**3.3 Northwest Minnesota - seventy-two cultivars, 4 environments**

Results from earlier research (Table 1; Figs.1, 2, and 3) provided preliminary, but convincing, evidence that grain yield and harvest seed [Fe] were closely related to planting seed [Fe] across several genotypes when conditions favorable for Fe deficiency prevailed. These studies, however, involved only 4 to 29 cultivars. To confirm this observation we evaluated 72 cultivars expressing a wide range of seed Fe concentrations. Our primary objective was to compare plant traits thought to represent measures of resistance to Fe deficiency. This collection of Maturity Group 00 (MG 00) genotypes had a range of planting seed Fe concentrations from 64 to 106 ug Fe g-1 seed. We assumed that the seed we obtained had been grown with adequate Fe availability. Plants were grown in nutrient solution as well as field nurseries. Nutrient solution culture procedures described by Chaney et al. (1992) were used to evaluate genotypes grown with moderate to severe Fe deficiency under controlled conditions. Other researchers have concluded that similar quantitative trait loci (QTL) are identified in nutrient solution and field tests and, therefore, both systems identify similar genetic mechanisms of iron uptake and/or utilization (Lin et al., 2000).

These seventy-two genotypes also were grown on high pH, highly calcareous soils at three locations in 2003 and one location in 2004. Measures of resistance to Fe deficiency were: harvest seed [Fe] (µg g-1 seed); harvest seed Fe content (µg 1000-1 seeds); and Fe removal (µg Fe m-2). Classification variables were: published visual chlorosis (VC); in-field visual chlorosis at V3 (V3); planting seed [Fe] (µg Fe g-1 seed) (FC); planting seed Fe content (µg Fe 1000-1 seeds) (FS); and relative chlorophyll concentration (SPAD values) (GC).

Importance of Seed [Fe] for

Fe

 Mean High

Mean Low

HI vs LOW

R1 vs R2 –

Rate 1 –

Rate 2 –

R1 vs R2 –

 Hi vs Low Fe

RR vs Conv.

Numerator Denominator

Label Estimates Standard

Numerator Denominator

grown at two rates of Fe-EDDHA under severe Fe deficiency.

Improved Agronomic Performance and Efficient Genotype Selection 39

**Type 3 Tests of Fixed Effects**

Rate 2797.27 247.64 6 11.30 <.0001

Rate 2363.11 247.64 6 9.54 <.0001

Fe 434.16 350.21 6 1.24 0.2614 RR 2806.51 181.92 6.99 15.43 <.0001 Conv. 2353.87 181.92 6.99 12.94 <.0001 RR vs Conv. 452.64 98.65 114 4.59 <.0001 Rate 1 – RR 2568.48 257.27 6.99 9.98 <.0001 Rate 2 – RR 3044.55 257.27 6.99 11.83 <.0001

RR -476.07 363.84 6.99 -1.31 0.2321

Conv. 2157.75 257.27 6.99 8.39 <.0001

Conv. 2550.00 257.27 6.99 9.91 <.0001

Conv. -392.25 363.84 6.99 -1.08 0.3168

**Contrasts**

Table 4. Analysis of variance of grain yield for 10 RR and 10 conventional soybean cultivars

Label df df F value Pr > F

(C1) 1 6 1.54 0.2614

(C2) 1 114 21.05 <.0001 C1 x C2 1 114 0.6717

**Estimates** 

error df t value Pr. [t]

Effect df df F value Pr > F

Fe Rate 1 6 1.54 0.2614 Entry 19 114 2.37 0.0026

Rate\*Entry 19 114 0.61 0.8950




† SPAD values are described as consistent and accurate measures of leaf chlorophyll content (umoles m-2 of leaf area) (Markwell, et al., 1995; Markwell and Blevins, 1999).

Table 3. Analysis of variance of SPAD† at V3 for 10 RR and 10 conventional soybean cultivars grown at two rates of Fe-EDDHA under severe Fe deficiency.

