacajus, Ceará, Brazil. All trees in the BAG Cashew are identified, documented, characterized, and subsamples of all accessions are deposited in the Germplasm Bank of Embrapa Recursos Genéticos and Biotecnologia, Brasilia, an accredited institution for reference samples.

Ten cashew nuts of each cultivar were manually harvested and detached from the peduncle and sent to the Laboratory of Biochemistry, at the Universidade Federal Rural do Semi-Árido, Mossoró, Rio Grande do Norte, Brazil. Plastic bags with kernels were kept at room temperature until the start of the experiments. The young leaves of the same cultivars were collected, stored in plastic bags and kept on ice until reaching the Laboratory of Plant Biotechnology, at the Universidade Federal Rural do Semi-Árido, Mossoró, Rio Grande do Norte, Brazil. Samples were then stored at -20 °C temperature and DNA isolation was performed next.

Physicochemical traits evaluation

The physicochemical traits were evaluated from previously steamed cashew kernels. Upon steaming, cashew kernels were dried at room temperature and then crushed in a mortar. The following parameters were analyzed, following methods described by the Adolfo Lutz Institute (^{IAL, 2008}): pH, titratable acidity, moisture, ash, lipids, and proteins (Table 2).

The BRS 253 cultivar was discarded from this part of the study because, at the beginning of the experiment, cashew kernels from this clone had high moisture due to the peel removal. During the execution of the experiments, it was hard to obtain more samples from this cultivar.

For pH analysis, a 3.0 g fresh nut was placed in 30 ml of distilled water, and the pH was measured directly with a potentiometer (Fisherbrand™, AE150, Massachusetts, USA). This same solution was used to determine acidity. For this purpose, three drops of phenolphthalein were added and then the solution was titrated with 0.1 M sodium hydroxide (NaOH). In the pH analysis, the concentration of free hydrogen ions (H₃O⁺) in solution is determined. Acidity is a measure of the organic acids present in the solution, which can be either weak acids (they do not completely dissociate in water) or strong acids (they completely dissociate in water).

Many acids cannot completely dissociate in water. These acids have a low dissociation constant (Ka), generating a low concentration of H₃O⁺ in solution. This can cause differences in the values of acidity and pH, as the acidity can be high, but the composition of acids can be mostly weak acids, generating little H₃O⁺ in solution and low pH value.

To determine moisture, 5.0 g of fresh nut weighed in a porcelain crucible were dried in an oven (at 105 °C for 4 h) and then in a desiccator. For ashes, 2.0 g of dried nut from moisture analysis weighed in a crucible were placed in a muffle at 550 °C and kept for 6 h, and then the sample was weighed, after being cooled in a desiccator.

Lipids were extracted using hexane in a Soxhlet extractor (HMK21043, Beijing, China). 1.0 g sample weighed (dried nut) in a cartridge was dissolved with 125 ml hexane. Then, the residue was placed in an oven at 105 °C and in a desiccator.

Crude protein was determined according to the Kjeldahl Method (^{Kirk, 1950}). 0.5 g of sample (dried nut) was digested with 200 mg of catalytic mixture (potassium sulfate and copper sulfate at 9:1 ratio) and 5 ml of sulfuric acid. Then, the digested solution was distilled using 20 ml of NaOH (40%), 10 ml of boric acid (4%), and a mixed indicator (bromocresol green and methyl red) to receive the ammonia released in the chemical reaction. Subsequently, the distillate obtained was titrated with 0.1 N hydrochloric acid (HCl).

The physicochemical characteristics of cashew kernels (pH, titratable acidity, moisture, ash, lipids, and proteins) were evaluated in triplicate in a completely randomized experimental design. Data were submitted to one-way analysis of variance (one factor = cashew cultivars), and the significant differences between the means of each genotype were evaluated by the Tukey test at 5% probability level in Past software v. 4.03 (^{Hammer et al., 2001}).

To assess the physicochemical diversity among cashew cultivars, we used an UPGMA cluster analysis, as implemented in the software Genes v. 2013 (^{Cruz, 2013}). A genetic dissimilarity matrix was constructed using the Euclidean coefficient and a cluster analysis by ANOVA was performed based on the mean of the distances. To validate this clustering, the cophenetic correlation coefficient (CCC) was calculated using the dissimilarity between the cultivars as a measure of genetic distance.

DNA analysis

DNA was extracted from fresh leaves of each cultivar using the CTAB method as described by ^{Ferreira and Grattapaglia (1995)}. Then, the DNA quality was checked using electrophoresis under 80 V for 30 min in 1% agarose gel stained with ethidium bromide (10 mg ml^-1). Subsequently, the gel was photographed under UV light on an image system (Bio-Rad, Gel Doc 2000, San Diego, USA) and the samples were compared with DNA quantification standard from DNA lambda for further dilution to 10 ng μl^-1.

