MORPHOMETRIC ANALYSIS OF FRUITS AND SEEDS OF Annona crassiflora Mart. (ANNONACEAE) FROM CENTRAL BRAZIL ANÁLISE MORFOMÉTRICA DE FRUTOS E SEMENTES DE ANNONA CRASSIFLORA MART. (ANNONACEAE) DO BRASIL CENTRAL

– With the existing reduction of the vegetation cover of the Cerrado , several tree species have their areas of occurrence diminished, and as a consequence, they lose part of their genetic diversity. To counter this scenario it is necessary to know the genetic diversity in order to base projects and practical conservation measures. Thus, the objective of this work was to investigate the genetic variability of Annona crassiﬂ ora by means of morphometric data of fruits and seeds in four populations in central Brazil. In total, 152 fruits were obtained from 73 matrices, of which the height, diameter, fruit mass, mass of 100 seeds and number of seeds per fruit were measured. The fruits were stripped and the seed removed to measure their height, width and thickness. The hierarchical cluster study showed the grouping of 11 small groups and two large groups, and in these two large groups there are representatives of the four populations sampled. This shows that there is no speciﬁ c division of populations, indicating high genetic variability. The fruits of the Buritis population were, as a rule, signiﬁ cantly larger than those of other populations, which may indicate genetic distancing or diﬀ erent environmental conditions of pollination and dispersion. This pattern was not observed in the seed size analysis, although there was a statistical diﬀ erence between the populations. In general, the largest dimensions were found in the populations of Planaltina and Buritis . Therefore, the analyzes indicate high genetic diversity and fruits/seeds with larger dimensions in the best-preserved state.


INTRODUCTION
In the past decade the Cerrado had around 40% of the vegetation altered by human activity, as shown in the study by Sano, et al. (2007) and currently 46% of its native coverage has already been lost (Strassburg, et al., 2017). This can be explained by the fact that the Cerrado has the smallest percentage of legally protected area, only 5.2%, in the form of protected areas until the middle of the last decade (Jepson, 2005), and about 7.5% today (Soares-Filho, et al., 2014). As a result, many species are threatened by disorderly anthropic occupation and predatory extractivism. This leads to irreparable losses in the area of health, welfare and economy of society, as native species of great potential for economic use such as fruit, medicinal, timber or ornamental are replaced by agricultural crops. This replacement happens without due care to good practices for the conservation of natural resources.
In addition to the creation of larger protected areas, the restoration of degraded areas and the restoration of the legal reserve of rural properties with native species seedlings help in the conservation of species with economic potential (Hopkin, 2004). Studies on the genetic variability of these species generate information to support diff erent management practices beyond the establishment of germplasm banks and early stages of selection and genetic improvement (Faleiro, 2007).
Cerrado fruit trees play an important role in the conservation of native environments, but can also be commercially exploited, presenting agronomic potential. The main challenge of these species involves production and commercialization, where specifi c eff orts improve knowledge and enable the advancement of this new market (Vieira, et al., 2010).
Because they are native species, their nutritional demands are compatible with the local soil and require little or no input for implantation or permanence. In addition to commercial use, these species can be employed in projects for the recovery of degraded areas, fl ora enrichment and restoration of environmental protection areas. The fruits of these species are traded in fairs in central Brazil region and highway margins with great consumer acceptance and competitive prices (Vieira, et al., 2010).
Based on its economic, nutritional, social and environmental potential, with a view to fostering its use by smallholder farmers and rural communities is the araticum (Annona crassifl ora Mart.), Also popularly known as bruto, cabeça-de-negro, cascudo, marolo and pinha-do-cerrado.
Given the native characteristics and economic and social importance of A. crassifl ora, there is a need for seed production with adequate genetic quality, quantity production and low cost, for implementation of native species seed orchards for environmental restoration or even economical (Higa and Duque-Silva, 2006). These orchards, in addition to serving as seed suppliers for restoration of areas, can meet the precept of genetic conservation ex situ, as well as providing information and genetic material for the feasibility of genetic improvement programs.
However, when dealing with genetic material for forest restoration, the use of seeds with wide genetic variability is necessary, maintaining the gene pool of the species in the region, based on the use of seeds from the same collection and use zone. This fact is important because the genetic variation of species is associated with their geographical distribution. Defi ning the boundaries of seed collection and use zones should be based on experimental data identifying genetic variation, or by analyzing environmental factors that are likely to have the greatest infl uence on the creative forces of such genetic variation (Bower and Aitkens, 2008).
In response to the great economic, social and environmental importance of araticum, this study aimed to investigate the population genetic diversity of Annona crassifl ora through a detailed study of fruit and seed morphometric data in four populations in central Brazil.