**Type 3 Tests of Fixed Effects**

Rate 35.0325 0.7198 6 48.67 <.0001

Rate 30.3862 0.7198 6 42.22 <.0001

Fe 4.6463 1.0179 6 4.56 0.0038 RR 33.8887 0.5781 9.94 58.62 <.0001 Conv. 31.5300 0.5781 9.94 54.54 <.0001 RR vs Conv. 2.3588 0.5482 114 4.30 <.0001 Rate 1 – RR 31.8850 0.8175 9.94 39.00 <.0001 Rate 2 – RR 35.8925 0.8175 9.94 43.90 <.0001

RR -4.0075 1.1562 9.94 -3.47 0.0061

Conv. 28.8875 0.8175 9.94 35.34 <.0001

Conv. 34.1725 0.8175 9.94 41.80 <.0001

Conv. -5.2850 1.1562 9.94 -4.57 0.0010

**Contrasts**

† SPAD values are described as consistent and accurate measures of leaf chlorophyll content (umoles m-

Table 3. Analysis of variance of SPAD† at V3 for 10 RR and 10 conventional soybean

cultivars grown at two rates of Fe-EDDHA under severe Fe deficiency.

Label df df F value Pr > F

Fe (C1) 1 6 20.83 0.0038

(C2) 1 114 18.51 <.0001 C1 x C2 1 114 1.36 0.2464

**Estimates**

Error df t Value Pr > F

Effect df df F value Pr > F

Fe Rate 1 6 20.83 0.0038 Entry 19 114 4.49 <.0001

Entry 19 114 0.67 0.8369

Label Estimates Standard

Numerator Denominator

2 of leaf area) (Markwell, et al., 1995; Markwell and Blevins, 1999).

Fe Rate \*

Mean High

Mean Low

HI vs Low

R1 vs R2 –

Rate 1 –

Rate 2 –

R1 vs R2 –

Hi vs Low

RR vs Conv.

Numerator Denominator




Table 4. Analysis of variance of grain yield for 10 RR and 10 conventional soybean cultivars grown at two rates of Fe-EDDHA under severe Fe deficiency.

Importance of Seed [Fe] for

used to measure resistance.

maturity, are evaluated.

written by Spehar (1994).

Improved Agronomic Performance and Efficient Genotype Selection 41

improvement in genotypic resistance mentioned earlier may be related to the plant trait

Classifying genotypes on the basis of relative chlorophyll concentration (GC) yielded homogeneous class variances; however, differences among class means were not statistically significant (Table 6). The severity of Fe chlorosis in nutrient solution culture was especially harsh and may have limited genotypic expression of resistance to those genotypes having high Fe-efficiency (Jessen, et al., 1988). Although other researchers have concluded that similar QTL for visual chlorosis are identified in nutrient solution and field tests (Lin et al., 2000), other QTL may be identified when integrated measures of resistance, manifest at

Classifying genotypes on the basis of planting seed Fe content (ng Fe seed-1 or ug Fe 1000-1 seeds) resulted in homogeneous class variances for each measure of resistance, whereas, differences among class means were significant only for harvest seed Fe content (Table 6; Fig. 10). Measures of planting seed Fe content should provide reliable measures of resistance to Fe deficiency defined as Fe accumulation. Using planting seed [Fe] to classify genotypes resulted in homogeneous class variances for Fe accumulation and harvest seed [Fe], but not for Fe removal. Differences among class means were significant for each measure of resistance. The heterogeneous class variances for Fe removal is a warning that planting seed [Fe] may not provide consistent, reliable estimates of resistance to Fe deficiency defined as Fe removal at harvest. Another interpretation is that Fe removal at harvest may not be an acceptable measure of resistance to Fe deficiency because it involves the primary yield component, seeds m-2. Grain yield is not necessarily a suitable measure of resistance to Fe deficiency, especially where Fe is not yield-limiting, such as on non-IDC prone sites (Helms, et al., 2010). Helms, et al. (2010) also noted that visual chlorosis scores could not identify the