The DNA samples were amplified using eight ISSR primers (Table 3). The amplification reactions comprised a 12 μl final volume, containing 1.2 μl of 10X reaction buffer (20 mM Tris-HCl pH 8.0, 100 mM KCl, and 3.2 mM MgCl₂), 1.0 μl of dNTP (0.25 mM), 2.0 μl of DNA (10 ng μl^-1), 2.0 μl of each primer (0.2 μM), 0.2 μl of Taq DNA Polymerase (5 U μl^-1), and ultrapure water (5.6 μl). The amplification reactions were carried out in a thermocycler (Bio-Rad, Gel Doc 2000, San Diego, USA), programmed for an initial denaturation (at 94 °C for 4 min), followed by 35 cycles of denaturation (at 94 °C for 40 s), annealing (at 40, 45 or 50 °C, depending on the primer), and an extension (at 72 °C for 1 min), followed by a final extension (at 72 °C for 2 min). Successful amplification was checked by agarose electrophoresis (1.5% agarose gel ran for 220 min at 110 V, stained with 10 mg ml^-1 ethidium bromide, and visualized under UV light on an image system (Bio-Rad, Gel Doc 2000, San Diego, USA).

The RAPD markers were obtained from amplifications with 15 arbitrary primers (Table 3) in reactions with a final volume of 12 μl. Reactions were prepared under the same conditions as described for the ISSR markers, except that primer concentration was 0.33 μM. Amplification was performed in a thermocycler (Bio-Rad, PTC-100®, California, USA) programmed for an initial denaturation at 94 °C for 5 min followed by 40 cycles consisting of: denaturation at 94 °C for 30 s, annealing at 40 °C, and extension at 72 °C for 1 min; followed by a final extension at 72 °C for 7 min. The amplification products were subjected to agarose electrophoresis under the same conditions as described for the ISSR markers.

Statistical analyses

Based on the analysis of the band pattern produced by each primer (RAPD and ISSR), a binary matrix was built with both molecular markers, attributing number one to the presence and zero to the absence of polymorphic bands. From the presence/absence matrix, a genetic dissimilarity matrix was obtained using the Euclidean coefficient, which was used as input for cluster analysis using the UPGMA clustering method (^{Faleiro, 2007}), as implemented in the software Statistica v. 8 (^{Weiß, 2007}). To compare the genetic diversity among the cultivars using the RAPD and ISSR markers and physicochemical attributes, a Mantel test approach to calculate Pearson’s correlation coefficient between the dissimilarity matrices was implemented in the software Statistica v. 8 (^{Weiß, 2007}).

Physicochemical traits

Cashew kernels from the twelve cashew cultivars evaluated showed pH values close to neutrality, 6.8 on average. Acidity was 0.81% on average, with values ranging from 0.70% to 0.90%, which corresponded to EMBRAPA 51 and BRS 274, respectively. The standard deviation was low (0.0007), so that no significant difference for acidity was observed among cultivars, according to the F test at 5% probability level. Moisture in cashew kernels was 6.69% on average, with 0.021 standard deviation. The contents of ashes, lipids and proteins in cashew kernels were 2.76, 47.2, and 28.6% on average. Table 2 shows the results of the physical and chemical properties of each variety. Figure 3A and Table 4 show the similarity relationships among accessions based on physicochemical traits. The supplementary material (Appendix) shows the results of ANOVAs for each characteristic that was tested to differentiate the cultivars.

Figure 3: Dendrograms obtained from statistical analyses based on physicochemical (A) and molecular (B) data for the twelve cashew cultivars. A. the dendrogram was constructed using the UPGMA clustering method in the software Genes; B. the dendrogram was obtained in Statistica software. The red line indicates the cutoff point for arrangement of groups. 

Genetic diversity

Analyses with the RAPD and ISSR markers generated a total of 163 DNA bands, from which 159 were polymorphic (Table 3). Thus, the joint analysis provided 97.5% of polymorphism. The average of bands per primer was 8.13 polymorphic bands for RAPD and 5.1 for ISSR.

Using ISSR markers, TriCAT 3'RC was the most informative primer, amplifying seven bands (Fig. 4). Using RAPD markers, the most informative primer was OPA 02, producing 14 bands (Fig. 5).

Figure 4: Amplification pattern of twelve cashew cultivars using the ISSR 04 primer. Numbers 06 to 275 correspond to the studied cultivars. 

Figure 5: Amplification pattern of twelve cashew cultivars using the RAPD OPA 02 primer. Numbers 06 to 275 correspond to the studied cultivars. 

