Genetic variability in Brazilian castor (Ricinus communis) germplasm assessed by morphoagronomic traits and gray mold reaction

Abstract

The characterization and conservation of castor accessions in germplasm bank are essential in order to breeding programs achieve its goals. Despite Brazil having the 4^th largest castor germplasm bank in the world, castor diversity in Brazil remains little explored. Thus, this study aimed at characterize castor accessions collected in different Brazilian regions by means of 31 morphoagronomic traits and gray mold reaction. Forty accessions of the Universidade do Estado de São Paulo (UNESP), Botucatu, SP, Brazil, germplasm bank were evaluated. Genetic parameters were estimated for the quantitative traits, and the accessions were grouped by Ward method using the standardized Euclidean distance and the simple coincidence index for quantitative and qualitative data, respectively. Qualitative and quantitative traits were important to understand and differentiate castor accessions. The accessions showed a high variation regarding the castor gray mold reaction. The accessions assessed in this study have been preserved and can be used as a source for genetic variability in the development of new castor varieties in breeding programs.

Key words
Genetic improvement; genetic variability; qualitative descriptors; quantitative descriptors; REML/BLUP

INTRODUCTION

Castor (Ricinus communis L.) is an important oleaginous from Eastern Africa and currently is cultivated in different tropical and subtropical regions around the world (Allan et al. 2008, Severino et al. 2012). Castor seeds have from 40 to 50% of oil content, which is the only commercial source of ricinoleic acid, which is used in high-quality lubricants, cosmetics, pharmaceutical products and polymers (Suhail et al. 2015, Venegas-Calerón et al. 2016). Although India has been responsible for more than 80% of the world production of castor seeds, the crop also plays an important role in Brazil, which produces around 25,000 metric tons per year (FAO 2016).

In Brazil, castor crop is especially important for family smallholders, mainly in the northeast region (Florin et al. 2012). As the national industry of castor has been significantly suffering from the decrease of raw material in the past decades, the Brazilian government launched a program of incentive in order to promote the castor production in other regions (Ribeiro & Raiher 2013). Due to this fact, Brazilian breeding programs start to focus in the development of new dwarf, short cycle and highly productive castor varieties and hybrids with potential to be used in the Brazilian cerrado (Severino et al. 2012).

Studies of genetic diversity characterization are very important for the conservation and utilization of these genetic resources in breeding programs (Saadaoui et al. 2017). Different morphoagronomic traits and molecular markers have been widely used for characterization of the castor germplasm (Wang et al. 2016, Simões et al. 2017, Silva et al. 2017, Rukhsar et al. 2018). However, there are few data about the reactions of these accessions to Botryotinia ricini (Godfrey) Whetzel, which is the causal agent of the castor gray mold, the most import castor disease (Dange et al. 2005, Sussel et al. 2009, Soares 2012).

The castor gray mold can cause yield losses up to 100% when highly susceptible cultivars are growing under favorable conditions to the disease development, even though, there are only few studies about castor gray mold management and, so far, there is no effective ways to control this disease (Sussel et al. 2009, Soares 2012, Sá et al. 2015, Oliveira Datovo & Soares 2019). Attempts to identify sources of resistance to this pathogen have been made since the first reported castor gray mold outbreak, and some works published in the early 20^th century, have claimed, based on field observation, that “spontaneous varieties” or wild genotypes are highly resistant to the pathogen, however, such claims were never corroborated by further studies (Soares 2012).

Although partial resistance to castor gray mold had been reported in some castor cultivars and hybrids (Anjani et al. 2018), only few studies have been conducted aiming to develop resistant genotypes, coupled with the agronomic traits required by the intensive agricultural systems, such as Brazilian cerrado (Severino et al. 2012, Soares 2012). The main reason for this is the lack of reliable genetic resistance source for castor gray mold (Soares 2012, Soares et al. 2010). Thus, the identification of genetic resistance sources is crucial for castor crop expansion through Brazilian cerrado.

The present study had the objective to characterize castor accessions, collected in different regions of Brazil, by several morphoagronomic traits, as well as verify the reaction of these accessions to castor gray mold.

MATERIALS AND METHODS

Plant material

Forty wild accessions of castor which has been maintained in the germplasm bank of the Universidade do Estado de São Paulo (UNESP), Botucatu, Brazil, were evaluated. The accessions were collected on three different States (São Paulo, Minas Gerais and Rio Grande do Norte) of the Southeast and Northeast Brazil, comprising 23 different municipalities, during expeditions carried out in 2015 (Figure 1).

Figure 1
Identification and origin of the collection of the 40 accessions of castor (Ricinus communis L.) characterized by morphoagronomic descriptors and castor gray mold reaction.

Morphological characterization

The research was conducted in greenhouse conditions in the Department of Production and Genetic Improvement of Universidade do Estado de São Paulo (UNESP), School of Agriculture, Botucatu, SP, Brazil (22°50’59.0”S and 48°25’55.6”W and altitude of 786 m) in 2016. The experiments were performed using a completely randomized design with three repetitions. The plots were constituted by one plant per hole, with a spacing of 0.5 m between plants and 1.0 m between rows. Weed control, hydric and mineral supplementation, insecticide and fungicide applications were performed according to the plants need in order to keep satisfactory levels of plant health.