The severity of Fe deficiency among environments ranged from almost no chlorosis (Fisher, MN, 2003; VCS=1.2) to mild chlorosis (Crookston, MN, 2003; VCS=2.3) to moderate chlorosis (Ada, MN, 2003; VCS=3.0) to severe chlorosis (Crookston, MN, 2004; VCS=4.2). Across this wide range of IDC severity, our results emphasize the difficulty of identifying superior genotypes when visual observations, either VC or V3, are used to classify genotypes, whereas, the importance of seed [Fe] for efficient genotype selection (consistency and reliability) is underscored. Genotypes were also ranked in each environment for three putative measures of resistance to Fe deficiency and for four classification variables thought to represent potential measures of resistance to Fe deficiency. These values were then

correlated to determine genotypic rank correlations among environments (Table 7).

When genotypes were ranked using published visual chlorosis scores or in-field visual chlorosis scores, there was little association with measures of resistance to Fe deficiency. In contrast, when genotypes were ranked using planting seed [Fe] or planting seed Fe content, there often was a close association with measures of resistance to Fe deficiency. These generalizations, however, do not include Crookston, 2004 where IDC was especially severe and some genotypes barely survived. Nonetheless, planting seed [Fe] and content are substantially superior to measures of visual chlorosis for identifying consistent, reliable estimates of resistance to Fe deficiency (Fig. 10). The consistency and reliability of using seed [Fe] as a measure of resistance to Fe deficiency in soybean is further illustrated in the article

Using results from this study of 45 cultivars of soybean grown on partly- and fully-limed acid soils in Brazil, it is possible to calculate a genotypic rank correlation coefficient

highest-yielding genotype even where Fe was yield-limiting.

We acknowledge that in our studies genotype and seed [Fe] are confounded and that Fe availability, as well as genotype, likely will influence final seed [Fe]. Although seed [Fe] and genotype are confounded, this is not unlike visual chlorosis score and genotype. A better approach would have been to use several genotypes each having a range of seed [Fe] from 50 to 120 µg g-1 seed. We were unable to identify or create these treatments. Nonetheless, we know from earlier research (Wiersma, 2005, 2007, and 2010) that large (25-50%) differences in agronomic performance, within the same genotype, often are associated with rather small (5-10%) differences in harvest seed [Fe]. In this case, correlations among agronomic characters and seed [Fe] are relatively small (r <0.4). We also know from field experiments that differences between cultivars of <10 µg Fe g-1 seed are not likely to be statistically significant (Wiersma, 2005, 2007,and 2010). Similarly, Shen et al. (2002) concluded that in wheat "In fact, it is impossible to distinguish completely the role of the genotype vs seed Fe content in the early response to Fe deficiency because the difference in seed Fe content can be an aspect of genotypic difference to Fe deficiency". It is reasonable to think that seed [Fe], seed Fe content, or Fe removal can also be considered aspects of genotypic differences in response to Fe deficiency.


\*, NS Significant at 5% level of probability, and not significant at 5% level of probability

Table 5. Genotypic rank correlations of 14 genotypes across several North Dakota locations during 2007, 2008, and 2009.

In-field visual chlorosis is better predicted using at planting seed [Fe] than the recorded visual chlorosis score, although the relationship is far from perfect. The complexity of using individual genotype means, without first classifying them into groups, is illustrated in Fig. 9. The extent of yellowing (VCS) among plots within a nursery often approaches a continuous distribution from green to yellow. Historically, this range of expression has been sub-divided into classes prior to analysis (Fehr, 1982). Thus, to evaluate relationships among characters, the 72 genotypes in each environment were first divided into 5 classes on the basis of several characters: recorded visual chlorosis (VC); visual chlorosis at V3 (V3); seed Fe concentration (FC); seed Fe content (FS); and growth chamber SPAD (GC). Then, Levene's F test (Littell et al., 2006) was used to assess homogeneity of error variances across classes for each measure used in classifying genotypes. Welch's F test was used to test the equality of means across the levels of the single class terms (Littell et al., 2006). Ideally, plant traits used to classify genotypes would have homogeneous error variances (non-significant Levene's F) and significant differences among class means (significant Welch's F).

