Plant material and its morphological description

Samples and photos of fruit and leaf were collected during the period 2012-2013 by randomly selecting one tree per orchard in 140 of the 423 gardens reported for the guava production area in Calvillo, Aguascalientes (^{SAGARPA, 2012}). The fruit and leaf morphology was described according to the classifications established by ^{UPOV (1987)} and supplemented with other descriptions (^{Sánchez and Peña, 2011}; ^{Sanabria, 2005}). In total, eight morphological descriptors were used for fruit: weight, size, equatorial diameter, number of carpels, shape, apex shape, base shape and cavity shape. For leaf: length, width, shape (of the complete leaf), apex shape and shape of the base. For the quantitative data, classification ranks were established in descriptive categories such as “small” , “medium” and “large”; or ordinal type labels according to the case (Table 1).

Table 1 Morphological characters of leaf and fruit observed in the gardens of Calvillo.  

Extraction of genomic DNA

The collected leaves were the source of the DNA, extracting it by the method of ^{Doyle and Doyle (1990)} with some modifications. The method consisted of grinding 100 mg of clean sheet with mortar and liquid nitrogen. The powder obtained was placed in a microcentrifuge tube containing 750 μl lysis buffer (100 mM Tris-HCl pH 8; 50 mM EDTA pH 8; 2% CTAB; 2M NaCl, 2% PVP 40; 3% β-mercaptoethanol) preheated to 65 °C and stirred for three minutes. Subsequently, a volume of chloroform-isoamyl alcohol (24:1) was added, carefully mixing for five minutes to pass a 15 min centrifugation at 13 000 rpm.

The aqueous phase was transferred to a new tube and a volume of chloroform-isoamyl alcohol was added, repeating the above procedure. Once the supernatant was taken into a new tube, 5 μL of RNase (10 mg mL^-1) was added and incubated for 30 min at 37 °C.

A volume of cold isopropanol was added, incubated for 10 min at -20 °C and centrifuged for 10 min at 13 000 rpm. The supernatant was discarded and the pellet was washed with 500 μL of 70% ethanol vortexing. After centrifuging the tubes at 13 000 rpm for 10 min, the supernatant was discarded. The excess ethanol was dried by evaporation and the pellet was suspended in 50 μL of TE (10 mM Tris-HCl pH 8; 1 mM EDTA pH 8). The DNA concentration in the sample was determined by a spectrophotometer (UV-6100A UV/VIS, Metash, Shanghai) and the samples were stored at -20 °C.

Procedure for performing the RAPD

The equipment with prepared DNA amplification reagents (GoTaq^® Green Master Mix, PROMEGA) was used by PCR, following the manufacturer’s instructions. After several tests to define the optimal conditions of DNA concentrations, primer concentrations and alignment temperatures, the parameters for the PCR reaction mixture (Table 2) and for the program of the thermocycler Techne TC-512 Thermal Cycler (Bibby Scientific Ltd., UK) (Table 3). A total of 22 random decamer primers designed by Operon Technologies and produced by IDT (Integrated DNA Technologies, USA) (Table 4) were tested. The amplification products were routinely run on 2.5% agarose gel with 1x TBE buffer for 2:45 h by applying 65V. Approximate molecular weight was referenced by molecular weight markers (50 bp, invitrogen).

Table 2 PCR reaction mix. 

Table 3 Program for the thermal cycler.  

Table 4 Primers used* (operon technologies).  

Elaboration of dendrograms and their analysis

The categorized data of molecular or morphological origin (based on absence/presence, or at predetermined range scales) were analyzed using the software Primer 5 (version 5.2.8 for Windows). The method of analysis of similarity was the one of clusters with the algorithm of Bray-Curtis with transformation of square root. With the generated similarity matrix, dendrograms were constructed based on the group average method (^{Clarke and Gorley, 2001}). The Spearman correlation range of the variables considered in the similarity matrix was performed in multiple runs, and the best calculated value was reported by taking five variables at a time per run for each of the dendrograms.