The accessions characterization was done using 32 descriptors (Savy Filho et al. 1999, MAPA 2008, Milani 2008): 23 qualitative and nine quantitative traits. The quantitative traits evaluated were: plant height (cm), primary raceme insertion height (cm), stem diameter (cm), internodes number, commercial raceme number, primary raceme length (cm), 100-seed weight (g), spores number of Botryotinia ricini, and seed oil content (%) using a bench-top NMR spectrometer model SLK-100 (SpinLock, Cordoba, Argentina), determined by nuclear magnetic resonance in time domain using ~15 g of seeds of each treatment were employed. The determination of the oil content was performed using the calibration curve for castor oil.

The qualitative traits were: anthocyanin pigmentation on the hypocotyl, stem waxy, stem color, face format of the limb, leaf vein pigmentation, face waxy limb, upper face color of the limb, stigma color, presence of male flowers on the raceme, predominance of male flowers on the raceme, raceme density, raceme format, capsule waxy, capsule color, presence of spines on the capsule, spines density, spines color, capsule dehiscence, main seed color, secondary seed color, type of secondary color, seed format and caruncle protuberance.

Reaction to castor gray mold

The reaction of castor accessions to castor gray mold was verified using the Soares methodology (Soares et al. 2010), with modifications. Green capsules, between stages V and VII (Greenwood & Bewley 1982), were collected, conducted to the laboratory facilities, washed in running water, and then kept during 30 seconds in alcohol 70% followed by 30 seconds in sodium hypochlorite 0.5%. After that, the capsules were washed with sterilized distilled water and dried in room temperature (25±2°C) for 2 hours. After dried, the capsules were sprayed with a manual atomizer driven by compressed air pump calibrated for a pressure of 1.5 bar with B. ricini spores suspension adjusted for 2×10⁵ spores.ml^-1. After inoculation, the capsules were placed in acrylic boxes, which were sealed with plastic film and maintained in growth chamber at 25±1 °C with a photoperiod of 12 h for 7 days. The boxes had previously been sterilized using sodium hypochlorite 0.5%, left to dry, and then lined with a double layer of sterilized wet filter paper and a polyethylene mesh used to avoid direct contact of the capsules with the wet paper. The experiment was performed using a complete randomized block design with four replicates, each replicate consisting of a box with four capsules. After the incubation period, the four capsules of each replicate were vigorously shaken in 100 mL of alcohol 50% to remove the spores. The obtained suspensions were then filtered in double-layered cheesecloth to remove the mycelial mat. The number of spores per mL, of each suspension, was determined using a Neubauer chamber by means of two independent readings. In order to standardize the spore readings due to the difference in capsule size, the number of spore were divide by the average capsule volume of each accession.

Data analysis

The quantitative traits were analyzed by restricted maximum likelihood (REML) and best linear unbiased prediction (BLUP) methods using the software Selegen-REML/BLUP (Resende 2016). The predicted genotypic values were calculated after verification of normality and homogeneity of data by Shapiro & Wilk (1965) and Hartley (1950) tests, respectively. The data that did not present the assumptions of normality and homogeneity where changed by Box-Cox transformation (Box & Cox 1964). The deviance analysis was performed considering the following statistic model:

$y=\mathit{Xb}+\mathit{Za}+e$

where: y is the data vector, a is the vector of block effects (assumed as fixed) added with the total average, b is the vector of genotypic effects (assumed as aleatory), e is the vector of error (aleatory), X and Z represents the matrices of incidence for b and a, respectively.

The estimators REML for attainment of phenotypic (σ² _p), genotypic (σ² _g) and environmental (σ² _e) variance using the algorithm EM (Expectation-Maximization) were:

${\sigma }_e^2=[y\text{'}y-b^{\text{'}}{X}^{\text{'}}y-g^{\text{'}}{Z}^{\text{'}}y]/[N-r\left(X\right)]$

${\sigma }_g^2=[g^{\text{'}}g+{\sigma }_e^2\mathit{tr}{C}^{22}]/N_g$

${\sigma }_p^2={\sigma }_g^2+{\sigma }_e^2$

where: $N_g$ is the number of aleatory elements (individuals), $\mathit{tr}$ is the dot matrix operator (sum of elements of the diagonal matrix), $N$ is the total number of data, $r\left(X\right)$ is the number of independent linear columns $X$ and, $C^{22}$ is the formula:

$\left[\begin{array}{cc}{C}^{11}& C^{12}\\ C^{21}& C^{22}\end{array}\right]={\left[\begin{array}{cc}X\text{'}X& X\text{'}Z\\ Z\text{'}X& Z^{\text{'}}Z+A^{-1}({\sigma }_e^2/{\sigma }_g^2)\end{array}\right]}^{-1}$

Broad-sense heritability (h²) and the selective accuracy of genotypes (AS_g) were calculated in the following way:

$h^2=\frac{{\sigma }_g^2}{{\sigma }_g^2+{\sigma }_e^2}$

${\mathit{AS}}_g={(h^2)}^{\frac{1}{2}}$

The coefficient of genotypic (CV_g) and environmental (CV_e) variation was determined, respectively, by the following formulas:

${\mathit{CV}}_g\left(\text{\%}\right)=\frac{100\sqrt{{\sigma }_g^2}}{\overline{m}}$