Matrix selection, seed collection and benefi ciation
During March 2017, the process of selection of matrices was carried out in conserved or partially conserved forest environments of the Federal District, the municipality of Planaltina, state of Goiás and the municipality of Buritis, state of Minas Gerais. The sampling regions were characterized by typical Cerrado vegetation but, in all regions, there were areas close to degraded vegetation, mainly for agriculture, livestock and roads. The collections were performed in the following areas: Buritis -The matrices sampled in this area are located in pasture areas with remnants of nearby cerrado. The collection area is all surrounded by cerrado sensu stricto which allows good pollination, dispersion and random crossings between the matrices. Located in the municipality of Buritis-MG. 30 matrices.
FAL -Similar to Buritis area. It has a nearby urban area as well as agricultural research facilities such as grazing and planting of commercial cultivars. Located in the Água Limpa Farm of the University of Brasília-DF. 26 matrices.
Planaltina -Preserved area with rural occupation in its surroundings. Located in the municipality of Planaltina-GO. 23 matrices.
Sobradinho -The matrices of this region are remnants of urban occupation in the area. They are native individuals who were preserved to compose green areas and preservation of residential condominiums. Located in the administrative region of Sobradinho-DF. 16 matrices.
In all, seeds were collected from 73 matrices in four populations, having as criteria of selection the tree health, minimum distance of 50 meters between matrices, and seed production in the year of collection. After the ripe fruits were collected, they were placed in individual containers by matrix, with their respective identifi cations, and kept in a shady place. A total of 152 fruits were obtained from 73 matrices (1 to 5 fruits collected per matrix). All were measured (diameter and height) with a tape measure and weighed using precision digital scales. The seeds were removed from the fruits and separated from the impurities with later cold storage. At least 10 seeds from each matrix were separated for length, width and thickness measurements using a digital caliper. The mass of 100 seeds (per matrix) was evaluated using precision digital scales and following the guidelines of the Seed Analysis Rules of eight repetitions of 100 seeds each (Brasil, 2009).

Data analysis
Fruit and seed measurements were recorded in the Excel spreadsheet, Microsoft 2010, and statistical analyzes were performed using the IBM SPSS (2012) software. Descriptive analysis of quantitative variables was presented as mean, standard deviation, minimum and maximum values, quartiles and frequency histogram.
The comparison of fruit measurement data was performed by one-way analysis of variance, using the four study populations as a factor, Pearson correlation test, and hierarchical cluster measurement to evaluate intra and inter-population similarities. The hierarchical cluster study among the 73 matrices of the four populations studied was performed using the Ward method and Euclidean distance as a measure.
For seed data analysis, a 2-way ANOVA (population and matrix within each population) was used to evaluate the diff erence in measurements between the populations and between the matrices of the respective populations. Correlation analysis was also performed between the studied variables. The confi dence level adopted was 5%.

