Biological material

The samples used belong to the genebank of the Colegio de Postgraduados, San Luis Potosí Campus, México (Table 1). During the 2013 year, seeds of twelve castorbean accessions were completely random collected in the states of San Luis Potosí, Aguascalientes, Zacatecas and Jalisco, México; the castorbean plants were in zones between 1 400 and 2 400 masl. The twelve accessions were collected following transects in the different regions of influence, from plants without presence of pests and diseases, vigorous, with seed supply and agronomic production characteristics. Geo positioning data, characteristics of the plant and agri-environmental aspects as well as the description and identification of the collected materials were recorded (^{Isaza et al., 2017}).

Table 1 Origin of the R. communis L. accessions.  

Proximal chemical analysis of R. communis L. accessions

For the proximal chemical analysis, castorbean seeds with shell were used. The determinations were performed using techniques recommended by ^{AOAC (2002)}. The protein content (954.01) was determined in a Kjeldahl system (Labconco, USA) using a protein conversion factor of 6.25. The crude fiber content (962.09) was obtained by digestion with sulfuric acid and sodium hydroxide in a digester equipment (Ankom, USA). Moisture was determined by gravimetric method (7.003), using a drying oven (Felisa, México). For the ash content, calcination (923.03) was used using a terrgen muffle (Novatech, Mexico). The oil content was determined according to the methodology described by ^{Loredo et al. (2012)} in an oil extraction system (Soxtec, USA). Finally, carbohydrates were calculated by difference of other components already determined (^{Bello-Pérez et al., 2001}).

Extraction and quality assessment of R. communis L. oil

The oil was extracted by mechanical method through a prototype of stainless steel with a maximum capacity of 400 g of seed, equipped with a cylinder to contain the seed and a pressure piston coupled to a hydraulic press (Mikel, México) with a maximum pressure capacity of 700 kg F cm^-2. For the evaluation of the oil quality, parameters such as viscosity, density, acidity index (IA) and percentage of free fatty acids (%AGL) were determined. For the determination of viscosity and density, a Stabinger VM 3000 equipment (Anton Paar, Austria) was used, following the methodology reported in the ASTM D 445 standard. The density was calculated as the ratio between the mass and volume (g cm^-3) at the same pressure and temperature conditions. The percentage of free fatty acids (AGL) and acidity index (IA) were determined according to ISO660: 1983 and to the methodology described by ^{Firestone (1996)}, using titration with 0.1N potassium hydroxide and phenolphthalein as indicator. IA was expressed as the amount of KOH in milligrams, which is required to neutralize the free fatty acids in one gram of oil (mg KOH g^-1).

Statistic analysis

The results were expressed as the mean of three independent experiments ±standard deviation (SD). Differences between means were analyzed using the Tukey test (p< 0.05). A principal component analysis (ACP) was performed by the Ward method and plotted on a dendrogram. Statistical analyzes were performed using the Minitab 17^® and R^® programs.

Proximal chemical analysis

The results obtained for the proximal chemical composition of the accessions are shown in Table 2. The content for each of the parameters evaluated in the castorbean seeds showed statistically significant difference (p< 0.05). These differences can be attributed to several factors among which the genetic variability, the conditions of the collection site, seasonality and other environmental and growth factors stands out (^{Hidalgo et al., 2009)}. Due to these differences in all response variables, it is encouraging to implement multivariate strategies.

Los resultados son expresados como la media de tres réplicas ± desviación estándar (expresados en base seca). Diferentes letras en la misma columna expresan diferenciassignificativas (p< 0.05) en la prueba de Tukey. *= factor de conversión: 6.25; CHO= carbohidratos obtenido por diferencia.

Table 2 Proximal chemical analysis of R. communis L accessions. 

Oil

Most of the accessions of castorbean had an average oil content of 50%, however accessions such as SLPS3C1, AGSS3C1 and AGSS4C1 showed a decrease up to 8%. Similar contents were reported by ^{Armendáriz et al. (2015)}, who registered values between 42% and 50.5% for castorbean seeds collected in several Mexican states. Meanwhile, ^{Goytia-Jiménez et al. (2011)} recorded values between 12.2% and 64.84% in 151 accessions collected in the state of Chiapas. Authors such as ^{Bello-Pérez et al. (2001)}, obtained similar values to those reported in this paper in oilseeds such as peanut (Arachis hypogaeae) and sunflower (Helianthus annuus) (47% and 51%, respectively) values. Also, ^{Martín et al. (2010)} showed similar content in seeds with bioenergy potential such as Mexican pinion (Jatropha curcas) (49.1%), neem (Azadirachta indica) (39.7%), moringa (Moringa oleifera) (38.1%), trisperma (Aleurites trisperma) (62%) and cashews (Aleurites moluccana) (56.3%). Accordingly, it is confirmed that castorbean seed is a raw material for the production of oil which can be used in the pharmaceutical, cosmetological, lubricants industries and bioenergy production (^{Mosquera et al., 2016})

