Crop Science
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Web of Science (9)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sinebo, W.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Sinebo, W.
Agricola
Right arrow Articles by Sinebo, W.
Related Collections
Right arrow Crop Growth and Development
Right arrow Other Crop Management
Crop Science 42:428-437 (2002)
© 2002 Crop Science Society of America

CROP ECOLOGY, PRODUCTION & MANAGEMENT

Yield Relationships of Barleys Grown in a Tropical Highland Environment

Woldeyesus Sinebo*

Holetta Agric. Res. Center, P.O. Box 2003, Addis Ababa, Ethiopia

* Corresponding author (iar{at}telecom.net.et or wsinebo{at}hotmail.com)


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
Grain yield in barley (Hordeum vulgare L.) can be improved by understanding the interrelationships among yield, yield components, vegetative growth, and growth durations. The objective of this study was to determine for barley genotypes adapted to low-input tropical highlands the extent to which grain yield was related to: (i) vegetative and mature heights, straw yield, and vegetative and grain-filling durations and (ii) yield components determined by correlation and ontogenetic path analysis. Data were obtained from 26 barley genotypes tested in factorial combinations of N (0 and 11.5 g m-2) and P (0 and 2 g m-2) in 1998 and 1999 at Holetta, Ethiopia. Grain yield was correlated positively with straw yield, vegetative and mature heights, and grain-filling duration. Grain yield was correlated positively with harvest index and correlated negatively with vegetative duration in the cooler season. Vegetative duration influenced grain yield negatively under low N and in the cooler season but positively under high N. Mature height influenced grain yield negatively under high N. Vegetative height influenced vegetative duration negatively. Spikes per square meter followed by kernels per spike largely determined grain yield. However, spikes per square meter had a strong negative effect on kernels per spike. Kernel weight had little effect on grain yield. Early shoot height association with yield and time to maturity may suggest an adaptive strategy for capturing the early flush of mineralized N and an escape mechanism from drought towards season end. Early shoot height can serve as an indirect selection criterion for high grain yield and early maturity for this gene pool grown in a tropical highland environment.

Abbreviations: BLUP, best linear unbiased predictor • GDD, growing degree days • GFD, grain-filling duration • G x Y x N x P, genotype x year x nitrogen x phosphorous • HED, vegetative duration • HI, harvest index • KPS, kernels per spike • KW, kernel weight • MHT, mature plant height • SPK, spikes per square meter • STR, straw yield • VHT, vegetative shoot height • YLD, grain yield


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
GRAIN YIELD IN BARLEY can be expressed as a function of spikes per square meter, kernels per spike, and kernel weight, which together are referred to as yield components. Yield components are formed during successive stages in the ontogeny of plants, and earlier formed components influence those that develop later (Hamid and Grafius, 1978). Previous studies, using ontogenetic path analysis, have shown that morpho-phenological traits such as phyllochron, leaf number, leaf area, stem diameter, and vegetative and grain-filling durations interact with yield components, and thereby in the end affect grain yield (Hamid and Grafius, 1978; García del Moral et al., 1991; Dofing, 1997). Grain yield in barley depends mainly on spikes per square meter and kernels per spike (Hamid and Grafius, 1978; Riggs et al., 1981; García del Moral et al., 1991; Dofing, 1997). The effect of kernel weight on grain yield is either negligible (García del Moral et al., 1991) or inconsistent (Dofing, 1997). The association of spikes per square meter with kernels per spike is negative (Rasmusson and Cannell, 1970; Hamid and Grafius, 1978; García del Moral et al., 1991; Balkema-Boomstra and Masterbroek, 1993) but with kernel weight is either positive (Rasmusson and Cannell, 1970; García del Moral et al., 1991) or negative (Hamid and Grafius, 1978; Balkema-Boomstra and Masterbroek, 1993). Kernels per spike and kernel weight are correlated negatively (Grafius, 1978; Hamid and Grafius, 1978; Balkema-Boomstra and Masterbroek, 1993).

Large dry matter production and its partitioning to the grain are traits of importance in cereal improvement. Elsewhere, much of cereal grain yield increases in the past have accrued from increase in harvest index (Donald and Hamblin, 1976; Riggs et al., 1981; Martiniello et al., 1987; Boukerrou and Rasmusson, 1990; Sinclair, 1998). With harvest index approaching its limit, future grain yield increases may result from increase in vegetative biomass (Riggs et al., 1981; Boukerrou and Rasmusson, 1990; Damisch and Wieberg, 1991; Schittenhelm et al., 1996; Okeno, 1999). In Ethiopia, straw is used as a livestock feed and therefore high straw yield without compromising grain yield is an advantage. Mature plant height and straw yield are correlated positively (Boukerrou and Rasmusson, 1990; Martinez and Foster, 1998), and therefore, taller height may increase grain yield indirectly when positive association between the yield traits exist (Boukerrou and Rasmusson, 1990). In many instances, however, reduction in plant height is associated with progress made in barley breeding for grain yield (Riggs et al., 1981; Martiniello et al., 1987; Boukerrou and Rasmusson, 1990; Schittenhelm et al., 1996; Okeno, 1999).