Classifying genotypes on the basis of visual observations, either VC or V3, rarely (16%) yielded homogeneous class variances, although differences among class means were almost always (83%) significant (Table 6). The heterogeneous class variances indicate that VCSs may not be the most appropriate measures of Fe efficiency and suggest that the slow

We acknowledge that in our studies genotype and seed [Fe] are confounded and that Fe availability, as well as genotype, likely will influence final seed [Fe]. Although seed [Fe] and genotype are confounded, this is not unlike visual chlorosis score and genotype. A better approach would have been to use several genotypes each having a range of seed [Fe] from 50 to 120 µg g-1 seed. We were unable to identify or create these treatments. Nonetheless, we know from earlier research (Wiersma, 2005, 2007, and 2010) that large (25-50%) differences in agronomic performance, within the same genotype, often are associated with rather small (5-10%) differences in harvest seed [Fe]. In this case, correlations among agronomic characters and seed [Fe] are relatively small (r <0.4). We also know from field experiments that differences between cultivars of <10 µg Fe g-1 seed are not likely to be statistically significant (Wiersma, 2005, 2007,and 2010). Similarly, Shen et al. (2002) concluded that in wheat "In fact, it is impossible to distinguish completely the role of the genotype vs seed Fe content in the early response to Fe deficiency because the difference in seed Fe content can be an aspect of genotypic difference to Fe deficiency". It is reasonable to think that seed [Fe], seed Fe content, or Fe removal can also be considered aspects of genotypic differences in

Year

\*, NS Significant at 5% level of probability, and not significant at 5% level of probability

Year 2007 2008 2009

2007 1 0.6176\* 0.5472\* 2008 1 0.3714NS 2009 1

Table 5. Genotypic rank correlations of 14 genotypes across several North Dakota locations

In-field visual chlorosis is better predicted using at planting seed [Fe] than the recorded visual chlorosis score, although the relationship is far from perfect. The complexity of using individual genotype means, without first classifying them into groups, is illustrated in Fig. 9. The extent of yellowing (VCS) among plots within a nursery often approaches a continuous distribution from green to yellow. Historically, this range of expression has been sub-divided into classes prior to analysis (Fehr, 1982). Thus, to evaluate relationships among characters, the 72 genotypes in each environment were first divided into 5 classes on the basis of several characters: recorded visual chlorosis (VC); visual chlorosis at V3 (V3); seed Fe concentration (FC); seed Fe content (FS); and growth chamber SPAD (GC). Then, Levene's F test (Littell et al., 2006) was used to assess homogeneity of error variances across classes for each measure used in classifying genotypes. Welch's F test was used to test the equality of means across the levels of the single class terms (Littell et al., 2006). Ideally, plant traits used to classify genotypes would have homogeneous error variances (non-significant Levene's F) and significant differences among class means

Classifying genotypes on the basis of visual observations, either VC or V3, rarely (16%) yielded homogeneous class variances, although differences among class means were almost always (83%) significant (Table 6). The heterogeneous class variances indicate that VCSs may not be the most appropriate measures of Fe efficiency and suggest that the slow

response to Fe deficiency.

during 2007, 2008, and 2009.

(significant Welch's F).

improvement in genotypic resistance mentioned earlier may be related to the plant trait used to measure resistance.