${\mathit{CV}}_e(\text{\%})=\frac{100\sqrt{{\sigma }_e^2}}{\overline{m}}$

Pearson linear correlation, Principal Components Analysis (PCA) and Singh’s relative importance (Singh 1981), analysis were performed using the predicted genotypic values. The estimation of the genetic distance matrix among the accessions for the quantitative traits was performed using the standardized Euclidean distance, while for the qualitative traits it was used the simple coincidence index. The Ward method (Ward 1963), was used in the hierarchical groupings of the accessions for both quantitative and qualitative data. The correlation between the matrices of the quantitative and qualitative data was verified by Mantel test (Mantel 1967), using 2,000 permutations. The optimal number of groups, formed in the dendrograms, was choose using the Milligan and Cooper methods (Milligan Cooper 1885). Statistical analyses were performed using Genes (Cruz 2013) and R software (R Core Team 2019) through packages ‘NbClust’ (Charrad et al. 2014), ‘dendextend’ (Galili 2015), ‘corrplot’ (Wei et al. 2017) and ‘FactoMineR’ (Lê et al. 2008).

RESULTS

Qualitative descriptors

Polymorphism was observed for all qualitative traits evaluated, except for anthocyanin pigmentation on the hypocotyl and male flowers on the raceme (Table I). Stem wax was present on 67.5% of the accessions, while 32.5% did not show wax. Stem color varied from light green (10.0%), green (32.5%), pinky green (42.5%), pinky (10.0%), and red (5.0%). Funneled limb was observed on 72.5% of the accessions, while the other 27.5% were not funneled. The presence of wax on the limb was observed in 97.5% and only 2.5% of the accessions did not showed this trait. Limb color varied from light green (17.5%), green (65.0%) and dark green (17.5%), while the vein colors were greenish (52.5%) and reddish (47.5%). Concerning the inflorescence traits, 10.0% of the accessions showed stigmas with greenish color, 27.5% orange, 37.5% reddish and 25.0% pinky. The male flowers were mainly observed on the lower part of the raceme (95.0%) and only 5.0% of the accessions showed male flowers interspersed with female flowers. The predominant racemes form was conical (60.0%), followed by globose (32.5%) and cylindrical (7.5%), while their densities were predominantly intermediate (42.5%), followed by sparse (37.5%) and compact (20.0%).

Table I
Accessions number per group for qualitative morphoagronomic traits in each of the three groups (G1, G2 and G3) formed by the Milligan Cooper (1985) method from 40 castor (Ricinus communis L.) accessions.

Trait	Groups			Trait	Groups
	G1 (12)	G2 (18)	G3 (10)		G1 (12)	G2 (18)	G3 (10)
Anthocyanin on hypocotyl				Capsule color
Presence	12	18	10	Light green	1	12	1
Absence	-	-	-	Green	3	3	8
Male flowers on the raceme				Dark green	7	2	1
Presence	12	18	10	Red	1	1	-
Absence	-	-	-	Spines on capsules
Stem waxy				Presence	12	17	10
Presence	1	16	10	Absence	-	1	-
Absence	11	2	-	Spines density
Stem color				Low	-	1	1
Light green	-	4	-	Medium	8	10	8
Green	3	5	5	High	4	6	1
Pinky green	5	7	5	Spines color
Pinky	2	2	-	Light green	1	14	1
Reddish	2	-	-	Green	5	2	9
Limb face				Dark green	3	2	-
Plain or not funnel-shaped	4	6	1	Pink	2	-	-
Funneled or funnel-shaped	8	12	9	Red	1	-	-
Leaf vein pigmentation				Capsule dehiscence
Greenish	6	12	3	Dehiscent	10	15	10
Reddish	6	6	7	Semi-dehiscent	1	1	-
Limb waxy				Indehiscent	1	2	-
Presence	11	18	10	Seed main color
Absence	1	-	-	White	2	2	-
Foliar lamina color				Yellow	1	2	1
Light green	2	5	-	Light brown	-	2	-
Green	8	12	6	Brown	1	-	-
Dark green	2	1	4	Dark brown	-	1	-
Stigma color				Grayish	8	11	9
Light green	-	1	3	Black	-	-	-
Orange	1	9	1	Seed secondary color
Reddish	4	7	4	White	-	1	-
Pinky	7	1	2	Yellow	-	-	1
Male flower predominance				Light brown	1	-	2
Lower part predominant	12	16	10	Brown	6	7	6
Interspersed with female flowers	-	2	-	Dark brown	1	10	1
Raceme shape				Reddish brown	3	-	-
Globose	7	4	2	Black	1	-	-
Cylindrical	-	2	1	Secondary color type
Conical	5	12	7	Painted	-	1	-
Raceme density				Striped	12	17	10
Sparse	6	4	5	Seed shape
Intermediate	3	13	1	Rounded	7	8	4
Compact	3	1	4	Ellipsoid	5	10	6
Capsule waxy				Caruncle protuberance
Absence	10	3	-	Inconspicuous	2	1	2
Presence	2	15	10	Conspicuous	10	17	8