Protein

The SLPS11C1 accession had the highest protein content (16.02 ±0.36%), this result is consistent with ^{Reveles et al. (2010)}, who reported values of 16.7% and 18.6% in wild castorbean seeds Durango-México. ^{Onyeike and Acheru (2002)}, who describe average contents of 14.4% in Nigerian castorbean seeds. Meanwhile, higher contents were reported by ^{Perea et al. (2011)} who found up to 28.48% in castorbean seeds of the Tiripiteo variety. However, differences in protein content may depend on the genetic makeup and environmental conditions of the plant (^{Ortega and Rodríguez, 1979}). On the other hand, the protein content obtained in the evaluated accessions could potentially be used as a complement in the diet of cattle, sheep, goats and fish, among others. However, it must be taken into account that the presence of toxic substances (alkaloids, allergens, among others) is a limitation for the direct use of these seeds in animal or human food, which would require a prior detoxification.

Raw fiber

The average crude fiber values ranged from 12.62% to 21.15%, being lower for the SLPS6C1 accession and higher for JAL3C1. ^{Makkar, (1998)} reported similar values of crude fiber in Mexican pinion seeds (14.6% and 16.4%). Likewise, similar data are recorded in some fodder types for animals, such as soybean meal (12%) and sunflower meal (19%), although lower to alfalfa hay (45%) (^{Van Soest et al., 1991}). Based on the data obtained, it is recommended the use of detoxified castorbean meal, in ruminant animals, since the intake of food with high fiber content in mono-gastric animals can be a cause of digestion problems.

Ashes

The ash content obtained for the twelve accessions of castorbean ranged between 2.24% and 3.41%, being higher for the SLPS6C1 accession and lower for AGSS2C1. ^{Lucena et al. (2010)} reported values between 3.01% and 6.95% in a variety of R. communis brazilian at different maturity stages. The amount of ash in the castorbean accessions shows the total content of minerals, allowing to know in this way the presence of inorganic elements to generate co-products that could generate added value to this raw material. Meanwhile, ^{Severino et al. (2004)} emphasize the mineral content in the seeds of castorbean as an excellent conditioner for the soil.

Humidity

Moisture contents between 4.49% and 5.75% were obtained in the different accessions of castorbean. ^{Perdomo et al. (2013)}; ^{Perea et al. (2011)} reported values of 3.89% and 5.64%, respectively, in collections of Mexican origin. Variations in the moisture content could be due to the climatic conditions and development of the crop, as well as the harvest time of the fruits. Nevertheless, the results obtained for the castorbean seeds show a low content of humidity (<6%), characteristic that makes them less susceptible to deterioration processes by microorganisms actions, being able to be stored and conserved during a determined time without affecting its viability (^{Souza et al., 2016}).

Carbohydrates

Four accessions (SLPS6C1, SLPS3C1, AGSS3C1, AGSS4C1) of the twelve evaluated showed values higher than 20%. ^{Annongu and Joseph (2008)} found contents of 24.88% for carbohydrates in previously defatted castorbean seeds. According to the results, the castorbean seeds can be a good source of energy and complement to animals nutrition, considering, of course, its previous detoxification, as already mentioned above. If these carbohydrate contents are checked in studied seeds, this by-product could be used (paste resulting from the oil extractio) in ethanol production (^{Melo et al., 2008}).

Oil quality

The results obtained for the evaluation of oil quality of twelve accessions of castorbean are shown in Table 3.

Los resultados son expresados como la media de tres réplicas ± desviación estándar. Diferentes letras en la misma columna expresan diferencias significativas (p<0.05) en la prueba de Tukey. Viscosidad y densidad fueron evaluados a 40 °C.

Table 3 Oil quality of the twelve accessions collected from R. communis L.  

Viscosity

Viscosity values were determined in R. communis oil from 250.04 to 265.84 mm² s^-1. These results are within the reported by authors like ^{Costa and Rossi (2000)}; ^{Scholz and Silva (2008)}, who report values of 248.8 mm² s^-1 and 296.87 mm² s^-1 for oils obtained from castorbean seed, respectively. It has been reported that the presence of a hydroxyl (OH) group in carbon 12 would be responsible for the high viscosity of castor oil (^{Scholz and Silva, 2008}), these characteristics confer extra stability to oil and its derivatives to prevent the formation of hydroperoxides (^{Ogunniyi, 2006}). A high viscosity and lubricity over a wide range of temperatures and insolubility in fuels and aliphatic petrochemicals make it directly applicable as a lubricant for equipment operating under extreme conditions, as well as in the use in paints and varnishes, among others (^{Mutlu and Meier, 2010}).