Classifying genotypes on the basis of relative chlorophyll concentration (GC) yielded homogeneous class variances; however, differences among class means were not statistically significant (Table 6). The severity of Fe chlorosis in nutrient solution culture was especially harsh and may have limited genotypic expression of resistance to those genotypes having high Fe-efficiency (Jessen, et al., 1988). Although other researchers have concluded that similar QTL for visual chlorosis are identified in nutrient solution and field tests (Lin et al., 2000), other QTL may be identified when integrated measures of resistance, manifest at maturity, are evaluated.

Classifying genotypes on the basis of planting seed Fe content (ng Fe seed-1 or ug Fe 1000-1 seeds) resulted in homogeneous class variances for each measure of resistance, whereas, differences among class means were significant only for harvest seed Fe content (Table 6; Fig. 10). Measures of planting seed Fe content should provide reliable measures of resistance to Fe deficiency defined as Fe accumulation. Using planting seed [Fe] to classify genotypes resulted in homogeneous class variances for Fe accumulation and harvest seed [Fe], but not for Fe removal. Differences among class means were significant for each measure of resistance. The heterogeneous class variances for Fe removal is a warning that planting seed [Fe] may not provide consistent, reliable estimates of resistance to Fe deficiency defined as Fe removal at harvest. Another interpretation is that Fe removal at harvest may not be an acceptable measure of resistance to Fe deficiency because it involves the primary yield component, seeds m-2. Grain yield is not necessarily a suitable measure of resistance to Fe deficiency, especially where Fe is not yield-limiting, such as on non-IDC prone sites (Helms, et al., 2010). Helms, et al. (2010) also noted that visual chlorosis scores could not identify the highest-yielding genotype even where Fe was yield-limiting.

The severity of Fe deficiency among environments ranged from almost no chlorosis (Fisher, MN, 2003; VCS=1.2) to mild chlorosis (Crookston, MN, 2003; VCS=2.3) to moderate chlorosis (Ada, MN, 2003; VCS=3.0) to severe chlorosis (Crookston, MN, 2004; VCS=4.2). Across this wide range of IDC severity, our results emphasize the difficulty of identifying superior genotypes when visual observations, either VC or V3, are used to classify genotypes, whereas, the importance of seed [Fe] for efficient genotype selection (consistency and reliability) is underscored. Genotypes were also ranked in each environment for three putative measures of resistance to Fe deficiency and for four classification variables thought to represent potential measures of resistance to Fe deficiency. These values were then correlated to determine genotypic rank correlations among environments (Table 7).

When genotypes were ranked using published visual chlorosis scores or in-field visual chlorosis scores, there was little association with measures of resistance to Fe deficiency. In contrast, when genotypes were ranked using planting seed [Fe] or planting seed Fe content, there often was a close association with measures of resistance to Fe deficiency. These generalizations, however, do not include Crookston, 2004 where IDC was especially severe and some genotypes barely survived. Nonetheless, planting seed [Fe] and content are substantially superior to measures of visual chlorosis for identifying consistent, reliable estimates of resistance to Fe deficiency (Fig. 10). The consistency and reliability of using seed [Fe] as a measure of resistance to Fe deficiency in soybean is further illustrated in the article written by Spehar (1994).

Using results from this study of 45 cultivars of soybean grown on partly- and fully-limed acid soils in Brazil, it is possible to calculate a genotypic rank correlation coefficient

Importance of Seed [Fe] for

Improved Agronomic Performance and Efficient Genotype Selection 43

Fig. 10. Relationships between seed Fe content at harvest and four classification variables. Each regression equation is represented by the same symbol and line type for each

environment in each panel: filled circle, dotted line is Fisher, MN, 2003; filled square, solid line is Ada, MN, 2003; open circle, short dash line is Crookston, MN, 2003; open square, long

dash line is Crookston, MN, 2004.

Fig. 9. Regression of varietal rank for visual chlorosis at V3 on varietal rank for published visual chlorosis score (VC) and planting seed [Fe].

Fig. 9. Regression of varietal rank for visual chlorosis at V3 on varietal rank for published

visual chlorosis score (VC) and planting seed [Fe].