Regarding the capsule traits, the presence of wax was observed in 67.5% of the accessions. Color varied from light green (35.0%), to green (35.0%), to dark green (25.0%), and red (5.0%). Capsule dehiscence was observed in 87.5% of the evaluated accessions, while semi-dehiscence and 7.5% indehiscence were observed on 5.0 and 7.5% of the accessions, respectively. The presence of spines on the capsules was observed in 97.5% of the accessions. Spine density varied from low (5.0%) to high (27.5%), being the medium the most observed density (67.5%). Color of spines was light green (40.0%), green (40.0%), dark green (12.5%), pinky (5.0%), and red (2.5%). Concerning the seeds, the accessions showed seeds with rounded (47.5%) or ellipsoid (52.5%) shape. The caruncle protuberance was conspicuous for 87.5% and inconspicuous on 12.5% of the accessions. Seed primary color was white (10.0%), yellow (10.0%), light brown (5.0%), brown (2.5%), dark brown (2.5%), and grayish (70.0%). Striped seeds (97.5%) were most frequently observed, and only 2.5% of the seeds were painted. The secondary colors observed were white (2.5%), yellow (2.5%), light brown (7.5%), brown (47.5%), dark brown (30.0%), reddish brown (7.5%) and black (2.5%).

Quantitative traits

The predicted genotypic values regarding the nine quantitative traits are presented in Figure 2. The plant height varied from 64.2 (CB12) to 248.7 cm (CB14), while the insertion height from the primary racemes and the length of the raceme varied from 52.2 (CB12) to 185.6 cm (CB13) and 19.1 (CB23) to 56.7 cm (CB14), respectively. Stem diameter varied from 1.63 (CB17) and 3.73 cm (CB14). Regarding the number of internodes and racemes, the genotypic means varied from 13.0 (CB37) a 24.0 (CB40) and 1.63 (CB20) to 10.33 (CB3), respectively. The weight of 100 seeds varied from 10.96 to 41.29 g, considering that accessions CB4, CB30 and CB40 showed the heaviest seeds. The seed oil content varied between 39.18 and 50.83%, considering that accessions CB10, CB30 and CB40 showed the higher percentage. The spores number of B. ricini varied from 2.19 to 6.14 spores.mL^-1 among the accessions evaluated, considering that accession CB40 was considered the most tolerant to the castor gray mold and the CB10 the most sensible.

Figure 2
Boxplots of the predicted genotypic values from the nine traits evaluated in 40 accessions from the castor (Ricinus communis) grouped in all accessions (white), group 1 (blue), group 2 (gray), and group 3 (green).

Deviance analysis and genetic parameters

The deviance analysis and the genetic parameters obtained are presented in Table II. The deviances were highly significant (P0.01) for all the traits assessed. The CV_g varied between 6.91 and 58.31% to the seed oil content and number of commercial raceme, respectively. The highest CV_e were observed in the number of commercial racemes (14.38%) and in spores number of B. ricini (12.90%), while the lowest values were verified to the 100-seeds weight (4.99%) and in the seed oil content (4.68%). The h² estimates oscillated between 49.12% and 94.59%. The plant height (94.22%) and the commercial racemes numbers (94.59%) showed the highest h² estimates, while the lowest estimated were observed in the number of spores of B. ricini (49.12%) and stem diameter (71.44%). The estimates of AS_g were elevated to all assessed traits varying between 0.70 and 0.97 to the number of spores of B. ricini and plant height, respectively.

Table II
Deviation analysis, estimate of variance components and genetic parameters of nine quantitative traits in 40 accessions of castor (Ricinus communis L.).

Parameters2 /	Morphoagronomic traits1 /
	NS	PH	IR	SD	NI	NR	LR	WS	OC
Deviance	145.95*	795.56*	754.55*	238.42*	331.71*	168.97*	656.76*	143.05*	466.71*
σ2 g	0.60	1590.95	1020.01	0.21	4.46	6.00	52.14	46.99	9.91
σ2 e	0.43	98.76	137.84	0.09	1.54	0.37	10.51	0.15	0.05
σ2 f	1.03	1689.71	1157.85	0.30	6.00	6.37	63.65	47.14	9.96
h2 (%)	49.12	94.22	88.45	71.44	74.12	94.59	83.01	91.82	89.44
ASg	0.70	0.97	0.89	0.84	0.86	0.97	0.91	0.95	0.94
CVg (%)	16.90	28.19	27.36	19.71	12.74	58.31	19.96	34.92	6.91
CVe (%)	12.90	7.02	10.05	12.62	7.48	14.38	8.96	4.99	4.68
Average	4.58	141.45	116.73	2.34	16.58	4.21	36.16	16.62	45.34

• ¹/NS: number of spores of Botryotinia ricini, PH: plant height, IR: primary raceme insertion, SD: stem diameter, NI: number of internodes, NR: number of commercial racemes, LR: length of primary raceme, WS: 100-seeds weight and OC: seed oil content.
  ²/σ² _g: genotypic variance, σ² _e: environmental variance, σ² _f: phenotypic variance, h²: broad-sense heritability, AS_g: selective accuracy of genotypes, CV_g: coefficient of genotypic variation and CV_e: coefficient of residual variation.
  *: significant by the Chi-square test with a degree of freedom at the 1% probability level.