Dendrogram analysis based on morphological data

When analyzing the 16 morphological characters of leaf and fruit guava specimens were found that were classified in two or three categories according to the descriptor. Eleven characters correspond to mature fruit and five to developed leaf (Table 1). Analysis of the morphological data set showed a similarity range of 87 to 100% (Figure 1), with a Spearman correlation range of r= 0.719, which differs substantially from the results obtained with RAPDs. However, these results are comparable to those obtained by ^{Muñoz (2012)}, who used six descriptors for fruit and three for leaf of 79 guava trees belonging to an equal number of orchards. According to their results the range of similarity of morphological characters of Calvillo guavas ranges from 87 to 100% with a correlation range of r= 0.909. In the same way, we have concordance with the results of ^{Kujal et al. (2005)}, since they found that the RAPD data obtained by them presented much greater divergence than the morphology data. This could be evidence that during the evolutionary process the markers of neutral DNA polymorphisms have no effect on the ability of living beings to survive in the environment, and therefore, are not subject to selection pressure as occurs with the coding sequences. The results of ^{Domínguez et al. (2005)}, also indicate that changes occurring at the genome level in guava do not always correspond to those observed in the phenotype. ^{Gomes et al. (2010)} on the other hand, found partial agreement between the methods of clustering of RAPD markers and the morphological descriptors of guava. In contrast, ^{Feria (2008)} was able to associate three specific bands of a RAPD pattern with the accumulation of quercetin in guava.

Figure 1 Cladistic analysis based on morphological data. Spearman correlation r = 0.719; A= fruits with an elliptical basis;B= genotypes with mean leaf length and width; C= round fruits with equatorial diameter >44 mm and thickness of mesocarp >7 mm; D= fruits with four carpels and weight (40-70 g); E= fruits of piriform aspect with white pulp.  

Because the present work did not find a clear relation between the molecular markers and the guava descriptors, the morphological characters were analyzed separately, being found that five groups were generated in the similarity range between 91 and 94% described with their labels (Figure 1)

Dendrogram analysis based on RAPDs

From the preliminary tests with the primers polymorphic bands were found in several primers of the series OPA and OPH produced as a function of the genotype (Figure 2). The 10 polymorphic bands were used to elaborate the dendrogram based on RAPDs. The similarity values calculated for the 140 genotypes ranged from 25 to 100% and a Spearman correlation coefficient r= 0.794 (Figure 3), which coincides with that obtained by ^{Muñoz-Rodríguez (2012)}, who found a range of 30 to 100% similarity with a correlation range of r= 0.927 using RAPDs of only 26 genotypes from the Calvillo producing zone with series primers OPA (10, 05 and 07) and OPB (01 and 09).

Figure 2 Example of gel showing polymorphic bands (within the yellow rectangles) by RAPD in guava using the OPH-04 primer. At the top of the gel the codes assigned to the samples were recorded in the order in which they were sampled.MPM= molecular weight marker.  

Figure 3 Cladistic analysis based on RAPD performed with 140 guava DNA samples from Calvillo, Aguascalientes. Labels correspond to consecutively numbered samples. Correlation of Spearman observed r= 0.794.  

There is not clear relationship between the presence of a specific band with the morphological characteristics of the leaves or fruits of the sampled trees. This also coincides with the results of ^{Muñoz (2012)} even though the number of samples was equivalent to one fifth of the number of samples of the present work. The RAPDs in guava have been proposed to relate them to morphological characters or level of domestication. For example, ^{Tseng-Wei et al. (2007)} were able to identify two major guava groups from Taiwan using dendrogram derived from RAPD: commercial genotypes and wild genotypes. The latter group included two subgroups that generally distinguished between white pulp guavas and red pulp guavas. No RAPD band could be related to these groups.

In another case, with primers OPA-02 and 03, ^{Ahmed et al. (2011)} observed 33.2% polymorphic banding, which served to determine the genetic distances of 33 guava genotypes in a small range (0.5253-0.6631) range index. According to the authors, one of the clusters found is closely related to the morphology of the cultivated genotypes. ^{Bajpai et al. (2008)} obtained two different dendrograms of 22 guava genotypes using RAPD and direct DNA minisatellite amplification (DAMD) with genetic distances of 5 to 43%. The authors mentioned that the genotypes of the plains of the Indo-Ganges region are compacted into a single group, thus showing that RAPD markers can give a pattern of identification at least for that group.