Fig. 10. Relationships between seed Fe content at harvest and four classification variables. Each regression equation is represented by the same symbol and line type for each environment in each panel: filled circle, dotted line is Fisher, MN, 2003; filled square, solid line is Ada, MN, 2003; open circle, short dash line is Crookston, MN, 2003; open square, long dash line is Crookston, MN, 2004.

Importance of Seed [Fe] for

Measure of

ug Fe 1000-1 seeds

ug Fe g-1 seed

Fe removal (ug Fe m-2)

ug Fe 1000-1 seeds

ug Fe g-1 seed

Fe removal (ug Fe m-2)

ug Fe 1000-1 seeds

ug Fe g-1 seed

Fe removal (ug Fe m-2)

Classification variable

Published visual chlorosis score

(VC)

Field visual chlorosis (V3)

Planting seed [Fe] (FC)

Improved Agronomic Performance and Efficient Genotype Selection 45

deviation‡

VC1=9.2 VC2=8.5 VC3=8.6 VC4=7.9 VC5=7.3

VC1=19.3 VC2=20.2 VC3=17.5 VC4=14.9 VC5=13.2

VC1=3.7 VC2=3.7 VC3=2.8 VC4=2.0 VC5=1.4

V31=1.8 V32=2.5 V33=2.4 V34=1.8 V35=2.3

V31=13.4 V32=20.5 V33=20.1 V34=14.3 V35=13.6

V31=0.4 V32=3.3 V33=3.7 V34=2.6 V35=3.3

FC1=2.2 FC2=2.3 FC3=2.1 FC4=2.4 FC5=2.0

FC1=14.6 FC2=17.4 FC3=18.2 FC4=19.7 FC5=18.6

FC1=2.4 FC2=2.8 FC3=3.5 FC4=4.0 FC5=3.5 Levene's

<1.0NS 2.36NS

5.01\*\* 3.66\*

9.02\*\* 12.68\*\*

3.16\* 7.50\*\*

18.42\*\* 14.71\*\*

3.42\*\* 21.87\*\*

<1.0NS 2.89\*

1.88NS 7.16\*\*

5.01\*\* 4.24\*\*

F§ Welch's F¶

resistance Mean† Standard

VC1=9.2 VC2=8.5 VC3=8.6 VC4=7.9 VC5=7.3

VC1=74.7 VC2=70.3 VC3=64.7 VC4=62.0 VC5=59.8

VC1=10.6 VC2= 9.5 VC3= 8.2 VC4= 6.6 VC5= 6.0

V31=9.9 V32=8.0 V33=8.5 V34=8.2 V35=7.9

V31=82.8 V32=62.7 V33=66.6 V34=63.9 V35=63.6

V31=5.6 V32=7.6 V33=9.8 V34=8.4 V35=7.2

FC1=7.5 FC2=8.4 FC3=8.4 FC4=8.8 FC5=9.4

FC1=55.6 FC2=64.5 FC3=69.0 FC4=72.2 FC5=75.8

FC1=7.0 FC2=7.9 FC3=8.6 FC4=9.5 FC5=10.6

Fig. 11. Seed [Fe] determined from genotypes grown on fully-limed acid soils can be used to predict seed [Fe] of those same genotypes when grown on partly-limed soils (A); similarly, seed [Fe] determined from genotypes grown on partly-limed soils can be used to predict seed [Fe] of those same genotypes when grown on fully-limed soils (B).

Fig. 11. Seed [Fe] determined from genotypes grown on fully-limed acid soils can be used to predict seed [Fe] of those same genotypes when grown on partly-limed soils (A); similarly, seed [Fe] determined from genotypes grown on partly-limed soils can be used to predict

seed [Fe] of those same genotypes when grown on fully-limed soils (B).


Importance of Seed [Fe] for

resistance to Fe deficiency.

Crookston,

Crookston,

‡ Missing data.

**4. Conclusions** 

[Fe], and planting seed Fe content.

ppm, which led to nearly identical results.