Correlations and relative importance of traits

The correlation coefficients among the nine quantitative traits are presented in Figure 3. The plant height presented a positive and significant correlation with the variables: insertion of primary raceme (0.82**) and stem diameter (0.77**). A positive and significant correlation was also observed among traits such as stem diameter with insertion in the primary raceme (0.61*), the number of commercial raceme with stem diameter (0.60*) and seed oil content with the 100-seeds weight (0.64*). By means of the analysis of relative importance of traits of Singh (1981) it can be observed that the variables: number of racemes (16.96%) and seed oil content (15.52%) were the traits the most contributed to the differentiation of the accessions evaluated while the number of internodes (3.53%) and 100-seeds weight (7.42%) showed less importance (Figure 4).

Figure 3
Estimation of the genotypic correlation coefficients with their respective 95% confidence intervals in nine traits evaluated in 40 accessions of castor (Ricinus communis).

Figure 4
Relative importance by the Singh (1981) method of nine traits evaluated in 40 accessions of castor (Ricinus communis).

Principal component analysis (PCA)

The first two principal components (PC) explained 77.03% of the total variation among the nine quantitative traits assessed (Figure 5). The first component (PC1) responded by 55.44% from the variation attributed to the traits such as plant height, number of commercial racemes and internodes number. On the other hand, the second component (PC2) absorbed 21.59% of the total variation, and are associated with seed oil content and the 100-seeds weight. In the two-dimensional graphic of PCA, it can be observed the distinction of 40 accessions of castor in three different groups. Generally, the groups 1 (blue) and 2 (gray) presented the highest and lowest averages, respectively, to the traits of primary raceme insertion, plant height, stem diameter and the number of commercial racemes. The groups 3 (green) presented associations with vectors of the variables seed oil content, 100-seeds weight, internode number and length of primary raceme, presenting the highest averages to those traits.

Figure 5
Principal component analysis (PCA) of nine traits assessed in 40 castor accessions (Ricinus communis).

Grouping analysis

The genetic dissimilarity among the accessions obtained by the simple coincidence index had the average value of 0.38 (±0.08). The lower distance (0.12) was observed among accessions CB19 and CB24, being those accessions collected in the towns of Campos do Jordão and Natal (São Paulo and Rio Grande do Norte States, respectively). The CB13 (Botucatu, São Paulo State) and CB37 (Santa Isabel, São Paulo State) accessions were the more distanced (0.67). Using a dendrogram obtained by the Ward method, it can be observed the formation of three distinct groups (Figure 6a). The group I (green) was constituted by 12 accessions, presenting prevalence of accessions with an absence of waxy coating in the stem and capsules. Eighteen accessions constituted the group II (gray), presenting in common accessions with a prevalence of light green capsules, plants with an intermediary density of racemes and stigmas with orange and red colors. The group III (blue) was formed by ten accessions that showed waxy coating in the stems and capsules, spines with green coloration and dehiscent capsules.

Figure 6
Genetic dissimilarity among 40 accessions of castor (Ricinus communis) obtained by the Ward method based on standard Euclidian distance (quantitative) and simple coincidence index (qualitative).

The genetic dissimilarity obtained by the standard Euclidean distance presented the average value of 0.31 (±0.08). The smaller distance was observed among the accessions CB5 and CB6 (0.04), being both accessions from Bofete city (São Paulo State). On the other hand, the accessions CB14 (Botucatu, São Paulo State) and CB37 (Santa Isabel, São Paulo State) were the most distanced (0.69). By the dendrogram obtained by the Ward method, it can be observed the formation of three distinct groups (Figure 6b). The group I (blue) was constituted by 16 accessions being those characterized for presenting low seed oil content and elevated averages of plant height and number of commercial racemes. Seventeen accessions constituted the groups II (gray), presenting in common with low plant height and number of commercial racemes, and intermediated values to the other traits. The group III (green) was formed by seven that presented a low number of commercial racemes and higher seed oil content, 100-seeds weight, and internodes number.

Comparing the dendrograms obtained by the quantitative and qualitative descriptors (Figure 6), it is verified that there was not concordance among the groups formed in both dendrograms. The discordance of the formed groups can be confirmed by the absence of the correlation in both distances matrices verified by the Mantel test (r_m = 0.12; P0.05).

DISCUSSION

The elevated polymorphism among the qualitative descriptors as well as the significance of deviance analysis indicated a wide genetic variability among the 40 accessions of castor collected in the 23 Brazilian municipalities. The variability observed regarding the different morphoagronomic traits evaluated was already expected and corroborate with previously studies (Silva et al. 2017, Goodarzi et al. 2012, Oliveira et al. 2013, Bezerra Neto et al. 2010, Rukhsar et al. 2017, 2018, Rodrigues et al. 2015). For instance, authors observed values of plant height between 108.0 and 256.0 cm and racemes length between 13.71 and 44.38 cm (Bezerra Neto et al. 2010). Others reported values from 14.77 to 37.16 g to 100-seeds weight, 32.19 to 50.81% to the seed oil content, 4.17 to 7.33 number of racemes and 12.31 to 20.53 number of internodes in the castor accessions (Rukhsar et al. 2017). Generally, castor breeding programs seek to identify genotypes with high yield, high seed oil content, dwarf plants to facilitate mechanized harvest, beyond resistant to the main diseases, specially Fusarium wilt, charcoal rot and castor gray mold (Severino et al. 2012, Rodrigues et al. 2015, Saadaoui et al. 2017, Anjani et al. 2018).