The genetic dissimilarity coefficients among the twelve cashew cultivars ranged from 4.0 to 14.0. BRS 274 and BRS 275 were the cultivars with the highest similarity degree (4.0), while CCP 09 and BRS 189 were the most genetically distant (14.0, Table 5). Moreover, cutoff at 70% genetic similarity in the dendrogram arranged the cultivars into five subgroups (Fig. 3B).

Subgroup I clustered CCP 06, CCP 09, CCP 76, CCP 1001, EMBRAPA 50, and EMBRAPA 51. Subgroup II grouped BRS 226, BRS 274, and BRS 275. Subgroups III, IV, and V were composed only of one clone: BRS 253, BRS 189, and BRS 265, respectively.

Correlation analysis between molecular markers and physicochemical characteristics

The correlation coefficient (r=0.02) between the similarity matrices generated by the RAPD markers and physicochemical attributes was positive but not significant according to the Mantel test at 5% probability (p=0.49). Also, the correlation coefficient (r=-0.04) between the ISSR and physicochemical matrices was negative, and, although significant at 1% probability by the Mantel test, no correlation between these two matrices existed (p=0.01).

Physicochemical traits

The pH we observed (6.8) was similar to that reported by ^{Liu et al. (2019)} in cooked (6.14) and raw (6.20) cashew kernels. Moisture in cashew kernels is relatively high, and moisture content differed significantly among cultivars. However, the BRS 253 cultivar was irrelevant for the study because, at the beginning of the experiment, cashew kernels from this clone had high moisture due to the peel removal.

Peel removal under water vapor conditions may explain the higher initial moisture in kernels. Also, the other factors may affect moisture content evaluation, such as the method used. Moreover, the size of cashew kernels may affect water retention and removal, and, as the size is genetically inherited, moisture differed significantly among cultivars. ^{Akinhanmi (2008)} and ^{Liu et al. (2019)} reported 7.2% and 3.04 moisture, respectively.

Regarding ashes, the content was close to the value of 2.80% found by ^{Akinhanmi (2008)}. Ashes are the inorganic substances present in food, that is, the mineral content. Minerals are essential for diet because they act as cofactors in many physiological processes. Cashew kernels are rich in minerals, mainly selenium, which acts as a powerful antioxidant (^{Souza et al., 2017}).

The lipid content found (47.2%) was similar to those found by ^{Akinhanmi (2008)} and ^{Liu et al. (2019)}, 49.1 and 54.1%, respectively. Cashew nuts are rich in oleic and linoleic acids, fatty acids with therapeutic properties against cardiovascular diseases, besides controlling glycemia (^{Mohan et al., 2017}).

On the other hand, the protein content (28.6%) in the present study was higher than that obtained by ^{Griffin and Dean (2017}) in raw kernels (17.2%), and lower than that found by ^{Akinhanmi (2008)}, 36.3%. Despite the differences in these contents, results support that cashew nuts are an abundant source of protein and amino acids. Rich in these compounds, such cashew kernels are useful for the food industry in providing value-added products (^{Liu et al., 2018}).

Genetic diversity

According to ^{Colombo et al. (1998)}, generating 50 to 200 polymorphic fragments in such analyses is sufficient for estimating genetic relationships within and between species. Similarly, using RAPD and ISSR markers to assess genetic diversity among 40 cashew cultivars from India, ^{Thimmappaiah et al. (2016)} found 163 bands, from which 129 were polymorphic, corresponding to 79.6% polymorphism. ^{Castro et al. (2016)}, using the same markers to study the genetic diversity among 56 grapevine cultivars, observed that 145 fragments were amplified, from which 116 were polymorphic bands, an 80% polymorphism. Thus, such results showed that both markers are effective in assessing genetic diversity.

Subgroup I junction in the dendrogram (Fig. 3B) is explained by the fact that CCP 09, CCP 76, and CCP 1001 were obtained from an individual phenotypic selection from a CCP 06 population. Also, because EMBRAPA 50 was generated from a cross between CCP 09 and CCP 76, the branch in the dendrogram joining the two cultivars is short, showing their genetic similarity. Moreover, EMBRAPA 51 originated within a cross-bred progeny of CCP 76, followed by clonal selection. Therefore, these latter cultivars are close to the group containing CCP 76 (^{Vidal Neto et al., 2013}).

In Subgroup II, cultivars have different origins from the previous ones. BRS 226 was generated from an individual phenotypic selection of MAP-42, a mother plant of early dwarf cashew from Fazenda CAPISA, Piauí (^{Paiva and Barros, 2004}). BRS 274, in turn, originated from an individual phenotypic selection of common cashews in a segregating population (^{Paiva, 2007a}). On the other hand, BRS 275 was obtained by selection within a cross-pollination progeny, originating from cross between an early dwarf (CCP 1001) and a common cashew mother plant (CP 12) (^{Paiva, 2007b}).

