**3. Results and Discussion**

#### **3.1 Morphological and phenological evaluation**

Mean data of morphological and phenological evaluations of amaranth are shown in table 2 and 3. From our long-term observations, genotypes with number of days from emergence to flowering higher than 100 days likely does not mature before early frost in autumn. The vegetation period in evaluated collection ranged from 92±0.00 to 163.00±0.00 days. Also height of plants in maturity and length of inflorescence is a very useful character. Both are

The electrophoresis was performed on 90 mA (45 mA / gel) and let to run for about 4 hours. The gels were stained with a solution of 0.1% (w/v) Coomasie Brilliant Blue (CBB) R250, 50% (w/v) methanol, 10% acetic acid, 0.02% (w/v) bromphenol blue salt for 1 day and destained with a solution of 25% (w/v) denatured alcohol and 3.5% (w/v) acetic acid, what lasted also 1 day. Gels were preserved in solution: 45% (w/v) denatured alcohol, 3% (w/v) glycerol for 2 hours, then dried and stored into cellophane sheets. The whole procedure including the test of the different extraction concentrations, the protein fraction separation procedure and the electrophoresis was repeated for the control of the correct experiment

All the extracted protein fraction samples were analyzed by chip capillary electrophoresis using commercial Experion Pro260 Analysis Kit for 10 Chips and the Experion automated electrophoresis system (Bio-Rad Laboratories, USA) for protein quantification according to the manufacturer's instructions. Experion automated electrophoresis station performs automatically all the steps of the gel-based electrophoresis (samples separation, staining,

For the statistical evaluation of morphological traits, analysis of variance (ANOVA) and the Tukey HSD test were used (software -Statistica 7.0 CZ). In the case of protein fraction proportion in accessions with different seed colour, the basic statistics of R statistics 2.10.0 software were used for calculation of mean x, standard deviation sx and p-values (adjusted

The SDS-PAGE spectra of total seed storage proteins and protein fractions were compared and confronted with the spectra of the chip capillary electrophoresis. The bands in the spectra were analyzed regarding the positions of the bands and also the relative intensity of the bands. The intensity of the bands was analyzed individually for each sample considering the intensity of the internal markers of the chip electrophoresis and the general intensity of all the bands in the sample. The intensity of the bands was expressed as the relative protein concentration measured by chip capillary electrophoresis what was the multiplication of numbers 0, 1, 2, 3 used in our statistics (0– no band, 1- light band, 2 – medium intensity band, 3 – dark band). The spectra expressed as the numerical values were analyzed by R statistics 2.10.0 software. The relationships between accessions were expressed by Pearson correlation using single linkage. The hierarchical clustering dendrogram was cut at the level

Mean data of morphological and phenological evaluations of amaranth are shown in table 2 and 3. From our long-term observations, genotypes with number of days from emergence to flowering higher than 100 days likely does not mature before early frost in autumn. The vegetation period in evaluated collection ranged from 92±0.00 to 163.00±0.00 days. Also height of plants in maturity and length of inflorescence is a very useful character. Both are

performance.

**2.3.3 Chip electrophoresis** 

**2.4 Statistical analysis** 

destaining, imaging, band detections, and data analysis).

by Holm correction, two sided Welch Two Sample t-test used).

of correlation 0.99 to show the well defined clusters.

**3.1 Morphological and phenological evaluation** 

**3. Results and Discussion** 


SD-standard deviation

Analysis of variance (ANOVA) and the Tukey HSD test were used for statistical evaluation (software - Statistica 7.0 CZ).

Different letters in the same row are statistically significant at p ˃ 0.05.

Table 2. Phenological evaluation of amaranths


Different letters in the same row are statistically significant at p ˃ 0.05.

SD-standard deviation

Analysis of variance (ANOVA) and the Tukey HSD test were used for statistical evaluation (software - Statistica 7.0 CZ).