In general, all others traits presented elevated estimates of SA_g and h², except the spores number of B. ricini for h². The SA_g it is a parameter of big relevance in the experimental quality assessment, taking into consideration not only the number of repetitions and the environmental quality but also the relation between the genetic and residual variations, being considered the most important parameter in the context of selective assessment (Ribeiro et al. 2017). The h² is a parameter of high importance in breeding programs since it is used to measure the phenotypic variation occurred by a genetic factor which means that reflects the proportion of phenotypic variance inherited (Falconer Mackay 1996). Authors studying genetics parameter in castor, reported h² values varying from 8.0 to 97.2%, being these values concordant to the values obtained in the current study for plant height, length of primary racemes, 100-seed weight and oil content (Rukhsar et al. 2018).

Although the castor gray mold is considered the main castor disease in Brazil, information regarding the genetics parameters and the heritage of castor resistance B. ricini is very scarce. The presence and the distribution of spines influence the disease development and the genotypes with a low number of spines were the most resistant (Lima Soares 1990). Controversial results were observed in the current study, since accession CB40, herein the most tolerant, showed an average density of spines, on the other hand accession CB10, herein the most susceptible to castor gray mold pathogen, had no spines. The spine density is considered a recessive monogenic characteristic conditioning by the gene s (“spineless”), where a SS plant has high spine density, and Ss has intermediated density (Gurgel 1945). In the presence of S, a not determined number of modifiers genes affect the final number and distribution of spines on capsules. Others morphoagronomic traits can influence the disease development under field conditions, such as plant architecture, raceme compaction, and internodes length (Milani et al. 2005). However, usually in an opposite way to that seeking by breeding programs, i.e., dwarf plants, with compact racemes usually are more susceptible to the pathogen.

The knowledge of the relationship among different traits is of great importance in breeding programs, mainly if the selection for one of those traits is difficult due its low h², measurement takes and/or identification, especially for perennial crops such castor. In the current study, the seed oil content and 100-seed weight traits were positively correlated. In this way, the 100-seed weight can be considered an important trait to indirect selection of genotypes with higher seed oil content, since it is considered a highly inherited trait and of easily measured. This correlation had already been reported by previous studies (Rukhsar et al. 2017, Adeyanju et al. 2010).

Using Singh’s relative importance analysis (Singh 1981) it was possible to classify the variables studied according to their respective contribution to the genetic divergences among the accessions and to identify the traits that contributed the most and the less to differentiate the accessions. In the present study, the number of commercial racemes and seed oil content were the variables that most contributed to the differentiation of the accessions, while the 100-seed weight and the number of internodes were less important. Discordant results were presented by other authors, where 100-seed weight was the variable that most contributed for the differentiation of castor accessions evaluated (Cavalcante et al. 2008).

The first two components of PCA explained more than 75% of the total variation observed, and the main groups formed were in agreement with those obtained for the quantitative traits using the Ward method. Based on the dendrograms obtained by the qualitative and quantitative descriptors, the 40 accessions of castor were allocated in three different groups in each dendrogram. However, it was not possible to observe a relation among the formed groups with the geographic origin of the accessions. The discordance among the formed groups and the accession geographic origin was also reported by others authors, using morphoagronomic traits and single nucleotide polymorphism (SNP), respectively (Arif et al. 2015, Foster et al. 2010).

The absence of correlation among the distance matrices from the qualitative and quantitative traits indicates that both kind of traits should be used for the differentiation among castor accessions. In general, studies showing the genetic diversity of castor germoplasm banks by means of qualitative descriptors are very scarce. On the other hand, was reported the complementarity of molecular and phenotypic markers in the study of castor diversity in India, emphasizing that the use of both kind of markers allowed a more precise distinction of genetic diversity present among the accessions evaluated (Rukhsar et al. 2017).

CONCLUSION

The present study revealed the presence of high genetic variability among the 32 traits assessed for the 40 castor accessions collected in Brazil. Those accession are been preserved and can be used as genetic variability source for castor breeding programs. The genetic parameters obtained in the present study, confirm that the variability observed was due to genetic factors rather than just environmental influence. The seed oil content and the 100-seed weight showed positive correlation inferring the possibility of indirect selection between these traits. The grouping analysis allocated the accessions in three different groups for the quantitative and qualitative traits. However, there was no concordance among the formed groups, indicating that both kind of traits are important in the accessions differentiation. Additionally, it was also not possible to verify concordance among the formed groups and origin of the accessions.

ACKNOWLEDGMENTS

The authors would like to thank the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) for financial support.

