Study area

The research was carried out in the municipality of Santa María Zoquitlán, Oaxaca, in the location named El Chilar, located at the coordinates 16°34’ N and 96°2’ W and at an altitude of 1 340 m, in plots with 10 to 50 % slopes, rocky soils and a warm, dry climate, an annual mean temperature of 22 °C, and precipitations of 300-500 mm (^{Inegi, 2008}).

Seed tree selection and germplasm collection

Both the selection of Bursera heliae trees and the collection of fruits were carried out in October and November, 2015. In plots with 48 % slope and an average altitude of 1 498 m. In the selection of parent trees, individuals that stood out for their size or the quality of their stem were sought (^{Vallejos et al., 2010}). Furthermore, this species was privileged because of its phenotype, the shape of its crown, its health and vigor, and of the fact that it was blooming and fruiting.

Once 17 trees were identified; 17 circular sites of 400 m² including the seed specimens were established in a targeted manner, according to the recommendations of the methodology described by ^{Zobel and Talbert (1988)}. An important criterion was that the trees should have fruits and should stand out among other individuals of the same species located in their vicinity. Their height (m) was determined using a (SILVA) clinometer, and the stem diameter at ground level (cm), the diameter at 1.30 (cm), the clear trunk height (m), number of forkings, number of branches, crown diameter (CD, m), with a STIHL measuring tape that was used to calculate the crown area (π/4 ×DC²). A GARMIN Drive^TM 40LM GPS was used to obtain general data of the site: coordinates, altitude, slope (%) and exposure. The number of shrubs, the number of associated trees per site, and the average distance between the trees and the shrubs were also registered. Fruits were collected from each tree; the impurities were removed, and the fruits were stored separately in brown paper bags, and the parent tree was identified, according to the procedures described by ^{Vallejos et al. (2010)}.

Fruit and seed characterization

A random sample of 50 fruits per tree was used to measure the equatorial diameter (ED) and the polar diameter (PD, cm) using a (TITAN^TM) vernier caliper, the weight of the fruit (g) was calculated using a VE-303 Model high-precision analytic scale with external calibration (VELAQUIN). The estimation of the shape coefficient --PD/ED-- was based on the diameters.

The fruits were then dried by placing them outdoors in paper bags with the corresponding identification, and exposed to direct solar radiation for a period of 10 to 15 days (^{Parrotta, 1994}). Once the fruits were opened, the seeds were taken and separated by parent tree in brown paper bags; their weight, PD and ED were quantified by groups of 50 seeds of each group

Statistical analysis of the data

The data of the fruits and seeds were systematized in a Microsoft Excel^TM spreadsheet, and all the analyses were processed with the Statistical Analysis System software package (^{SAS, 2004}). In order to find out whether or not the registers met the normality and variance homogeneity assumptions, the variables were subjected to the Shapiro and Bartlett test. The Arcsen (x) transformation was applied to those variables that did not meet the assumptions, such as the PD. A variance analysis was carried out using a completely randomized statistical model; a Tukey’s mean test was also applied (α=0.05), using as the independent variable each individual tree from which the seeds were collected. In order to determine the relationship, a correlation analysis was carried out between the morphometric variables of trees, fruits, seeds, and site, using the PROC CORR procedure. The similarities between morphometric variables of fruits and seeds, and the tree and site dasometric variables were estimated by means of a cluster analysis.

Tree characteristics

The 17 trees from which fruits were collected had heights ranging between 2.5 and 7 m; stem diameters at ground level of 55 to 149 cm; diameters at 1.30 of 30 to 156 cm; 1 to 3 forkings; 2 to 12 branches; a crown diameter (CD) of 2 to 8.5 m, and crown areas of 3.14 to 56.74 m². The great differences in size may be attributed to: 1) genotypes with different growth capacities; 2) the unknown ages, and 3) differences between the environments in which they grew. Notably, there is no published background on this topic for the Bursera genus. However, the influence of the genotype and the environment on the phenotypic manifestation of all living beings has been acknowledged (^{Torres and Magaña, 2001}).