Classification Harvest seed

Environment variable (ug Fe g-1 seed) (ug Fe 1000-1

Improved Agronomic Performance and Efficient Genotype Selection 47

limed soil: linear, r2=0.860, F=24.5\*\* (Fig. 11, B). Differences in class sizes (5 vs 10 ppm) are related to the range of seed [Fe] values observed for the independent variable. Nonetheless, the same genotypes usually were included in the same classes whether based on 5 or 10

Presuming that seed [Fe] can be regarded as an integrated measure of resistance to Fe deficiency that is manifest at maturity, then Spehar's data provides additional evidence that individual genotypes have a seed [Fe] "threshold" that is genetically predetermined, yet seldom exceeded, and that seed [Fe] should supplement or replace VCS as a measure of

[Fe]

Ada, 2003 VC† -0.2464\* -0.2007NS -0.0254NS V3 -0.0984NS -0.0137NS 0.1932NS FE 0.4382\*\* -0.2505\* 0.3687\*\* FS 0.3454\*\* 0.1730NS 0.3646\*\*

2003 VC -0.2214NS -0.0554NS 0.0186NS V3 -0.3616\*\* -0.3838\*\* -0.1832NS FE 0.5698\*\* 0.4548\*\* 0.4402\*\* FC 0.1701NS 0.4038\*\* 0.2148NS

2004 VC -0.1282NS 0.0320NS -0.0099NS V3 -0.1158NS -0.0887NS -0.1048NS FE 0.2211 NS -0.0298NS 0.0482NS FC 0.0044NS -0.0950NS -0.0365NS

† VC, V3, FE, and FC are published visual chlorosis score, in-field visual chlorosis score, planting seed

Table 7. Correlations among genotypes, across four environments, ranked on the basis of three measures of resistance to Fe deficiency and genotypes ranked on the basis of four classification variables though to represent potential measures of resistance to Fe deficiency.

Conclusions from the research discussed in this chapter are: (1) soybean seed [Fe] and/or seed Fe content provide reliable and consistent measures of genetic differences in resistance to Fe deficiency; (2) seed [Fe] is tightly controlled genetically; (3) it is not likely that

Fisher, 2003 VC -0.1722NS -0.0659NS .‡ V3 -0.0999NS 0.0722NS . FE 0.3568\*\* 0.0975NS . FC 0.2468\* -0.0057NS .

Measures of resistance to Fe deficiency

Harvest Fe

content Fe removal

seeds) (ug Fe m-2)


† Mean of resistance measure for each level of classification.

‡ Standard deviation of resistance measure for each level of classification.

§ Test to assess homogeneity of error variances across levels of classification

¶ Test to assess the equality of classification means.

NS, \*, \*\* Not significant at 5% level of probability, significant at 5% level of probability, and significant at 1% level of probability.

Table 6. Classification of genotypes into 5 levels of selected plant traits thought to represent measures of resistance to Fe deficiency, defined as Fe accumulation (ug Fe 1000-1 seeds), harvest seed [Fe] (ug Fe g-1 seed), and Fe removal (ug Fe m-2).

between partially- and fully-limed conditions of r=0 .686 (P<0.001) using nearly all genotypes. However, some genotypes were missing seed [Fe]s under one treatment or another and 7 cultivars were not included in later calculations. Using 38 genotypes and 6 seed [Fe] classes (5 ppm), seed [Fe] on partly-limed soil could be predicted from seed [Fe] on fully-limed soil: linear, r2=0.883, F=30.2\*\* (Fig. 11, A). Using 38 genotypes and 6 seed [Fe] classes (10 ppm), seed [Fe] on fully-limed soil could be predicted from seed [Fe] on partlylimed soil: linear, r2=0.860, F=24.5\*\* (Fig. 11, B). Differences in class sizes (5 vs 10 ppm) are related to the range of seed [Fe] values observed for the independent variable. Nonetheless, the same genotypes usually were included in the same classes whether based on 5 or 10 ppm, which led to nearly identical results.