Table 3. Morphological evaluation of amaranth

54.00±1.41de 137.50±3.54bcd 0.75±0.01cdefghijk pale 56.50±0.71de 137.50±45.96bcd 0.69±0.08bcdefghij pale 29.50±0.71bc 117.50±3.54abcd 0.68±0.04bcdefghij pale 34.00±1.41bcd 90.00±7.07abc 0.58±0.03abcdefg pink 44.50±0.71cde 127.50±3.54bcd 0.76±0.08cdefghijk black 66.00±1.41e 152.50±3.54cd 0.75±0.07cdefghijk black 30.50±0.71bc 167.50±3.54d 0.90±0.00ijk pale 36.00±0.00bcd 150.00±0.00cd 0.74±0.04cdefghijk pale 24.50±0.71a 147.50±3.54cd 0.74±0.00cdefghijk pale 29.00±0.00bc 100.00±0.00abc 0.88±0.00hijk pale 29.00±0.00bc 142.50±3.54bcd 0.85±0.00hijk pale 37.50±0.71bcd 137.50±3.54bcd 0.66±0.06abcdefghi black 45.50±0.71cde 102.50±3.54abcd 0.78±0.05efghijk pale 52.50±0.71cde 132.50±3.54bcd 0.86±0.15hijk pale 46.50±0.71cde 122.50±3.54bcd 0.93±0.10jk pale 41.50±0.71cde 130.00±8.49bcd 0.71±0.01cdefghijk pale 60.00±0.00e 132.50±3.54bcd 0.91±0.10ijk pale 36.50±3.54bcd 109.00±4.24abcd 0.84±0.08ghijk pale 47.00±0.00cde 125.00±0.00bcd 0.50±0.00abcd black 51.00±1.41de 142.50±3.54bcd 0.84±0.00ghijk pale 35.50±0.71bcd 92.50±3.54abc 0.63±0.11abcdefgh black 34.00±1.41bcd 92.60±3.54abc 0.40±0.04a black 43.50±0.71cde 152.50±3.54cd 0.96±0.00k pale 44.00±0.00cde 110.00±0.00abcd 0.70±0.00bcdefghijk black 38.00±0.00bcd 127.50±36.06bcd 0.5±0.02abc pink 42.50±0.71cde 92.50±3.54abc 0.63±0.04abcdefgh black 22.00±0.00a 53.00±0.00a 0.52±0.00abcde black 34.50±0.71bcd 77.50±3.54ab 0.44±0.04ab black 49.50±0.71cde 155.00±7.07cd 0.70±0.00bcdefghijk black 52.00±1.41de 137.50±3.54bcd 0.52±0.11abcde black 53.00±2.83de 127.50±3.54bcd 0.84±0.01ghijk pale 51.50±2.12de 95.00±42.43abc 0.73±0.03cdefghijk pale 36.50±2.12bcd 115.00±21.21abcd 0.54±0.17abcdef pale 47.00±0.00cde 102.50±3.54abcd 0.80±0.00fghijk pale 54.50±0.71de 137.50±3.54bcd 0.82±0.06ghijk pale 51.50±0.71de 127.50±3.54bcd 0.75±0.14cdefghijk pale 46.50±4.50cde 114.50±23.33abcd 0.77±0.01efghijk pale 44.50±3.54cde 119.50±17.68bcd 0.81±0.08fghijk pale 45.50±3.54cde 92.50±38.89abc 0.77±0.01defghijk pale 39.50±2.12bcd 107.50±38.89abcd 0.79±0.02efghijk pale

**Genotype** Mean±SD Mean±SD **Mean±SD** 

**2008** 42.71±9.90a 122.85±26.21a 0.72±0.15a **2009** 43.22±9.87a 119.02±26.36a 0.73±0.14a

Analysis of variance (ANOVA) and the Tukey HSD test were used for statistical evaluation (software -

Different letters in the same row are statistically significant at p ˃ 0.05.

Table 3. Morphological evaluation of amaranth

**Year** 

SD-standard deviation

Statistica 7.0 CZ).

**Inflorescence length (cm) Plant height (cm) WTS (g) Colour of seed** 

important for mechanized harvest by combine harvester. Lower plants with mean inflorescence are better for grain production and mechanized harvest. From our collection it is for example accession '120' with 95.00±42.43 cm height and 51.50±2.12 cm length of inflorescence. Taller genotypes are useful to develop varieties for feed utilization (Wu eta l., 2000). On the other hand, plant height could be influenced by increasing of number of plant per m2 (Jarošová et al., 1997). The value of weight of thousand seeds (WTS) is shown in table 3. In the relation with seed colour is clear, that the biggest WTS was observed in pale seeded samples. The seed size of the genera ranges from 0.37 to 1.21 g per 1000 seed weight according to Espitia-Rangel (1994). He noted that the low value corresponding to wild and weedy species and the high values to cultivated grain species. In our experiments the WTS ranged from 0.39 to 0.96 g.

