Index | Search | Home

link to pdf file

Guillen-Portal, F.R., D.D. Baltensperger, L.A. Nelson, and N. D'Croz-Mason. 1999. Variability in 'Plainsman' grain amaranth. p. 184–189. In: J. Janick (ed.), Perspectives on new crops and new uses. ASHS Press, Alexandria, VA.


Variability in 'Plainsman' Grain Amaranth

F.R. Guillen-Portal, D.D. Baltensperger, L.A. Nelson, and N. D'Croz-Mason*


  1. METHODOLOGY
  2. EXPERIMENTAL RESULTS
  3. CONCLUSION
  4. REFERENCES

Grain amaranth is a pseudo-cereal that played an important role as human food in the ancient civilizations of America. Current interest in amaranth resides in the fact that it exhibits a high nutritional value, a C4 photosynthetic pathway, a great amount of genetic diversity, and phenotypic plasticity (Dowton 1972; Hauptli 1977; Jain et al. 1979; Kauffman 1981; National Research Center 1984). In 1992, a cooperative amaranth breeding program between the Rodale Research Center and the University of Nebraska resulted in the release of the cultivar 'Plainsman', an interspecific hybrid between Amaranthus hypochondriacus L., a gold seeded selection from Mexico, and Amaranthus hybridus L., a black seeded selection from Pakistan. Breeding objectives for 'Plainsman' development were early maturity, light seed color, and short plant height. Single plant selection in earlier generations (F2 to F5) and mass selection in advanced generations (F6 to F7) was used (Baltensperger et al. 1992). 'Plainsman' maturity is very early in Western Nebraska, reaching maturity 110 days after planting. Evaluation tests in Nebraska, Colorado, Missouri, Minnesota, and South Dakota indicated that 'Plainsman' was one of the most promising amaranth cultivars for the United States. In 1994, approximately 1200 ha of 'Plainsman' were grown (Baltensperger 1992; Myers 1994) but field observations indicated a great deal of variation. The pollination mechanism of amaranth is complex in nature, varying from low to high outcrossing rates and furthermore being strongly affected by the environment (Jain et al. 1982; Hauptli and Jain 1985). The complexity of its pollination mechanism along with the breeding strategy employed in its development provided some evidence to suspect that some residual genetic variability is still present in 'Plainsman'. However, the large phenotypic plasticity observed in the amaranths might be the cause of such variability. Phenotypic plasticity has been defined as the physiological and/or morphological alteration by an organism in response to environmental differences (Schilichting 1986) and plastic variance as the amount of variation due to the environment and due to the genotype by environment interaction (s2Pt = sE2 + sG×E2). Plasticity has been defined as the ratio of plastic variance to total phenotypic variance (Scheiner and Goodnight 1984). Further improvement by selection within 'Plainsman' requires more knowledge about the cause of variability in seed production traits. The objectives of this study were to investigate the amount of morphological variation present in 'Plainsman', to estimate the genetic and environmental components of variance in agronomic traits in a random sample of 'Plainsman', and to predict the response to selection within 'Plainsman' under a selfing-selection scheme.

METHODOLOGY

'Plainsman' foundation seed from the Foundation Seed Division of the University of Nebraska was planted on 6 Jan 1995 in 140 pots. Sixty days after planting, plants were self pollinated by covering the panicles with pollinating bags. The panicles were hand-harvested 120 days after planting, allowed to dry at 32°C for three days, threshed, and cleaned by hand. Morphological traits measured included color of the main stem during the seed-filling period, inflorescence color during seed-filling period, branching index based on presence or absence of primary, secondary, or tertiary branches developed from the main stem, and inflorescence compactness just before harvesting.

In 1995, the 140 self pollinated families were planted at the High Plains Agricultural Laboratory near Sidney, Nebraska, and the Panhandle Research and Extension Center at Scottsbluff. At Sidney, a non-irrigated trial was planted on 16 June in a Duroc loam (Pachic Haplustol) soil and an irrigated trial on 17 June in a Keith silt loam (Aridic Arguiustoll) soil. The irrigated trial at Scottsbluff was planted on 18 July in an Otero loam (Fluventic Haplustol) soil using a Wintersteiger air seeder. Transplanting and hand thinning was necessary to achieve uniform populations at each experiment. Distance between plants was 0.15 m for the Sidney non-irrigated plots (51,000 plants/ha), 0.3 m for the Sidney irrigated plot (25,500 plants/ha), and 0.3 m at Scottsbluff (25,500 plants/ha). For all experiments, nitrogen was applied before planting at a rate of 100 kg/ha and 85 kg/ha was applied at panicle emergence at Scottsbluff. The plots were hand-weeded.