RESULTS
The fruits obtained had a diameter ranging between 9.10 and 19.40 cm and height between 8.37 and 19.01 cm. The diameter / height ratio ranged from 0.95 to 1.10, with a mass between 548.65 and 2157.89 g. The smallest amount of seeds per fruit was 30 seeds and the largest amount found was 169 seeds in the same fruit. The mass of 100 seeds ranged from 36.67 to 176.91 g (Table 1).
Pearson's correlation analysis (r) showed a high correlation between characteristics such as diameter and height (r = 0.830, p-value <0.001), diameter and mass (r = 0.806, p-value <0.001) and height and mass (r = 0.746, p-value <0.001). These data are expected considering that larger fruits tend to have greater mass. The other correlations were not signifi cant at 5% signifi cance.
In the hierarchical cluster study, 11 small groups and two large groups were grouped, and in these two large groups there are representatives of the four studied populations of. This shows that, based on the phenotypic characteristics of the fruits, there is no specifi c division of populations, indicating high genetic diversity within populations and among these populations of A. crassifl ora of central Brazilian cerrado (Figure 1).
Based on the Euclidean distances for the characteristics, it was possible to evaluate that the fruit mass contributed the most to the cluster formation, with 47.31% for the dissimilarity found in this analysis.
Seed length ranged from 13.18 to 23.67 mm, width 6.82 to 14.38 mm and thickness ranged from 5.39 to 10.65 mm ( Table 2). The ratio of width to thickness was calculated to evaluate the seed cylindricity, which ranged from 0.86 to 2.36 (Table 2). This shows that the cylindrical form is not predominant in the seeds of this species, but spherical or fl attened forms.
Correlation analysis was also performed between the variables analyzed for A. crassifl ora seeds.
There was signifi cant correlation between seed characteristics: length and width (r = 0.316, p-value <0.001), length and thickness (r = 0.234, p-value <0.001), width and thickness (r = 0.158, p-value <0.001). All correlations were signifi cant at 5% signifi cance, even with moderate to weak correlation between variables. This indicates that larger values of one seed measure tend to have larger values of the other parameter, even though this increase is not so large but signifi cant.
To assess whether there was a statistically signifi cant diff erence between the populations studied in relation to morphometric data, one-way or twoway analysis of variance was performed for fruit and It was observed that, for fruit measurements, there was a statistically signifi cant diff erence in relation to all variables except the 100 seed mass variable (p-value = 0.901) (Figure 2). For the diameter variable, the null hypothesis of equality of the diameters averages in the four populations was rejected (p-value <0.001). Only the Buritis population was statistically diff erent from the other (signifi cantly larger) populations, and they did not diff er from each other. For height (p-value = 0.011), the same result was found: the Buritis population presented statistically diff erent fruit height than the other populations (signifi cantly higher), and they did not diff er from each other. For diameter / height (p-value = 0.007), the Buritis population diff ered only from the fruits of the Sobradinho population (signifi cantly larger). The fruit mass presented p-value <0.001, showing that the fruit mass of the Buritis population was statistically higher than the fruit mass of the Sobradinho and FAL population. For seeds per fruit, this diff erence was also observed: p-value = 0.049, with the number of seeds per fruit of the Buritis population being statistically higher only than that of the Sobradinho population ( Figure 2).
Finally, for the mass of 100 seeds there was no division into groups, considering that the null hypothesis of diff erence of the average mass of 100 seeds of the four populations studied was not rejected.
Regarding the morphometric data of A. crassifl ora seeds, two-way analysis of variance (population and matrix) was used to assess whether there was a statistically signifi cant diff erence. The model to represent seed morphometric data was signifi cant for the intercept and the population and matrix factors for all traits analyzed. Regarding the corrected model, p-value <0.001 was evidenced for all characteristics: length (p-value <0.001), width (p-value <0.001), thickness (p-value <0.001) and L / E ratio. (p-value = 0.002). As with the model, the intercept and both factors were signifi cant (p-value <0.050) for all dependent variables.
The post hoc group comparison test used was the Tukey HSD test, which is indicated for data seed data, respectively. The unit evaluated in this case was the matrix, where in those with more than one fruit, the average morphometric data were used in the statistical analysis. The post hoc group comparison test used was the Tukey HSD test, whose test is indicated for data with homogeneous variances. Data with homogeneous variances. The data presented homogeneity of variances by the Levene test, considering that there was no rejection of the assumption of variance homogeneity (p-value> 0.05) for any variable dependent on seed measurements.
Regarding length, two groups were divided in the post hoc test: Buritis and Planaltina were separated from Sobradinho and FAL. That is, the seed length did not diff er statistically between Buritis and Planaltina, nor between Sobradinho and FAL, but diff ered between these groups, where the seed lengths of Buritis and Planaltina were signifi cantly shorter than those of Sobradinho and FAL.
For the width variable, the result of the post hoc test shows that there was division into four groups. This indicates that seed width diff ered statistically among all populations. Buritis presented the smallest widths and Sobradinho the largest.
Regarding the thickness variable, there was division into two groups, where only the Buritis population diff ered statistically from the others. Buritis seed thickness was statistically smaller than seed thickness of all analyzed populations. Finally, for the variable width / thickness ratio the seeds diff ered statistically between the Sobradinho population and the Planaltina and Buritis populations (Figure 3).