#### **3.2 Protein content and content of protein fractions**

The results of the protein content analysis showed that the highest protein content (17.32 ± 0.82%) had *A. cruentus*accessions followed by *A. caudatus* (17.24±0.65%) and *A. hypochondriacus* (16.89±0.80%). It corresponds with other published data. Segura-Nieto et al. (1994) published, that the range of protein content is following: *A. cruentus* 13.2 – 18.2%, *A. hypochondriacus* 17.9% and *A. caudatus* 17.6 – 18.4%. The range of the total protein content into our collection (12.43 – 17.33%) was similar to the results of other authors investigating various amaranth genotypes (Barba de la Rosa et al., 2009). The amaranth albumins, globulins and prolamins formed 9.2 – 14.65%, 9.78 – 13.81% and 1.76 – 3.3% of total seed protein, respectively (Table 4). The glutelins with the residual nitrogen were the most abundant. It was in accordance with the results of Bressani & Garcia-Vela (1990) and Bejosano & Corke (1999a). The very low content of prolamins (1.76 – 3.3%) confirmed the results of several authors (Gorinstein et al., 1991a; Bejosano & Corke, 1999a; Petr et al., 2003). However, another group of authors reported several times more prolamins (Correa et al., 1986; Zheleznov et al., 1997; Vasco-Mendez & Paredes-Lopez, 1995). The differences between the results of these two groups of authors might be due to the different extraction methods (Fidantsi & Doxastakis, 2001). Significant differences between black, pale and pink coloured seeds in the content of albumins were detected. Content of albumins of the black seeded group (9.64 ± 0.40%) was significantly lower (p-value 4.10-3) than of the pale seeded group (13.21 ± 1.45%) and also lower than of the pink seeded group (11.39 ± 0.00; p-value 2.10-2). Bresani & Garcia-Vela (1990) did not observed any differences in the protein fractions distribution among species or cultivars of the same species, independent of the fractionation sequence used. However, our results showed that the black seeded varieties had the lowest albumin content. No significant differences in other protein fractions were detected.


Table 4. Total seed protein content and protein fraction content (in % of DW) of investigated accessions with respect the seed colour.

#### **3.3 Methodical approach to protein extraction**

According to our results, the chip capillary electrophoresis could replace the standard SDS-PAGE procedure, because it produced comparable results and what is more it could be performed routinely also in small laboratories thanks to its rapid performance. On the other hand, the chip capillary electrophoresis showed wider range of proteins spectra (up to 260 kDa).

The test of different concentrations was used for selection of the best extraction approach for chip and SDS-PAGE electrophoresis. By the chip capillary electrophoresis, the bulked samples of 100 seeds in 400 μl of extraction buffer were also tested. The chip capillary electrophoresis showed the high sensitivity and therefore the high concentration of the protein in the main bands resulted in their illegility. The protocol of chip electrophoresis does not provide many possibilities to chase the loaded amount of the sample. The satisfactory results of the chip electrophoresis brought the use of the single seeds.

For the SDS PAGE there were used single seed samples, bulked samples of 10 seeds extracted in 50 and 100 μl and bulked samples of 100 seeds extracted in 200 and 400 μl of extraction solution were used. The protein patterns of the samples extracted from the single seeds did not show the intensity required for the analysis of all the bands in the spectra (Figure 1). On the other hand, samples obtained by extraction of 10 seeds in 50 μl and 100 seeds extracted in 200 μl of extraction solution did not show clearly separated bands, what resulted in their illegility. In comparison with the spectra of the less concentrated samples (single seeds, 10 seeds in 100 μl of extraction solution), the main bands of the more concentrated samples were thick and joined together. The bands, which were in the less concentrated samples less intensive, were expressed so intensively that formed dark background what resulted in the impossibility of identification of the individual bands in the protein spectra. The protein spectra of the samples obtained by the extraction of 100 seeds in 400 μl were also over expressed, but the less intensive bands did not form the background, so the mayor bands were more easily identified, but several mayor bands joined together.

As the best approach for the total seed storage protein extraction for classical SDS-PAGE we selected bulked samples of 10 seed extracted in 100 μl of extraction solution. The bulked samples of 10 seeds extracted in 100 μl to be the most suitable tools, because of their clear expression of protein patterns and moreover they can be used when samples with higher number of seeds are not available. This selected approach differed from methodology selected by Drzewiecki (2001) who used 50 μl or by Gorinstein et al. (2005) who used 62.5 μl of extraction solution for 10 seeds bulked samples. The need for using more extraction solution in our study might be to consequence of higher protein extraction as a result of the proper seed crushing performed in our study which was not mentioned in the methodology description of other authors (Drzewiecki, 2001; Gorinstein et al., 2005).