Presuming that seed [Fe] can be regarded as an integrated measure of resistance to Fe deficiency that is manifest at maturity, then Spehar's data provides additional evidence that individual genotypes have a seed [Fe] "threshold" that is genetically predetermined, yet seldom exceeded, and that seed [Fe] should supplement or replace VCS as a measure of resistance to Fe deficiency.


† VC, V3, FE, and FC are published visual chlorosis score, in-field visual chlorosis score, planting seed [Fe], and planting seed Fe content.

‡ Missing data.

46 Soybean – Genetics and Novel Techniques for Yield Enhancement

deviation‡

FS1=2.0 FS2=2.4 FS3=2.1 FS4=2.6 FS5=2.1

FS1=18.4 FS2=18.4 FS3=19.2 FS4=17.1 FS5=18.8

FS1=3.2 FS2=3.2 FS3=3.5 FS4=3.2 FS5=3.3

GC1=2.2 GC2=2.3 GC3=2.5 GC4=2.2 GC5=2.6

GC1=17.7 GC2=18.3 GC3=17.8 GC4=20.6 GC5=24.8

GC1=3.1 GC2=3.1 GC3=3.8 GC4=4.3 GC5=4.7

Levene's

1.59NS 5.63\*\*

<1.0NS <1.0NS

<1.0NS 1.49NS

<1.0NS <1.0NS

<1.0NS <2.45NS

1.77NS 1.65NS

F§ Welch's F¶

resistance Mean† Standard

FS1=7.7 FS2=8.1 FS3=8.6 FS4=9.6 FS5=10.1

FS1=67.6 FS2=64.9 FS3=68.0 FS4=66.3 FS5=71.8

FS1=8.4 FS2=8.0 FS3=8.7 FS4=8.3 FS5=10.4

GC1=8.2 GC2=8.0 GC3=9.0 GC4=8.7 GC5=9.9

GC1=66.3 GC2=64.6 GC3=69.3 GC4=80.6 GC5=80.0

GC1=8.4 GC2=8.0 GC3=9.1 GC4=10.9 GC5=11.4

\*\* Not significant at 5% level of probability, significant at 5% level of probability, and significant at

Table 6. Classification of genotypes into 5 levels of selected plant traits thought to represent measures of resistance to Fe deficiency, defined as Fe accumulation (ug Fe 1000-1 seeds),

between partially- and fully-limed conditions of r=0 .686 (P<0.001) using nearly all genotypes. However, some genotypes were missing seed [Fe]s under one treatment or another and 7 cultivars were not included in later calculations. Using 38 genotypes and 6 seed [Fe] classes (5 ppm), seed [Fe] on partly-limed soil could be predicted from seed [Fe] on fully-limed soil: linear, r2=0.883, F=30.2\*\* (Fig. 11, A). Using 38 genotypes and 6 seed [Fe] classes (10 ppm), seed [Fe] on fully-limed soil could be predicted from seed [Fe] on partly-

Classification variable

Planting seed Fe content (FS)

Relative chlorophyll concentration SPAD (GC)

NS, \*,

1% level of probability.

Measure of

ug Fe 1000-1 seeds

ug Fe g-1 seed

Fe removal (ug

Fe m-2)

ug Fe 1000-1 seeds

ug Fe g-1 seed

Fe removal (ug

† Mean of resistance measure for each level of classification.

‡ Standard deviation of resistance measure for each level of classification. § Test to assess homogeneity of error variances across levels of classification

harvest seed [Fe] (ug Fe g-1 seed), and Fe removal (ug Fe m-2).

Fe m-2)

¶ Test to assess the equality of classification means.

Table 7. Correlations among genotypes, across four environments, ranked on the basis of three measures of resistance to Fe deficiency and genotypes ranked on the basis of four classification variables though to represent potential measures of resistance to Fe deficiency.