REFERENCES

• ADEYANJU AO, USMAN A & MOHAMMED SG. 2010. Genetic correlation and path-coefficient analysis of oil yield and its components in castor. Int J Appl Agric Res 5: 243-250.
• ALLAN G, WILLIAMS A, RABINOWICZ PD, CHAN AP, RAVEL J & KEIM, P. 2008. World wide genotyping of castor bean germplasm (Ricinus communis L.) using AFLPs and SSRs. Genet Resour Crop Evol 55: 365-378.
• ANJANI K, RAOOFA MA, PRASAD MSL, DURAIMURUGAN P, LUCOSE C, YADAV P, PRASAD RD, LALA JJ & SARADA C. 2018. Trait-specific accessions in global castor (Ricinus communis L.) germplasm core set for utilization in castor improvement. Ind Crops Prod 112: 766-774.
• ARIF M, KHURSHID H, SIDDIQUI SU, JATOI SA, JAN SA, ILYAS M & GHAFOOR A. 2015. Estimating spatial population structure through quantification of oil content and phenotypic diversity in Pakistani Castor Bean (Ricinus communis L.) germplasm. Sci Technol Develop 34: 147-154.
• BEZERRA NETO FV, LEAL NR, GONÇALVES LSA, RÊGO FILHO LDM & AMARAL JÚNIOR ATD. 2010. Quantitative descriptors to estimative genetic divergence in castor bean genotypes based on multivariate analysis. Rev Ciência Agron 41: 294-299.
• BOX GE & COX DR. 1964. An analysis of transformations. J Royal Stat Soc 26: 211-252.
• CAVALCANTE M, PAIXÃO SL, FERREIRA PV, DA SILVA MADALENA JA & DA COSTA JG. 2008. Divergência genética entre acessos de mamona em dez municípios de Alagoas. Caatinga 21: 111-115.
• CHARRAD M, GHAZZALI N, BOITEAU V, NIKNAFS A & CHARRAD MM. 2014. Package ‘NbClust’. J Stat Softw 61: 1-36.
• CRUZ CD. 2013. Genes: a software package for analysis in experimental statistics and quantitative genetics. Acta Sci Agron 35: 271-276.
• DANGE SRS, DESAL AG & PATEL SI. 2005. Diseases of castor. In: Saharan G et al. (Eds). Diseases of Oilseed Crops, New Delhi: Indus Publishing Company, New Delhi, India, p. 211-234.
• FALCONER DS & MACKAY TFC. 1996. Introduction to quantitative genetics. Longmans Green, Malaysia, 463 p.
• FAO. 2016. Faostat, http://www.fao.org/faostat/en/#data/QC/visualize Accessed september 26 2018.
  » http://www.fao.org/faostat/en/#data/QC/visualize
• FLORIN MJ, VAN ITTERSUM MK & VAN DE VEN GW. 2012. Selecting the sharpest tools to explore the food-feed-fuel debate: Sustainability assessment of family farmers producing food, feed and fuel in Brazil. Ecol Indic 20: 108-120.
• FOSTER JT, ALLAN GJ, CHAN AP, RABINOWICZ PD, RAVEL J, JACKSON PJ & KEIM P. 2010. Single nucleotide polymorphisms for assessing genetic diversity in castor bean (Ricinus communis). BMC Plant Biol 10: 10-13.
• GALILI T. 2015. Dendextend: an R package for visualizing, adjusting and comparing trees of hierarchical clustering. Bioinformatics 31: 3718-3720.
• GOODARZI F, DARVISHZADEH R, HASSANI A & HASSANZAEH A. 2012. Study on genetic variation in Iranian castor bean (Ricinus communis L.) accessions using multivariate statistical techniques. J Med Plants Res 6: 1160-1167.
• GREENWOOD JS & BEWLEY JD. 1982. Seed development in Ricinus communis (Castor bean). Descriptive morphology. Can J Bot 60: 1751-1760.
• GURGEL JTA. 1945. Estudos sobre a mamoneira (Ricinus communis L.). Tese: Livre docência - Piracicaba: Esalq.
• HARTLEY HO. 1950. The maximum F-ratio as a short-cut test for heterogeneity of variance. Biometrika 37: 308-312.
• LÊ S, JOSSE J & HUSSON F. 2008. FactoMineR: an R package for multivariate analysis. J Stat Softw 25: 1-18.
• LIMA EF & SOARES JJ. 1990. Resistência de cultivares de mamoneira ao mofo cinzento, causado por Botrytis ricini. Fitopatol Bras 15: 96-98.
• MANTEL NA. 1967. The detection of disease clustering and a generalized regression approach. Can Res 27: 209-220.
• MAPA. 2008. Ministério da Agricultura Pecuária e Abastecimento. Instruções para execução dos ensaios de distinguibilidade, homogeneidade e estabilidade de cultivares de mamona (Ricinus communis L.), 1st ed., Brasília: Embrapa, 10 p.
• MILANI M, NÓBREGA MBM, SUASSUNA ND & COUTINHO WM. 2005. Resistência da mamoneira (Ricinus communis L.) ao mofo cinzento causado por Amphobotrys ricini. Campina Grande: Embrapa Algodão, 24 p.
• MILANI M. 2008. Descritores de mamona utilizados pela Embrapa Algodão, Campina Grande: Embrapa Algodão, 39 p.
• MILLIGAN GW & COOPER MC. 1985. An examination of procedures for determining the number of clusters in a data set. Psychometrika 50: 159-179.
• OLIVEIRA DATOVO C & SOARES DJ. 2019. Sensitivity of field isolates of Botryotinia ricini to fluazinam and thiophanate-methyl. Trop Plant Pathol 44: 205-208.