When using total seed storage protein spectra for accessions identification by chip electrophoresis the single seed samples with several repetitions showed up as the best approach. These results were with accordance with Bradova & Matejova (2008) that compared whole seed storage proteins of wheat.

#### **3.4 Polymorphism of the glutelins**

The electrophoresis of the glutelin fraction is widely used for crop varieties identification. There were published several articles about wheat (Matejova&Bradova, 2008; Dutta et al., 2011), rice (Gorinstein et al., 2003), barley (Smith & Simpson, 1983), lupine (Vaz et al., 2004) etc. varieties identification based on glutelin patterns. Similarly amaranth glutelins showed polymorphism not only in position of bands but also in their intensity.

1 - a single seed extracted in 18 μl,

468 Genetic Diversity in Plants

According to our results, the chip capillary electrophoresis could replace the standard SDS-PAGE procedure, because it produced comparable results and what is more it could be performed routinely also in small laboratories thanks to its rapid performance. On the other hand, the chip capillary electrophoresis showed wider range of proteins spectra (up to 260 kDa). The test of different concentrations was used for selection of the best extraction approach for chip and SDS-PAGE electrophoresis. By the chip capillary electrophoresis, the bulked samples of 100 seeds in 400 μl of extraction buffer were also tested. The chip capillary electrophoresis showed the high sensitivity and therefore the high concentration of the protein in the main bands resulted in their illegility. The protocol of chip electrophoresis does not provide many possibilities to chase the loaded amount of the sample. The

satisfactory results of the chip electrophoresis brought the use of the single seeds.

For the SDS PAGE there were used single seed samples, bulked samples of 10 seeds extracted in 50 and 100 μl and bulked samples of 100 seeds extracted in 200 and 400 μl of extraction solution were used. The protein patterns of the samples extracted from the single seeds did not show the intensity required for the analysis of all the bands in the spectra (Figure 1). On the other hand, samples obtained by extraction of 10 seeds in 50 μl and 100 seeds extracted in 200 μl of extraction solution did not show clearly separated bands, what resulted in their illegility. In comparison with the spectra of the less concentrated samples (single seeds, 10 seeds in 100 μl of extraction solution), the main bands of the more concentrated samples were thick and joined together. The bands, which were in the less concentrated samples less intensive, were expressed so intensively that formed dark background what resulted in the impossibility of identification of the individual bands in the protein spectra. The protein spectra of the samples obtained by the extraction of 100 seeds in 400 μl were also over expressed, but the less intensive bands did not form the background, so the mayor bands were more easily identified,

As the best approach for the total seed storage protein extraction for classical SDS-PAGE we selected bulked samples of 10 seed extracted in 100 μl of extraction solution. The bulked samples of 10 seeds extracted in 100 μl to be the most suitable tools, because of their clear expression of protein patterns and moreover they can be used when samples with higher number of seeds are not available. This selected approach differed from methodology selected by Drzewiecki (2001) who used 50 μl or by Gorinstein et al. (2005) who used 62.5 μl of extraction solution for 10 seeds bulked samples. The need for using more extraction solution in our study might be to consequence of higher protein extraction as a result of the proper seed crushing performed in our study which was not mentioned in the methodology

When using total seed storage protein spectra for accessions identification by chip electrophoresis the single seed samples with several repetitions showed up as the best approach. These results were with accordance with Bradova & Matejova (2008) that

The electrophoresis of the glutelin fraction is widely used for crop varieties identification. There were published several articles about wheat (Matejova&Bradova, 2008; Dutta et al.,

description of other authors (Drzewiecki, 2001; Gorinstein et al., 2005).

**3.3 Methodical approach to protein extraction** 

but several mayor bands joined together.

compared whole seed storage proteins of wheat.

**3.4 Polymorphism of the glutelins** 

10a - bulk of 10 seeds extracted in 50 μl,

10b - bulk of 10 seeds extracted in 100 μl,

100b - bulk of 100 seeds extracted in 400 μl of the extraction solution.

M - wide range protein marker (bands in kDa).

Fig. 1. SDS – PAGE spectra of total seed storage proteins of sample obtained by different extraction approaches.

