Dimensions of the plants

Table 1 shows the dimensions of the plants. The crown area was the attribute with the highest coefficient of variation: 32.2 %. The mean slenderness index was low, indicating that most of the plants (80 %) were wilder than they were tall, and only 20 % were almost spherical; this was due to the formative prunings and to the fact that the individuals were in full recovery from the pruning. However, it is advisable to modify the architecture of the plants, since the conical shape of the crown is considered to facilitate the harvesting of the seeds (^{Milthorpe, 2004}).

Dimensions of the seeds

The seed production per plant was low and highly variable, ranging between 0.6 and 241.7 g per plant (Table 1). This variable exhibited the highest coefficient of variation (105 %). The average length reveals that jojoba seeds are longer than they are wide; therefore, their mean slenderness index was 1.67, and only two plants exhibited very elongated seeds with a SI above 2.0.

Dimensions of the leaves

Table 1 shows the dimensions of the leaves; the leaf area had the highest coefficient of variation (17.4 %). In general, the length of the jojoba leaves was twice their width; hence their average elongation index of 2.1. For the same reason, they exhibit low values for their roundness and narrowness indices (0.60 and 0.65, respectively), and a Feret diameter of 2.1 cm.

Formation of groups

The results (Figure 1) showed the grouping of the population into four (cubic clustering criterion) or five groups, according to Hotelling’s t² pseudostatistic; hence, the formation of five groups was considered, and the resubstitution analysis proved that the groups exhibit no atypical individuals.

Figure1. Pseudostatistical tests to determine the number of groups of Simmondsia chinensis (Link) C.K. Schneid plants. 

Therefore, the 56 jojoba plants formed five working groups, which ranged between two individuals in group V and 21 plants in group II. Furthermore, the Mahalanobis distance test (Table 2) confirmed that all the plant groups are different (P < 0.0001).

The dendrogram (Figure 2) shows the female jojoba plants, identified by line and plant numbers (L_1-25, P_1-12) in the plantation with similar morphological characteristics and located in each group, respectively; in this case, the cutting line has a semi-partial R² of almost 0.06.

Figure 2 Dendrogram of 56 female Simmondsia chinensis (Link) C.K. Schneid plants. using morphometric data. 

In the canonical discriminant analysis (Table 3), it was determined that the first three canonical functions (CF) accounted for 95.56 % of the morphological variability in female plants; the CF 1, for 58.3 %; the CF 2, for 27.3 %, and the CF 3, for 9.9 %.

The variables most correlated with the canonical root 1 (Can 1) were the slenderness index (X 10) and the width of the seeds (X 11) in relation to their size and shape; while in the canonical root 2 (Can 2) they were the leaf perimeter (X 12) and Feret diameter (X 17) in relation to the size and shape of the leaves; and in the canonical root 3 (Can 3), the largest crown diameter (X 2) and the mean crown diameter (X 4) of the plants (Table 4).

Figure 3 shows the distribution of the plant groups in terms of the canonical values; in Can 1, the most negative records corresponded to group V, followed by groups II and IV. The latter shows the most negative values in C 2. Positive values were observed in groups I and III. In Can 3, positive values were obtained by the plants of group V and by the majority of the plants of groups III and IV.

Figure 3 Distribution of female Simmondsia chinensis (Link). C.K. Schneid. plants in terms of the values of the canonical discriminant analysis of the morphometric values. 

Table 5 shows the averages per group for the most discriminating morphometric variables; group V had the highest slenderness index of the seeds --2.4, which are the longest and narrowest seeds--, with a difference of 58 % between the length and the width, while groups III, IV and I have the broadest seeds.

The plants of group III had the largest perimeter, followed by those of groups I and II, which coincide with the recorded Feret diameter of 21.0 to 22.2 mm; the plants of group IV had the smallest perimeter and Feret diameter.

The maximum and mean crown diameters are variable, with little contribution to the explained variance (9.9 %) and can be regarded as an effect of management, not as intrinsic traits of the plants, as these were exposed to formative prunings of various intensities at least once a year.

In the multivariate analysis of the canonical roots, a highly significant difference (P < 0.0001) was observed between the three canonical roots. When comparing the means (Table 6) in Can 1, differences were observed in almost all the groups except between groups II and IV, which exhibited negative values and differed from the rest of the groups; in Can 2, groups with positive values (II, III and V) exhibited differences in relation to the rest of the groups, but not between them; in Can 3, with the highest positive value, group V was the only one with differences with respect to the rest of the groups. There was similarity between group IV and groups III and II, and between the latter and group I.