The clone BRS 253 in Subgroup III was generated by individual phenotypic selection from an early dwarf mother plant (P96D), which originated from a cross-bred progeny (^{Paiva, 2004}). Thus, the origin explains why this clone was grouped apart.

Similarly, Subgroup IV grouped only the clone BRS 189. Although generated from the selection of progenies from crosses between CCP 1001 and CCP 76, BRS 189 was distant from its parents. Because the ISSR and RAPD markers can randomly access the DNA regions, common sequences between cultivars may not have been accessed in the analysis, which explains the way the cultivars were grouped (^{Vidal Neto et al., 2013}).

Finally, Subgroup V contained only a cultivar originated from an individual phenotypic selection within open-pollinated progenies from the CCP 76 early dwarf cashew. However, both cultivars were not grouped because the markers did not access the common DNA sequences between them (^{Vidal Neto et al., 2013}).

According to ^{Pessoni (2007)}, the highest genetic diversity among cashew plants occurs within populations of the genus Anacardium, while lower variation results from differences among populations. Likewise, ^{Carvalho et al. (2012)} showed that 93.36% of the molecular variation among bushy cashew (Anacardium humile A. St.-Hill.) cultivars occurred within the population, while 6.64% was among those populations.

Analysis with the physicochemical markers showed that the genetic similarity between the cashew cultivars ranged from 2.199 to 16.900. BRS 189 and BRS 253 were the most genetically distant genotypes, while EMBRAPA 51 and BRS 226 were the closest ones (Table 4). In addition, cutoff at 70% genetic similarity arranged cultivars into six subgroups, as shown in the dendrogram (Fig. 3A). Subgroup I included CCP 09, BRS 226, and CCP 76. Subgroups II, III, V, and VII were composed just of BRS 274, EMBRAPA 51, BRS 253, and CCP 1001, respectively. Moreover, Subgroup IV was composed of EMBRAPA 50, CCP 06, BRS 265, BRS 275, and BRS 189. Furthermore, 0.63 cophenetic correlation coefficient (CCC) can be considered significant and of medium magnitude, indicating the consistency of the clustering.

^{Magalhães et al. (2018)} found high genetic diversity among acerola (Malpighia emarginata DC.) cultivars, using physicochemical attributes. Similarly, using morphological characteristics, such as plant growth, flowering, fruit, and nut characters, ^{Jena et al. (2016)} found high variability among twelve cashew cultivars.

Regarding the origin, obtaining method, and respective parents of the twelve cultivars studied here (^{Paiva, 2004}; ^2007a, ^b; ^{Vidal Neto et al., 2013}), genetic diversity among the cultivars was low, considering the molecular analysis used. Also, the molecular data were not similar to the physicochemical ones.

Correlation analysis between molecular markers and physicochemical characteristics

^{Dasmohapatra et al. (2013)}, using ISSR markers and morphological traits to estimate genetic variability among 25 cashew varieties from India, obtained lower and non-significant correlation coefficient (r=0.005).

Environmental conditions influence the expression of physicochemical and other quantitative traits of agronomic interest. Also, such characteristics are subject to natural selection, which acts on genetic level, generating differences among plants of the same genus. Thus, plants with the same DNA sequences can express different DNA regions, which makes the correlation analysis difficult. However, despite the divergence between the generated clusters, some considerations must be made.

The cultivars BRS 189 and BRS 253 are distantly related, as shown in the dendrogram generated based on DNA markers (Fig. 3B). However, they had the highest protein and lipid contents, 34.2 and 54.6%, respectively. Such plants can be used in future breeding programs to improve the contents of these compounds.

CCP 1001 showed the lowest moisture (3.36%) among cultivars, pH close to neutrality (6.92), and low acidity (0.73%) (Table 2). Such attributes are important for the industry because they improve the conservation and taste of the cashew kernels. Also, moisture is difficult to determine because the process is influenced by peel removal and environmental conditions under which plants grow. In addition, because the cashew kernels from the CCP 1001 cultivar have low acidity, they can be eaten toasted, with good acceptance by consumers.

In conclusion, although the results obtained with the ISSR and RAPD markers were not similar, using more than one marker allows obtaining complementing results. It thus improves the strategies of conservation and the use of germplasm. There was no correlation between the molecular and physicochemical data and high variability was found among the cashew cultivars. The cashew kernels from the twelve cultivars studied here had low acidity and pH, considerable moisture and ash contents, and high protein and lipid contents. BRS 189, BRS 253, and CCP 1001 showed the best values for proteins, lipids, acidity, and pH. In addition, those cultivars have maximum distance genetic among them. Therefore, the results suggests that they are more suitable for breeding programmes.