In the cluster dendrogram (Figure2), there were clearly separated the grain and the wild monoecious and the wild dioecious accessions. All investigated amaranth species had in common three major bands of the MW 21 – 23 kDa, but remarkable differences in the rest of the spectra were the reason for the segregation into three main clusters. The glutelin spectra of the grain amaranth varieties were very similar to the total seed storage protein patterns, but the main polymorphic bands were better distinguished because of the washing off the first three fractions during fraction separation procedure which probably formed the "background" of the spectra. The principal polymorphism was detected in following band positions 38, 39, 54, 58, 60, 64 and 65 kDa with three intensity levels (1-3). The amaranth glutelins showed up as the most abundant protein fraction by SDS-PAGE analysis also in the study of Bejosano&Corke (1999). The division of the grain amaranth glutelins into three major groups reported also Gorinstein et al. (2004) and Barba de la Rosa et al. (2009).

Figure 2 indicated three well defined clusters: grain species, monoecious wild species and dioecious wild species. The grain species *A. cruentus*, *A. hypochondriacus*, *A. caudatus* closely matched together with one sample *A. mantegazzianus.* There were clearly segregated clusters with the wild monoecious species (*A.wrightii*, *A. delfexus* and *A. retroflexus*) and the wild dioecous species (*A. australis*, *A. cannabinus* and *A. tuberculatus*).

*A. caudatus* group presentedwas two accessions '21' and '101' characterized by the dark band 60 kDa and the light band 39 kDa in their glutelin spectra. The *A. cruentus* cluster was clearly separated in the dendrogram of hierarchical distancing by the presence of the dark band of 58 kDa and of the light band in the position of 39 kDa. *A. hypochondriacus* accessions were characterized by the lack of any band in the position 58 kDa and by the presence of the dark band 54 kDa and the light band 38 kDa. The typical band (in the position 54 kDa) used for *A. hypochondriacus* recognition was qualified as characteristic for *A. hypochondriacus* by several authors (Drzewiecki, 2001; Marcone, 2002; Gorinstein et al., 2005), but its position was determined differently: as 55 kDa (Marcone, 2002) or 52 kDa (Drzewiecki, 2001) or in the case of protein fractions as 55 kDa, too (Thanapornpoonpong et al., 2008). The characteristic presence of the band 58 kDa in *A. cruentus* spectra and of the band 54 kDa in *A. hypochondriacus* spectra was confirmed by the results of Thanapornpoonpong et al. (2008).

Some of the accessions possessed extra light band of 65 kDa and were aggregated close to the *A. hypochondriacus* cluster. Their similarity to the other *A. hypochondriacus* varieties was expressed by very high correlation 0.987.

The dark band of 54 kDa, the dark band of 64 kDa and the light band in the position 65 kDa showed up in the glutelin spectra of the accession '134'. The accession '80' had the same glutelin spectra, but its band of 54 kDa was of medium intensity. These two varieties might be the hybrids of *A. hypochondriacus* and other unknown species which could have dark band of 64 kDa and light band of 39 kDa or they might be *A. hypochondriacus* varieties with some special properties that were not considered in our study. The accessions '132' with the dark band of 60 kDa typical for *A. caudatus* accessions was also present in the spectra and therefore the correlation between these accessions and the *A. caudatus* accessions was as high as 0.911. These accessions also showed the light band of 38 kDa and the medium intensity band of 54 kDa (typical marker for *A. hypochondriacus* spectra).

The dioecious wild species *A. australis*, *A. cannabinus* and *A. tuberculatus* formed a totally distinct cluster. They possessed several major dark bands of lower molecular weight 32 - 50 kDa. From this group, *A. cannabinus* and *A. australis* were the most similar, their correlation was 0.675. The monoecious wild species (*A. wrightii*, *A. deflexus* and *A. retroflexus*) and the dioecious wild species had in common one light band in the position of 65 kDa. The major dark bands of the monoecious wild species were of MW 29 - 66 kDa. The spectra of the monoecious wild species had some similarities with the spectra of the grain species. The grain species spectra were characterized by the two bands of MW 31 and 33 kDa while in the spectra of *A. retroflexus* these bands were just "shifted up" to MW 32 and 34 kDa. Protein fractions spectra of the wild species had not been published yet by other researchers. The results indicated the high correlation of the spectra of *A. retroflexus* and *A. wrightii* what confirmed the similarity observed by the first morphological descriptions made by Watson (1877).