Between 23 and 179 shrubs --mainly catclaw mimosa (unarmed shrubs) and candelilla (Euphorbia antisyphilitica Zucc.) -- were counted in the 400 m² sites, with a density of 3 900 shrubs per hectare. 3 to 31 tree species --e.g. bursera (Bursera bipinnata (DC.) Engl.) and holy copal (B. copallifera (Sessé ex Moc.) Bullock), cuachalala (Amphipterygium adstringens (Schltdl.) Schiede ex Standl.), organpipe cactus (Stenocereus thurberi (Engelm.) Buxb.), pochotes (Ceiba aesculifolia (Kunth) Britten & Baker f.), cacti Stenocereus sp), cazahuate (Ipomoea murucoides Roem. & Schult.), shuego (Cereus spp), huizache (Acacia farnesiana (L.) Willd.), among others-- were also quantified, with a density of 450 trees per hectare. The average distance between trees and shrubs was 4 to 8.5 m (Table 1). The sites where the trees were found are located at an altitude interval of 1 310 to 1 664 masl, in plots with a slope ranging between 24 and 70; out of the 17 sites, 11 had southeastern exposure, three had southwestern exposure, two had western exposure and one site had northeastern exposure.

Table 1 Characteristics of Bursera heliae Rzed. & Calderón trees and of the associated vegetation. 

D = Diameter; H = Height; F = Forkings; CD = Crown diameter; CA = Crown area; Dff = Diameter at the first fork; Shrub AD = Average shrub density; Tree AD = Average tree density; Min = Minimum; Max = Maximum.

Fruit and seed characteristics

Bursera heliae develops drupe type fruits, whose number ranged between 1 000 and 3 000 in the selected individuals. The variance analysis for four characteristics of fruits and seeds (Table 2) evidenced that only the weight and the equatorial diameter of the fruits showed significant differences (P≤ 0.05). As for the variables of the seeds, the weight, the equatorial diameter and the shape coefficient exhibited significant differences (P≤ 0.05). In both cases, these variations were determined by the differences between the trees.

Table 2 Summary of the variance analysis of various characteristics of Bursera heliae Rzed. & Calderón fruits and seeds. The independent variable was the tree from which the fruits were collected. 

v.s = Variation sources; df = Degrees of freedom; pd = Polar diameter; ed = Equatorial diameter; sc = Shape coefficient; *= Probabilities of F≤ 0.05; ** = F value; is= Insignificant.

The fruits had differences in size and average weight; thus, fruits collected from tree No. 10 and from tree No. 4 had a polar diameter of 0.94 and 0.75 cm, an equatorial diameter of 0.83 and 0.67 cm, and a weight of 0.44 and 0.25 g, respectively; in every case these magnitudes were significantly different (Tukey, 0.05) (Table 3).

Table 3 Characteristics of fruits collected from different parent trees (mean ± standard deviation). 

F weight = Fruit weight; Fpd = Fruit polar diameter; Fed = Fruit equatorial diameter; SCff = Shape coefficient; Fruits kg^-1 = Fruits per kilogram.

The fruits of trees Nos. 9, 10, 11, 12 and 17 were larger in weight, equatorial diameter and polar diameter than those collected from trees Nos. 3, 4, 6 and 7.

Each fruit contains only one seed (Figure 1), whose weight ranges between 0.074 and 0.135 g according to the tree. Seeds from trees Nos. 10 and 24 had a larger average size than those collected from trees Nos. 3 and 8. As for the polar diameter, two groups were differentiated, seeds from tree No. 9 being significantly larger than those collected from trees Nos. 3 and 4.

Figure 1 Seeds extracted from fruits of Bursera heliae Rzed. & Calderón. 

Seeds collected from trees Nos. 10 and 7 had average equatorial diameters of 0.54 and 0.544 cm, significantly (Tukey, 0.05) larger than the equatorial diameter of 0.448 estimated for specimens collected from trees Nos. 3 and 24. Likewise, seeds from trees Nos. 5 and 7 had shape coefficients of 0.953 to 0.957, being almost spherical, while seeds from trees Nos. 11 and 24 were 0.779 to 0.760, more elongated (Table 4).

Table 4 Characteristics of seeds collected from different parent trees (mean ± standard deviation). 

S weight = Seed weight; Spd = Seed polar diameter; Sed = Seed equatorial diameter; SCff = Shape coefficient; Seeds kg^-1 = Seeds per kilogram.

Notably, seeds collected from trees Nos. 7, 9, 10 and 24 were larger in weight, equatorial diameter and polar diameter than those from trees Nos. 3, 8 and 11.