• OLIVEIRA RSD, SILVA SA, BRASILEIRO BP, MEDEIROS EP & ANJOS EVAD. 2013. Genetic divergence on castor bean using the ward-mlm strategy. Rev Ciência Agron 44: 564-570.
• RESENDE MDVD. 2016. Software Selegen-REML/BLUP: a useful tool for plant breeding. Crop Breed Appl Biotechnol 16: 330-339.
• RIBEIRO MDFS & RAIHER AP. 2013. Potentialities of energy generation from waste and feedstock produced by the agricultural sector in Brazil: the case of the State of Paraná. Energ Policy 60: 208-216.
• RIBEIRO ND, STECKLING SDM, MAZIERO SM, DA SILVA MJ, KLÄSENER GR & CASAGRANDE CR. 2017. Experimental precision of grain yield components and selection of superior common bean lines. Euphytica 213-290.
• RODRIGUES HCDA, DE CARVALHO SP, DE CARVALHO AA, DE CARVALHO FILHO JLS & CUSTÓDIO TN. 2015. Avaliação da diversidade genética entre acessos de mamoneira (Ricinus communis L.) por meio de caracteres morfoagronômicos. Ceres 57: 773-777.
• RUKHSAR, PATEL MP, PARMAR DJ, KALOLA AD & KUMAR S. 2017. Morphological and molecular diversity patterns in castor germplasm accessions. Ind Crops Prod 97: 316-323.
• RUKHSAR, PATEL MP, PARMAR DJ & KUMAR S. 2018. Genetic variability, character association and genetic divergence studies in castor (Ricinus communis L. Ann Agrar Sci 16: 143-148.
• SÁ RO, GALBIERI R, BÉLOT J, ZANOTTO MD, DUTRA SG, SEVERINO LS & SILVA CJ. 2015. Mamona: opção para rotação de cultura visando a redução de nematoides de galha no cultivo do algodoeiro. Cuiabá: IMAmt, 12 p.
• SAADAOUI E, MARTÍN JJ, TLILI N & CERVANTES E. 2017. Castor bean (Ricinus communis L.), Diversity, seed oil and uses. In: Parvaiz A (Ed). Oil Seed Crops: Yield and Adaptations under Environmental Stress, New Jersey: Wiley, New Jersey, USA, p. 19-33.
• SAVY FILHO A, BANZATTO NV, VEIGA RFA, CHIAVEGATO EJ, CAMARGO CEO, CAMPO-DALL’ORTO FA, GODOY IJ, FAZUOLI LC, CARBONELL SAM & SIQUEIRA WJ. 1999. Descritores mínimos para o registro institucional de cultivares: mamona, 1st ed., Campinas: IAC, 7 p.
• SEVERINO LS ET AL. 2012. A review on the challenges for increased production of castor. Agron J 104: 853-880.
• SHAPIRO SS & WILK MB. 1965. An analysis of variance test for normality (complete samples). Biometrika 52: 591-611.
• SILVA ARD, SILVA SA, SANTOS LAD, SOUZA DRD, ARAÚJO GDM & MOREIRA RFC. 2017. Genetic divergence among castor bean lines and parental strains using ward’s method based on morpho-agronomic descriptors. Acta Sci Agron 39: 307-313.
• SIMÕES KS, SILVA SA, MACHADO EL & SILVA MS. 2017. Genetic divergence in elite castor bean lineages based on TRAP markers. Genet Mol Res 16: 1-12.
• SINGH D. 1981. The relative importance of characters affecting genetic divergence. Indian J Gen Plant Breed 41: 237-245.
• SOARES DJ. 2012. Gray mold of castor: a review. In: Cumagun CJR. (Eds). Plant pathology, Rijeka: InTech Publishing, Rijeka, Croatia, p. 219-240.
• SOARES DJ, NASCIMENTO JF & ARAÚJO AE. 2010. Componentes monocíclicos do mofo cinzento (Amphobotrys ricini) em frutos de diferentes genótipos de mamoneira. In: CONGRESSO BRASILEIRO DE MAMONA E SIMPÓSIO INTERNACIONAL DE OLEAGINOSAS ENERGÉTICAS, 4., João Pessoa, Anais, Campina Grande: Embrapa Algodão, p. 957-962.
• SUHAIL AH, SAKURE AA, BHAROSE AA, UMALE AA, SUSHIL K & SUBHASH N. 2015. Identification and characterization of low and high ricin containing castor (Ricinus communis L.) genotypes. Int J Plant Res 28: 92-97.
• SUSSEL AA, POZZA EA & CASTRO HA. 2009. Elaboration and validation of diagrammatic scale to evaluate gray mold severity in castor bean. Trop Plant Pathol 34: 186-191.
• VENEGAS-CALERÓN M, SÁNCHEZ R, SALAS JJ, GARCÉS R & MARTÍNEZ-FORCE E. 2016. Molecular and biochemical characterization of the OLE-1 high-oleic castor seed (Ricinus communis L.) mutant. Planta 244: 245-258.
• WANG ML, DZIEVIT M, CHEN Z, MORRIS JB, NORRIS JE, BARKLEY NA, TONNIS B, PEDERSON GA & YU J. 2016. Genetic diversity and population structure of castor (Ricinus communis L.) germplasm within the US collection assessed with EST-SSR markers. Genome 60: 193-200.
• WARD JH. 1963. Hierarchical grouping to optimize an objective function. J American Stat Assoc 58: 236-244.
• WEI T, SIMKO V, LEVY M, XIE Y, JIN Y & ZEMLA J. 2017. Package ‘corrplot’. Statistician 56: 316-324.

Publication Dates

• Publication in this collection
  06 Oct 2021
• Date of issue
  2021

History

• Received
  26 Aug 2019
• Accepted
  23 Dec 2019