DISCUSSION
The data presented here for seed dimensions are close to those of Machado, et al. (2016), in a study with matrices from the Goiás state region, and of Pimenta, et al. (2013) in headquarters located in Mato Grosso. The variation in seed size of A. crassifl ora is normal for polyspermic fruits where there is competition for nutrients, which aff ects the fi nal size. Often, the seeds positioned at the ends of the endocarp, close to the integument, are smaller than those located inside the fruit (Rodrigues, et al., 2006). Thus, these measures may be more infl uenced by pollination (eff ectiveness) than by genetic diversity itself. Another point to be emphasized is the physiological seed maturity that directly refl ects the seed size. According to Bewley and Black (1994) during the seed fi lling phase it can vary greatly in volume and size which, in the context of this study, may have a refl ection on the morphometric diversity found.   The A. crassifl ora population in more conserved areas was able to produce larger, heavier and globular fruits, probably due to the greater availability of resources and pollinators. In this sense the larger number of seeds resulted in smaller seeds. The more anthropized regions presented smaller, lighter fruits, with fewer seeds per fruit, but with larger seeds, which may be a response to some environmental stress (Feller and Vaseva, 2014;Wani, et al., 2016).
Thus, it can be concluded that there is a great morphometric diversity between fruits and seeds of the four populations studied. There was a signifi cant diff erence in diameter, height, mass and number of seeds per fruit among the populations, and the population of Buritis, in general, had the largest fruits, followed by the population of Planaltina, FAL and lastly Sobradinho. Regarding seed morphometric data, the data were signifi cant, showing morphometric diff erences, where the seeds of Buritis and Planaltina presented the smallest sizes in relation to FAL and Sobradinho. This shows that Buritis and Planaltina presented larger fruits with smaller seeds, and were the Cerrado areas with the highest degree of conservation. The FAL region has many villas around with large land (2500 m²) and the Sobradinho region had even more villas (condominiums), with smaller areas (approximately 1000 m²) and bustling avenue. Thus, to improve A. crassifl ora fruit production for commercial and self-consumption purposes, the ideal is for the population to be in a nearby conserved environment, minimizing anthropogenic eff ects, since extractivism by itself does not negatively aff ect this populations (Orioli, 2017) and other Cerrado species (Giroldo and Scariot, 2015;Ferreira, 2016).
However, even though there were signifi cant diff erences between the morphometric/phenotypic characteristics of the fruits in the four populations, by multivariate evaluation of several characteristics, it was observed that there was no clear division between the populations (cluster analysis). The groups formed based on the phenotypic characteristics of the fruits showed groups with components (matrices) of all four populations. Thus, it can be concluded that there is high genetic diversity among all A. crassifl ora populations studied, as already reported by Zanella, et al. (2012). These authors also show that allogamous species, such as A. crassifl ora, tend to have high intrapopulation genetic diversity and little diversity between populations. This fact indicates that there is still genetic diversity of A. crassifl ora to be studied, conserved and explored.
The original occurrence area of A. crassifl ora has been suff ering anthropogenic pressures such as agricultural expansion, creation of housing and commercial sectors, burning and highways (Orioli, 2017). These pressures culminate in negatively aff ecting the vegetative, reproductive and productive performance of the species (Chacoff , et al., 2004). Natural replacement of individuals is reduced by all the pressures already mentioned and has serious consequences for local populations of A. crassifl ora. The low fruit production per individual, allied to the endogenous seed dormancy can create a scenario where the species cannot recompose its population. Thus, A. crassifl ora is losing more and more space for other competing species, especially forage grasses (Poaceae).
The creation and maintenance of native species germplasm banks is a low cost alternative and with great genetic results. Although it is a medium to long-term initiative, it encompasses aspects such as maintaining genetic variability through seed orchard seedlings (Pontes, et al., 2018), and the possibility of improvement and replacement of genetic material lost in other regions.
Fruit and seed morphometry can be monitored over the long term for comparative studies to monitor the size of propagules of the same population (Macedo, 2009). Such studies may show decreases in the size of fruits and seeds that come from losses of genetic diversity.

CONCLUSION
Cluster analysis showed two large groups, but with representatives of the four populations. This suggests in addition to the high genetic diversity, that this same diversity is at the intrapopulation level and not between populations.
The fruits of the Buritis population were signifi cantly larger than the other populations, except for the 100 seed mass variable. This larger size of the fruits suggests a greater genetic distance allied to conditions more favorable to pollination given the continuity and preservation of the adjacent Cerrado areas. This pattern was not observed in seed size.
Given this, the analyzes indicate high genetic diversity and larger fruits/seeds in the best conservation areas.

ACKNOWLEDGMENTS
The authors thank Água Limpa Farm from University of Brasília for their permission to carry out part of the collections and for their logistical and personal support for this work. We also thank the Department of Forest Engineering of the University of Brasília for the infrastructure used in this study.