A replications-in blocks experimental design with two replications at each location was used. At each location, the experiment included ten blocks each containing a set of 14 randomly chosen families. The sets were kept together between replications and locations with a different randomization of families within each set. Single-row plots 5 m long with 0.76 m row spacing were used. Morphological traits measured were the same as those measured in the greenhouse. In addition, days to panicle emergence, flowering, and maturity, plant height, stem diameter, panicle length, grain yield per plant (the last four averaged over 5 plants randomly chosen per plot), and 1000-seed weight were determined. Morphologic data was analyzed by using a Chi-square goodness-of fit test between observations collected in the greenhouse (G0 generation) and those collected in the field (G1 generation). The least square mean method was used for the analysis of the agronomic data using Statistical Analysis System (SAS Institute, Inc. 1989). The analysis of data combined over locations was performed using a random linear model where the hypothesis test for genetic differences among families within blocks was H:sf2 = 0. Components of variance were estimated by equating the observed mean squares to the expected mean squares from the combined analysis of variance. The standard errors associated with these estimates were calculated as described by Anderson and Bancroft (1952). Broad sense heritability (H) was calculated as the ratio of the genetic variance to the phenotypic variance, H = sf2 /sP2 where sf2 = genetic variance among families, and sP2 = phenotypic variance. Plasticity (Pt) was calculated as the ratio of the plastic variance to the sum of the variance components, Pt = sPt2 / sPt2 + sf2+ se2 where sPt2 = plastic variance, sf2 = genetic variance among families, and se2 = residual variance. Approximate standard errors estimated for H and Pt were computed as SE(H) = SE(sf2 )/ sP2 and SE(Pt) = SE(sPt2)/ sPt2+sf2+se2. The predicted genetic gain from selection Gp was calculated as Gp = k sf2/sP where k = standardized selection differential, sf2 = genetic variance among families, sP = estimated phenotypic standard deviation.

EXPERIMENTAL RESULTS

Frequency distributions of morphological traits of selfed plants grown in the greenhouse (G0) and of resulting progeny grown in the field in Scottsbluff (G1) had three or four classes for all traits in both G0 and G1 generations, with one class being larger than the others (Table 1). No branches was the most frequent class for branching index with 39% of the total across generations. Pink-base stem color was present in 85%, while dark amaranthine inflorescence color was present in 93% of the plants. Lax density was the most prevalent class for inflorescence density with 52% of the plants. A Chi-square test of goodness-of fit to a G0:G1 ratio was highly significant for all the characteristics studied (Table 1), indicating that expression of these traits was greatly affected by the environment. It is evident that 'Plainsman' is characterized by a large amount of morphological variation, which is higher for branching index and inflorescence density than for stem and inflorescence color.

Table 1. Chi-square test of differences of classes within four morphological traits in two generations of 140 selfed families from the cultivar 'Plainsman'.

Trait

Classz

Generationy

c2

G0

G1

Branching index

No branches

56

49

21**

 

Few branches at the bottom

37

26

 
 

Few branches at the top

41

39

 
 

Branches all along the stem

6

18

 

Stem color

Green

13

13

18**

 

Pink base

110

119

 
 

Red or darker base

17

0

 

Inflorescence color

Green

1

2

13**

 

Dark amaranthine

125

129

 
 

Light amaranthine

14

1

 

Inflorescence density

Lax

92

51

127**

 

Near lax

15

47

 
 

Near dense

15

32

 
 

Dense

18

2

 

z Classes within traits were established following Brenner (1994).
y G0 based on 140 individuals grown in the greenhouse, G1, Based on 132 individuals grown in the field (Scottsbluff).
** significant at 1% level.

Panicle emergence and flowering days, and days to maturity were uniform among families (data not shown). Substantial variation over locations for all the agronomic traits was found (Table 2). Weather conditions contributed to this variation during the growing season. In late June, a hailstorm occurred at the Sidney non-irrigated location decreasing the plant population at this experiment to 10,000 plants/ha.

Table 2. Mean, maximum, minimum values and their corresponding standard errors (se) of five agronomic traits measured in 140 selfed families from the cultivar 'Plainsman'.

Parameter

Plant height (cm)

Stem diameter (cm)

Panicle length (cm)

Grain yield per plant (g)

1000-seed weight (g)

 

Sidney (non irrigated)

Max.

152

3.0

66

15.7

1.1360

Min.

69

1.0

25

0.7

0.5480

Mean

118

1.8

46

9.1

0.6642

SE

13.4

0.35

10.05

4.45

0.0324

 

Sidney (irrigated)

Max.

174

6.8

75

--

--

Min.