Density

The average density obtained for oils the extracted of the twelve accessions was 0.9463 g cm^-3 (Table 3). Similar results are reported by ^{Perdomo et al. (2013)} who found average values of 0.9418 g cm^-3 in oils from seeds of Mexican origin. While, ^{Conceição et al. (2007)} reported values of 0.9573 g cm^-3 in seeds from Brazil. Differences in the determination of this parameter can be due to small traces of water or impurities in the oils, thus affecting the density of the substance.

Acidity index

The values for this quality index ranged from 0.5415 to 2.2178 mg KOH g^-1, ^{Pradhan et al. (2012)} characterized castorbean oil for the production of biodiesel with a value of 0.91 mg KOH g^-1. On the other hand, ^{Ogunniyi (2006)} indicates that castor oil is a polyhydroxylated compound of natural origin, which has as limitation a slight reduction of the hydroxyl number and the acid value during storage. The reduction of these values can be caused by the reaction between the hydroxyl and carboxyl groups of the oil molecule leading to the formation of stolidity. The acidity index in crude oil is influenced by post-harvest and storage treatments such as temperature, humidity and others.

Percentage of free fatty acids

The results obtained for the percentage of free fatty acids showed significant difference (p< 0.05) with a range from 0.272% to 1.1156% for the twelve accessions tested. Lower values of (%) AGL were found in the accessions SLPS11C1< ZACS2C1< JALS2C1. According to the requirements of the industrial sector, the percentage of free fatty acids should be less than 0.5%; however, for the obtaining of biodiesel, up to 3% is allowed, using homogeneous basic catalysts such as sodium or potassium hydroxide (^{Moser, 2009}).

Multivariate analysis

The Pearson correlation analysis at 99% confidence for the variables studied (Figure 1), express point values of the different relations between variables. The variation range of the correlation goes from -1 to 1. The results found allowed to determine that a higher content of crude oil, protein and fiber (67-86%) will have lower content of ash, moisture and carbohydrates in the seed and viscosity in the oil. While the acidity index and free fatty acids have a correlation of 100%.

Figure 1 Correlation coefficient between the chemical composition and quality of oil R. communis. Oil (%);protein (%); FC= crude fiber (%); ash (%), moisture(%); CHO= carbohydrate (%); viscosity (mm² s^-1);density (g cm^-3); IA= acidity index (mg KOH g^-1); AGL= free fatty acids (%).  

Principal component analysis

On the other hand, the results of the groupings and relations between the twelve accessions of castorbean studied are shown based on the analysis of main components of six variables of the proximal chemical analysis and four variables for the oil quality. In Table 4, it is concluded that with at least four main components, 88.1% of the total cumulative variance of the data is explained.

Table 4 Vectors and eigenvalues of main component analysis (CP).  

The main component 1 (CP1), allows to explain the 46.9% of the global variance; with a value of 4.69, this main component shows a greater association with the following four variables: carbohydrates (41.8%), protein (41.1%), lipids (38.2%) and crude fiber (38.2%), which have a high participation in the definition of CP1, so it can be interpreted that high values present in this main component tend to have high average values in these four nutrients mentioned above. The main component 2 (CP2), participates with its own value (1.98) and 19.8% of the explained variance, in which the variable humidity is found (60.7%), viscosity (55.5%) and density (55.3%), two variables that refer to the oil quality and one to the seed. In the main component (CP3), there is an acidity index (46.2%) and free fatty acids (46%), this component consists only of attributes of oil quality. Finally the last one in the CP4 is ashes which is not associated with any other variable, considering this component allows to add 7.6% of the variation of the explained process.

Analysis by clusters

According to the cluster analysis (Figure 2), the twelve accessions were classified into relatively homogeneous groups, using a similarity of 66.6%, by means of the Euclidean distance as a grouping method. Three predominant clusters were identified for the twelve accessions. The first one consists of: SLPS11C1, ZACS3C1 and AGSS2C1, the second consisting of: ZACS2C1, JALS1C1, ZACS1C1, JAL3C1 and JALS2C1; and the third by: SLPS6C1, AGSS3C1, AGSS4C1 and SLPS3C1.

Figure 2 Dendrogram of hierarchical grouping according to the chemical composition and oil quality of twelve castorbean accessions.