It is estimated that 6 % (^{Aryal and Ming, 2013}) of the angiosperms species (14 620 out of 240 000) are dioecious (^{Renner and Ricklefs, 1995}). This unisexuality of the plants stems from hermaphroditic ancestors and leads to sexual dimorphism between males and females, which is generally related to physiological and morphological characteristics of the life cycle (^{Kumar et al., 2014}).

Jojoba is a dioecious plant requiring cross-pollination; therefore, it exhibits a high variability in its phenology and in its morphological and performance parameters (^{Hassanein et al., 2012}); furthermore, according to the records, less than 1 % of the plants from seeds of wild populations have the potential to produce economically acceptable yields (^{Reddy and Chikara, 2010}).

This supports the results obtained for female jojoba individuals under conditions of irrigation, since these exhibit morphological and productive differences when grouped together. The size and shape of the seeds, the sizes of the leaves and of the plant crowns are the most discriminant variables in the makeup of the groups.

Apparently, polymorphism in jojoba seeds is an important adaptive strategy against the heterogeneity and unpredictability of the desert environment, so that large and medium-sized seeds germinate first, after adequate rainfall, and small seeds remain dormant longer; in general, however, jojoba seeds are large and heavy, and therefore they are exposed to immediate consumption by rodents (^{Matthews, 1994}).

According to the literature, the size of jojoba seeds varies according to the locality and the genotype; thus, the range of length is 0.8 to 2.33 cm, and the width is 0.5 to 1.33 cm. In addition, seed size has a positive correlation with seed weight, which ranges between 0.33 and 1.88 g seed^-1, with 660 to 3 300 seeds kg^-1 (^{Meyer, 2008}; ^{Inoti et al., 2015}; ^{Khan et al., 2015a}). During the present work this behavior was observed to occur, even under the same agronomic management; therefore, this variability is attributed to the genotype of the species and is consistent with a study according to which the vegetative morphology of jojoba is compatible with the idea of intralocus conflict, in which the same characteristic is expressed in a similar manner in both sexes and at least part of the observed variation is assumed to have a genetic basis (^{Prasad and Bedhomme, 2006}).

Large jojoba seeds are preferred for establishing plantations, since they have a higher percentage of germination than medium-sized and small seeds; besides, they produce more vigorous seedlings in the first months of growth, and there is a positive correlation between the size and the wax content (^{Meyer, 2008}; ^{Hassanein et al., 2012}; ^{Inoti et al., 2015}) --the most important feature for the cultivation and exploitation of jojoba. In average, the plants of group III have the broadest (10.0 mm) and heaviest (71.0 g) seeds.

A strong sexual dimorphism of the leaves has been observed in certain wild jojoba populations; in these, the leaves of female plants are larger and have a greater variability than those of the male plants. However, the degree of sexual dimorphism in the size of the leaves is reduced with the increase in the annual precipitation (^{Kohorn, 1995}). As for the leaf area, in the plants of group III it was 3.9 cm²; in groups I and II, it was 3.5 cm²; group IV exhibited the smallest leaf area. In all the groups, the leaf area of the plants was similar, as they were all under the same soil moisture regime.

Studies on the morphological characterization of jojoba are scarce, but they are necessary for determining the variability that exists in their area of origin or in plantations under agronomic management; most studies have focused on describing and comparing a single characteristic, such as the size of the plants, branches, leaves or seeds, or certain physical or chemical characteristics of the oil (^{Ayerza, 2001}; ^{El-Baz et al., 2009}; ^{Khan, et al, 2015b}).

Multivariate statistics are a suitable tool for assessing more than one attribute of the plant at the same time, and they are utilized to classify the plants according to their similarities or differences in all the measured characteristics, as well as to select those variables that contribute most to their characterization and to the detection of differences between and within groups (^{Núñez and Escobedo, 2014}).

With this procedure, it was possible to group female jojoba plants with similar morphological characteristics, in addition to identifying the most relevant variables for the characterization of the plants --related to the dimensions of the seeds and leaves--, which can be used as a basis for further studies in this and other locations. Furthermore, differences between the groups of female plants thus formed were successfully determined.