104

1.6

29

--

--

Mean

155

2.3

50

--

--

SE

8.52

0.3

6.81

--

--

 

Scottsbluff (irrigated)

Max.

228

3.8

93

52.3

1.0797

Min.

120

1.6

34

11.5

0.6130

Mean

170

2.6

59

26.8

0.7133

SE

10.54

0.25

6.08

7.90

0.0364

 

Combined over locationsz

Max.

228

6.8

93

52.3

1.1360

Min.

69

1.0

25

0.7

0.5480

Mean

147

2.2

51

18.3

0.6900

SE

11.02

0.32

7.94

6.57

0.0346

z Plant height, stem diameter and panicle length averaged over the three locations. Grain yield per plant and 1000-seed weight averaged over locations Sidney (non irrigated) and Scottsbluff.

Analysis of variance showed variation among locations for all of the traits (p < 0.01), suggesting that the growing conditions at each location were different (Table 3). Family differences were found for plant height and stem diameter (p < 0.01) and 1000-seed weight (p < 0.05). Family by location interactions were found for plant height (p < 0.01) and stem diameter (p < 0.05), suggesting these traits were more susceptible to environmental changes. It appears that 'Plainsman' possesses a genetic structure composed mainly by homozygous-heterogeneous lines. The analysis of variance components indicated that the genetic components were small compared to the other variance estimates (Table 4), which reinforces the idea about the homozygous structure of 'Plainsman'. The family by location (genotype by environment) interaction variance component was greater than the genetic variance component, suggesting that most of the variability in 'Plainsman' was plastic. A small negative estimate for grain yield per plant was obtained, but since it is small, it can be ignored. The phenotypic variance was greater than the genetic and the family by location components of variance for all the traits. Residual error variance estimates showed larger values than the rest of the components, except for plastic variance. The estimated error variances among locations were relatively uniform for 1000-seed weight, stem diameter, and plant height but not for panicle length and grain yield per plant where a 1:3 error variance proportion among locations was observed (data not shown). Plastic variance components as defined by Scheiner and Goodnight (1984) had the greatest value for all five agronomic characters, and plastic variance was at least 30 fold greater than the genetic variance. It is evident that separation of genetic variation from phenotypic variation is not easy in amaranth cultivars (Kauffman 1981).

Table 3. Combined analysis of variance of five agronomic traits measured in 140 selfed families from the cultivar 'Plainsman' tested at Sidney (irrigated and non irrigated conditions) and Scottsbluff, Nebraska, in 1995.

Source

Mean squares

df

Plant height (cm)

Stem diameter (cm)

Panicle length (cm)

df

Grain yield/plant (g)

1000-seed weight (g)

Locations (L)

2

161592.4**

36.34**

9201.7**

1

29171.3**

0.2321**

Blocks (B)

9

552.6**

0.58**

136.7*

9

160.2**

0.0025*

B × L

18

707.3**

0.26**

242.9**

9

65.5 NS

0.0006 NS

Replications/B/L

30

110.9 NS

0.17*

59.6 NS

20

63.7 NS

0.0023 NS

Families/B

130

227.5**

0.16**

72.4 NS

129

43.7 NS

0.0016*

F/B × L

251

169.5**

0.13*

71.4 NS

115

37.5 NS

0.0014 NS

Error

326

121.5

0.10

63

170

43.1

0.0012

C.V. (%)

 

7

15

15

 

36

5

*, ** significant at 5% and 1% levels.

Table 4. Estimates of components of variance and their associated standard errors (SE) of five agronomic traits in 140 selfed families from the cultivar 'Plainsman'.

Variance componentz

Plant height (cm)

Stem diameter (cm)

Panicle length (cm)

Grain yield/plant (g)

1000-seed weight (g)

Family
(sf2)

11.04 ± 6.05

0.006 ± 0.004

0.19 ± 2.08

1.99 ± 2.17

0.0000 ± 0.0001

Family by location
(sf×l2)

26.21 ± 9.74

0.015 ± 0.008

4.57 ± 4.39

-3.47 ± 3.88

0.0001 ± 0.0000

Phenotypic
(sP2)

43.31 ± 5.33

0.031 ± 0.004

13.78 ± 1.69

14.03 ± 1.61

0.0005 ± 0.0001

Plastic
(sPt2)

709.62 ± 485.49

0.168 ± 0.109

42.63 ± 27.98

157.06 ± 127.60

0.0013 ± 0.0010

Residual
(se2)

121.54 ± 9.49

0.108 ± 0.008

63.05 ± 4.92

43.15 ± 4.65

0.0012 ± 0.0000

zPlant height, stem diameter, and seed-head length estimates based on a combined analysis over three locations: Sidney (irrigated and non-irrigated) and Scottsbluff. Grain yield per plant and 1000-seed weight estimates based on a combined analysis over two locations: Sidney (non-irrigated) and Scottsbluff.

