Descriptive statistics

The mass from the population of soursop fruit evaluated showed high variation (25.7%), with values between 837.8 and 2513 g, the maximum value was for A6 (Table 2). ^{Evangelista-Lozano et al. (2003)} report soursop fruit with values between 408 and 513 g from Jiutepec, Morelos, Mexico, while in Brazil (^{Do Sacramento et al., 2003}) report soursop mass between 2.39 and 3.21 kg. This suggests that the fruits from the assessed orchard outweigh those produced in Morelos and some trees produce fruit with similar mass to selections from Brazil.

The mass ratio of pulp, peel and seed was 71.0, 20.5 and 8.5%, respectively; observing high coefficients of variation due to its mass, between 25.0 and 39% (Table 2). Most pulp was obtained in the A6 material with more than 1.5 kg (Table 2), materials with greater mass of seed and peel were A2 and A2 trees, both surpassing the 200 g (Table 2). ^{Do Sacramento et al. (2011)} indicate that outstanding materials from Brazil had proportions between 83.1 and 85.8% in pulp, 7.1 and 9.7% in peel and 3 and 5% in seed. This indicates that materials from Nayarit have higher proportion of peel and seeds, thereby reducing the proportion of the pulp.

Diameter variables showed low coefficients of variation between 9.9 and 13.0 cm (Table 2). Hidalgo (2003) suggests that variables with lower coefficient of variation of 20% indicate little variability in these characters. The longitudinal and equatorial diameter ratio indicates that overall the population was 1.6; i.e. greater tendency to an elliptical shape (Table 2). The average number of seeds in various populations ranged from 19 and 311 (Table 2), the coefficient of variation was one of the greatest of all variables (47.2%). These results are attributed to soursop has pollination problems and fertilization of its flowers, due to the characteristics of morphology of the flower and pollinators thereof (^{Franco et al., 2001}) which caused high seed variation in fruits evaluated.

Soursop peel color was green (h between 151.7 and 164.9) opaque (C* between 9.4 and 21.4) with low luminosity (L* between 30.2 and 45.8) (Table 2). The chromaticity parameter showed the highest coefficient of variation in fruit color. ^{Evangelista-Lozano et al. (2003)} indicate that soursop fruits grown in Jiutepec, Morelos, the hue values were between h= 92.9 ± 2.5 and 96.7 ± 2 and luminosity between L*= 46.3 ± 2.7 and 47.5 ± 3.7, indicating a bright green little intense. These results indicate that fruits from Nayarit have a tendency to green than fruits from Morelos, although less bright.

Firmness was the variable with higher coefficient of variation (71.4), the values were between 3.93 and 42.3 N (Table 3). ^{Márquez et al. (2012)} indicate that soursop on mature stage its firmness is between 4.7 and 7.4 N, while in the stage over mature the average values are 3.6 N. This suggests that the average firmness of the population match with a mature stage, since firmness had average values of 8.3 N values (Table 2).

Soluble solids from the population under study were between 7.1 and 14 °Brix (Table 2). ^{Borrero et al. (1995)} indicate that soursop at physiological maturity reaches values of 7.0 °Brix. Three days after harvesting the fruits reach values between 10 and 16% (^{Paull and Duarte, 2011}). The ^{Ministry of Agriculture from Brazil (1999)} indicates that soursop must have a minimum of 9 °Brix in consumption. ^{Do Sacramento et al. (2012)} determined values between 12.1 and 13.8 °Brix in some selections from Brazil; while fruits from Morelos and Nayarit have reported values between 11 and 12 (^{Evangelista-Lozano et al., 2003}). pH of the soursop fruits showed minimum values of 3.1 and maximum of 5.1 and on average the population showed values of 3.6 (Table 2). The ^{Ministry of Agriculture from Brazil (1999)} indicates that the minimum pH value for soursop is 3.5; which is very close to the one from this work on average for the entire population. Finally titratable acidity was on average 0.7%, with higher values in the coefficient of variation (26.8%) (Table 2); these values are lower than 0.92-1.00 in soursop fruits from Brazil (^{Do Sacramento et al., 2003}) but similar to those reported by ^{Márquez et al. (2012)} for soursop from Colombia.

Clustering and principal component analysis

Cluster analysis showed that with a mean index of taxonomic distance of 1.26, four groups were formed (Figure 1). Group I located 5 trees: A1, A4, A11, A12 and A13, these fruits showed low acidity (<0.6%), the highest firmness (21.6 N), pH (3.79) and intermediate values for total soluble solids (10.97 °Brix) (Table 3). Group II 6 materials: A2, A3, A5, A6, A7 and A10, these fruits showed the highest values in dimensions, number of seeds (187.6), total mass (1718.4 g) and peel mass (264.1 g) and pulp (1 027 g), also the the highest value for titratable acidity (0.8%) (Table 3); group III consisted of a single tree (A9), this had the highest seed weight (128.8 g), lower firmness (7.2 N) and high titratable acidity (0.8%), with similar values to the latter group (Table 3). Group IV was formed by tree A8, where its fruits showed the lowest number of seeds (40.3), lower fruit mass (968.6 g), the color with less tendency to green (h= 156.8), opaque (C*= 13.5) and less bright (L*= 34.08) also the highest amount of soluble solids (Table 3).

Figure 1. Dendogram from 13 soursop materials (Annona muricata L.) built by UPGMA (average linkage) from 15 attributes. 

^{Castillo-Animas et al., (2005)} classified soursop fruit, from an orchard of Peñita Jaltemba, Nayarit in three sizes: a) small between 350 and 550 g b) medium between 560 and 750 g and large between 760- 1200 g. In this work all the fruits were large (Table 3). The fruits from groups with the highest number and seed mass can be important to be used in propagation of rootstocks or to search for useful antioxidants in different industries (^{Correa et al., 2012}). The tree from group IV showed higher values for soluble solids and lower seed proportion which could be considered the most promising for fresh consumption and industrialization.

The principal component analysis (CP) indicates that CP1, CP2 and CP3 explained respectively 34.71, 23.13 and 16.39% of variability, so that together accounted for about 74.3% of total variability (Table 4).

CP1 was defined by the size of the fruit, total mass, fruit pulp and its structures (Table 5). CP2 was defined by chromaticity and inversely by firmness and titratable acidity.

Finally CP3 was defined by color brightness (L*) and hue (h). In other characterization studies in tropical fruits like mamey and Mexican plum, mesocarp mass characteristics and color have determined the greatest variability and helped in group formation (^{Gaona- García et al., 2008}; ^{Alia et al., 2012}). It is important to consider in future works to expand the exploration area and add variables of physiological interest such as respiration, ethylene production, production of fermentable volatile, antioxidant activity, concentration of total phenols and ascorbic acid to obtain materials with good flavor, longer shelf life and greater contribution to human health.