Accessions possessing several bands of different intensities in the polymorphic area were qualified as the hybrid accessions. The accession '99' had in its spectra several bands in the polymorphic area: the dark band of 54 kDa, light band of 58 kDa, medium intensity band in the position of 60 kDa and the light band of 65 kDa. Its similarity with *A. hypochondriacus*  was expressed as correlation 0.901 and to the accession '95'. The accession '95' differed from the accession '99' just in the intensity of the bands of 58 kDa and 60 kDa (correlation 0.971). Varieties '62' and '110' were designated as hybrid varieties. They had the both bands of 54 kDa (marker for *A. cruentus*) and 58 kDa (marker for *A. hypochondriacus*) of medium

*A. caudatus* group presentedwas two accessions '21' and '101' characterized by the dark band 60 kDa and the light band 39 kDa in their glutelin spectra. The *A. cruentus* cluster was clearly separated in the dendrogram of hierarchical distancing by the presence of the dark band of 58 kDa and of the light band in the position of 39 kDa. *A. hypochondriacus* accessions were characterized by the lack of any band in the position 58 kDa and by the presence of the dark band 54 kDa and the light band 38 kDa. The typical band (in the position 54 kDa) used for *A. hypochondriacus* recognition was qualified as characteristic for *A. hypochondriacus* by several authors (Drzewiecki, 2001; Marcone, 2002; Gorinstein et al., 2005), but its position was determined differently: as 55 kDa (Marcone, 2002) or 52 kDa (Drzewiecki, 2001) or in the case of protein fractions as 55 kDa, too (Thanapornpoonpong et al., 2008). The characteristic presence of the band 58 kDa in *A. cruentus* spectra and of the band 54 kDa in *A. hypochondriacus* spectra was confirmed by the results of Thanapornpoonpong et al. (2008). Some of the accessions possessed extra light band of 65 kDa and were aggregated close to the *A. hypochondriacus* cluster. Their similarity to the other *A. hypochondriacus* varieties was

The dark band of 54 kDa, the dark band of 64 kDa and the light band in the position 65 kDa showed up in the glutelin spectra of the accession '134'. The accession '80' had the same glutelin spectra, but its band of 54 kDa was of medium intensity. These two varieties might be the hybrids of *A. hypochondriacus* and other unknown species which could have dark band of 64 kDa and light band of 39 kDa or they might be *A. hypochondriacus* varieties with some special properties that were not considered in our study. The accessions '132' with the dark band of 60 kDa typical for *A. caudatus* accessions was also present in the spectra and therefore the correlation between these accessions and the *A. caudatus* accessions was as high as 0.911. These accessions also showed the light band of 38 kDa and the medium

The dioecious wild species *A. australis*, *A. cannabinus* and *A. tuberculatus* formed a totally distinct cluster. They possessed several major dark bands of lower molecular weight 32 - 50 kDa. From this group, *A. cannabinus* and *A. australis* were the most similar, their correlation was 0.675. The monoecious wild species (*A. wrightii*, *A. deflexus* and *A. retroflexus*) and the dioecious wild species had in common one light band in the position of 65 kDa. The major dark bands of the monoecious wild species were of MW 29 - 66 kDa. The spectra of the monoecious wild species had some similarities with the spectra of the grain species. The grain species spectra were characterized by the two bands of MW 31 and 33 kDa while in the spectra of *A. retroflexus* these bands were just "shifted up" to MW 32 and 34 kDa. Protein fractions spectra of the wild species had not been published yet by other researchers. The results indicated the high correlation of the spectra of *A. retroflexus* and *A. wrightii* what confirmed the

intensity band of 54 kDa (typical marker for *A. hypochondriacus* spectra).

similarity observed by the first morphological descriptions made by Watson (1877).

Accessions possessing several bands of different intensities in the polymorphic area were qualified as the hybrid accessions. The accession '99' had in its spectra several bands in the polymorphic area: the dark band of 54 kDa, light band of 58 kDa, medium intensity band in the position of 60 kDa and the light band of 65 kDa. Its similarity with *A. hypochondriacus*  was expressed as correlation 0.901 and to the accession '95'. The accession '95' differed from the accession '99' just in the intensity of the bands of 58 kDa and 60 kDa (correlation 0.971). Varieties '62' and '110' were designated as hybrid varieties. They had the both bands of 54 kDa (marker for *A. cruentus*) and 58 kDa (marker for *A. hypochondriacus*) of medium

expressed by very high correlation 0.987.

intensity. Moreover, they possessed the light band of 39 kDa. The presence of the light band 39 kDa (typical marker for *A. cruentus*) was the reason for their higher correlation with *A. cruentus* group (0.920) than with A. *hypochondriacus* group (0.892). The variety '111' was exceptional. Moreover, it had higher correlation with *A. hypochondriacus* varieties (0.960).

Fig. 2. Relations among amaranth samples expressed by Pearson correlation in dendrogram