Estimates of broad sense heritability showed the largest heritability for plant height, followed by stem diameter, grain yield per plant, and panicle length. A zero heritability was observed for 1000-seed weight (Table 5). These estimates do not agree with those reported in the literature. Espitia (1994), in a population of amaranth races, found very high heritability of 0.92 for plant height and a moderately high heritability of 0.43 for grain yield. Joshi (1986), studying Indian landraces of amaranth, found high heritability of 0.77 for 1000-seed weight, 0.63 for inflorescence length, and 0.61 for plant height. Since the heritability estimates were small for the five agronomic characteristics studied, it is concluded that improvement for these traits through selection would be limited within this cultivar. Estimates of plasticity were high for all of the agronomic traits studied, indicating that they were greatly affected by the environment.

Estimates of the predicted gain from selection per year using three standardized selection differentials corresponding to a selection of the upper 10%, 5%, and 1% families based on estimates over locations show a relatively high genetic gain for plant height, a low genetic gain for yield per plant and stem diameter, and no genetic gain for 1000-seed weight (Table 6).

Table 5. Broad sense heritability (H) and plasticity (Pt) estimates and their associated standard errors.

Parameter

Plant height (cm)

Stem diam. (cm)

Panicle length (cm)

Grain yield/plant (g)

1000-seed wt. (g)

Broad sense heritability (H)

0.25±0.14

0.19±0.13

0.01±0.15

0.14±0.15

0.00±0.20

Plasticity (Pt)

0.84±0.58

0.60±0.39

0.40±0.26

0.77±0.63

0.52±0.40

Table 6. Predicted gain (Gp) from selection per year using three different standardized selection differentials (k) in five agronomic traits from the cultivar 'Plainsman'. Numbers in parenthesis correspond to genetic gain expressed as a percentage of the mean.

Standardized selection differential (k)

Expected genetic gain

Plant height (cm)

Stem diameter (cm)

Panicle length (cm)

Grain yield/plant (g)

1000-seed weight (g)

2.64 (1%)

4.43 (3)

0.09 (4)

0.13 (0)

1.41 (8)

0.00 (0)

2.06 (5%)

3.46 (2)

0.07 (3)

0.10 (0)

1.10 (6)

0.00 (0)

1.75 (10%)

2.94 (2)

0.06 (3)

0.09 (0)

0.93 (5)

0.00 (0)

CONCLUSION

The breeding biology of amaranth is complex in nature being strongly affected by the environment (Jain et al. 1982; Hauptli and Jain 1985). Based on theoretical considerations (Allard 1960; Simmonds 1979; Fehr 1987) the breeding method employed in the development of 'Plainsman' seems appropriate, although self-pollination was assumed to be the prevalent reproductive system in the developing populations (Weber and Kauffman 1990; Kauffman 1981; Schulz-Schaeffer et al. 1991). Under a single plant selection scheme, it is generally accepted that at the F5 generation near homozygosity is reached, so preliminary yield trials may begin at F6 generation (Allard 1960; Simmonds 1979). However, some residual genetic variability is retained, and it might be present indefinitely. Evidence of this has been found in cotton (Gossypium hirsutum) and sorghum (Sorghum spp.). Homologous pairing, new mutations, and recombination of linkage blocks promoted by homozygozity have been suggested as possible causes (Simmonds 1979). Amaranths in general are considered to have an intrinsic ability to attenuate the effects of strong environmental variations (Kauffman 1981). It has been hypothesized that plasticity and heterozygosity act in an opposite way. According to one hypothesis, phenotypic plasticity should increase as heterozygosity decreases due to the increase in developmental instability caused by deleterious homozygous recessive genes. Another hypothesis considers plasticity and heterozygosity as two antagonistic conditions in the sense that they represent alternative methods to deal with environmental heterogeneity. Thus, a population in which a consistent plastic response is observed has no need for genetic variation, and vice versa (Schlichting 1986). Yet under another hypothesis, plasticity and heterozygosity might well be expressed together so that a population could respond to an extremely variable environment by becoming both more plastic and more genetically variable (Scheiner and Goodnight 1984). The study showed the large extent to which environmental conditions affect the expression of morphological and agronomic traits in a population exhibiting a small degree of genetic variability, and that genetic improvement through a selfing–selection scheme would be limited.

REFERENCES


*Univ. of Nebraska Agr. Res. Div. J. Series no. 12496. The authors express their gratitude to Dr. Kent Eskridge for assistance in the statistical analysis of the results, and to Dr. Stephen Mason for manuscript review.
back