Relationship between the fruits and seeds and characteristics of the site and of the parent tree

According to the linear correlation analysis carried out for the dasometric characteristics of the site and of the tree versus those of the fruits and seeds, the altitude above the sea level of the site had negative but significant correlations with the dry weight (p= 0.0002), the equatorial diameter (p= 0.004), the shape coefficient of the seeds (p= 0.0008), the fruit weight (p= 0.0001) and equatorial diameter (p= 0.0001). This indicates that, the higher the altitude of the sites where the trees were located, the smaller their fruits and seeds proved to be. The exposure of the site also evidenced positive low but significant correlations with the dry weight of the seed (p= 0.0045) and the fruit weight (p= 0.0001) and equatorial diameter (p= 0.0001).

The linear correlation analysis had a negative value, indicating the presence of smaller and lighter fruits as the altitude above the sea level increases.

In addition, the cluster analysis made it possible to distinguish between four groups (A, B, C and D) in the 17 samples of Bursera heliae trees, according to their geographic origin and to the characteristics of the tree and of the fruit.

Group A was characterized by Euclidian distances of 4 000; within it, the subgroups A1 and A2 were formed with different numbers of trees and Euclidian distances ranging between 1 500 and 3 000. Only two trees with very similar characteristics to those of subgroup A2 with small defects of height were integrated into subgroup A1. Subgroup A2 includes the trees with the largest dimensions and the fruits and seeds with the largest weight, polar diameter and equatorial diameter. In other species, the seeds have been shown to have a positive relationship with the vigor of the tree that grows from them (^{Khurana and Singh, 2001}). However, it should be mentioned that both the diameter and the weight of the fruits are characteristics that vary significantly between individuals, which are in turn are affected by the environmental conditions.

The studies carried out by ^{Cámara-Cabrales and Snook (2005)}, ^{Grogan (2001)} and ^{Gullison et al. (1996)} showed that other variables, such as height and the diametric category have an impact in the production of seeds in Bursera. According to other authors (^{Balcorta-Martínez and Vargas-Hernández, 2004}; ^{Niembro and Ramírez-García, 2006}), within the arboreal mass, the dominant specimens with the largest crown volumes tend to produce more fruits and better seeds, unlike co-dominant or intermediate trees, which show a reduction of their fecundity.

The level of variation in weight and size of both fruits and seeds of Quercus oleoides Schltdl. & Cham. has been observed to differ in populations located at different altitudes (^{Márquez-Ramírez et al., 2005}). That is, the variation in the fruits collected from different individuals of a given species is affected by the environmental situation to which each plant is subjected.

So far, the important effect of the weight of the fruits on the emergence, growth and survival of the seedlings has become evident, as proven by ^{Longe and Jones (1996)}, ^{Bonfil (1998)}, ^{Bonfil and Soberón, (1999)}, ^{Seiwa (2000)} in studies of holm oak trees; however, the manner in which the effect of the genotype and of the environment in which the tree grows and on the weight of its seeds is produced in these species is unknown (^{Rice et al., 1993}).

^{Villanueva (1993)} and ^{Alvarado-Vázquez et al. (2008)} point out that the analysis of the geographic patterns of the morphological variation between populations may contribute to identify locally adapted taxa; this would be helpful to increase the likelihood of success of reforestation programs intended to introduce adequate genotypes for each environment (^{González-Rodríguez and Oyama, 2005}).

Further research of the correlation of the fruit and seed weights and of the thickness, structure and function of the pericarp with the viability, germination, emergence and establishment of the seedlings is required to establish the ecological role of the seed as a source of reserves and of the fruits as dispersion units (^{Zavala, 2004}).

The variation in the same individual, between individuals and between populations of each species must also be assessed in order to corroborate what has been described for other taxa: “There are seeds of the same size that differ in their specific gravity and vice versa; likewise, seeds differing in size and specific gravity will also differ in germination as well as in the size and vigor of the seedlings” (^{Garrido et al., 2005}).

Since seeds of different sizes in relation to the various selected trees are available, it is recommended to carry out work in nurseries and plantations in contrasting environments in order to determine genetic and environmental effects.

Only two trees with similar characteristics of height and diameter to those of group A were integrated into group B; however, the trees of group A produced fruits smaller in weight and length. The remaining groups, C and D, consist of one tree each; these are specimens with small diameters, pronounced heights, and less thick crowns, which bore smaller fruits both in size and weight. The separation of these groups should be highlighted, between which phenotypic differences are found in the trees, the fruits and the seeds (Figure 2).

The letters show the groups formed: A, B positive variables; B, C negative variables in the characteristics of the fruits and seeds, dasometric and site variables

Figure 2 . Dendrogram of 17 collections of fruits of Bursera heliae Rzed. & Calderón; Ward’s minimum variance cluster analysis, based on 24 dasonomic, site and morphological characteristics