Early maturity is desirable in stress-prone areas, and is achieved either by the shortening of grain-filling period (Dofing, 1997) or vegetative period (Peltonen and Nissila, 1996), while maintaining high grain yield. Vegetative duration is correlated negatively with grain yield in growth limiting environments (Donald and Hamblin, 1976; van Oosterom and Ceccarelli, 1993) but is correlated either positively (Donald and Hamblin, 1976) or not at all with grain yield in favorable environments (van Oosterom and Ceccarelli, 1993). The effect of grain-filling duration on grain yield was usually not significant (Metzger et al., 1984; Dofing, 1997).

Ethiopian barley possesses favorable traits for performance under adverse environmental conditions such as low nitrogen and drought (Gróny, 2001). However, little is known on how yield-related traits interact in the process of yield formation for this gene pool in this and similar environments. Gebre et al. (1996) reported good seedling and vegetative vigor but low yield potential in Ethiopian barley landraces. Their claims for vigor appeared to be based on experiences from visual observation, as there was neither data nor information on how vigor was evaluated. In addition, whether and how the stated vigor apparent in this barley conferred ecological advantage and was correlated with economic traits was not reported.

Early shoot height is easy to evaluate, can offer an unbiased measure of vigor relative to subjective evaluation, and therefore can serve as a selection criterion if the association of genotypic differences for this trait with economic traits is established. The objective of this study was to determine for barley genotypes composed largely of landraces adapted to low-input tropical highlands the extent to which grain yield was related to: (i) vegetative and mature plant heights, straw yield, and vegetative and grain-filling durations and (ii) yield components determined by correlation and ontogenetic approach to path analysis.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
Data were obtained from an experiment consisting of 26 barley genotypes tested in 1998 and 1999 in a factorial combination of two levels each of N (0 or 11.5 g m-2) and P (0 or 2 g m-2). Each year, the experiment was conducted on red brown clay (a Eutric Nitosol, FAO classification; essentially a Typic Haplustult, USDA classification) at Holetta Research Center (9°03' N, 38°31' E, elevation 2400 m), 45 km west of Addis Ababa, Ethiopia. The environment is seasonally humid with long-term average annual rainfall of 1100 mm, 85% of which is received between the months of June and Sept., and mean maximum and minimum temperatures of 21 and 6°C, respectively.

The 26 genotypes were composed of 19 landrace lines (3302-14, 3371-03, 3371-18, 208038-90, 208040-02, 3441-19, 3441-83, 212959-62, 3432-76, 3432-98, 1829-32, 1829-40, 3381-04, 3381-10, 1622-05, Neff-98, 3304-11, 3369-19, and Baleme), five breeding lines (HB-544, HB-545, HB-278, HB-242, and HB-42), and two introductions (IBON 94/91 and Beka). HB-42 is an improved standard food barley cultivar released in 1984 while Beka is an introduction released in 1973 (Lakew et al., 1996). Baleme is a typical line from the farmers' local cultivar around Holetta. Twelve of the genotypes are irregulare (a spike form typical of some Ethiopian barleys; Reid, 1985), eight are six-row, and six are two-row type.

The experimental field was planted to wheat in 1996 but was a fallow in 1997. A split-plot design with fertility treatments as main plots and genotypes as subplots was used. Both the main plots and the subplots within the main plots were randomized in 1998. In 1999, however, the main plots were fixed and, therefore, the same arrangement as in 1998 was used, the goal being to increase the N and/or P deficiency stress further in the control plots of the respective nutrients. But the genotypes within the fertility treatments were again randomized in 1999. There were two replications in 1998 and three replications in 1999. The third replication in 1999 was also sown to a barley crop with 4.6 g m-2 N and 2 g m-2 P in 1998, was located next to the other two replications and was managed similar to the other replications in the preceding years. Each subplot was made up of two rows in 1998 and four rows of 2.5-m length in 1999. The rows were 20 cm apart. The adjacent subplots were separated by a blank row in both years. The trial was sown at a seed rate of 8 g m-2 on 20 June 1998 and 24 June 1999. All the experimental area was harvested each year.

Vegetative shoot heights, from the ground level to the tip of the shoot, were measured four times from five plants in each plot before the booting of any of the genotypes. The times of measurements coincided with 351 ± 10, 414 ± 10, 484 ± 6, and 552 ± 4 growing degree-days (GDD) after sowing for the first, second, third and fourth measurements, respectively, in the two years. In absolute day basis, the times were 33, 40, 48, and 55 d after sowing in 1998 and 40, 47, 53, and 61 d in 1999. The measurements were later by 2 d for the first and second, and earlier by 1 d for the last two measurements in 1999 than was required to conform on GDD basis with the measurements in 1998. Since average growth in height within the range of the four measurements was linear, and also results of separate analysis of vegetative height data for each of the measurements were similar, plot mean values from the four measurements were averaged to give only a single value per plot. The mean values obtained in this way correspond to measurements made at about 450 GDD (mean of the four times of measurements), avoid cluttering, and ease understanding. Mature plant height was measured from the ground level to the tip of the spike excluding the awns at about physiological maturity. Heading date was recorded when 50% of the spikes in a plot had fully extruded out. Heading in this environment occurs within about a day after anthesis, and therefore anthesis and heading dates were considered the same for practical reasons. Physiological maturity was recorded when 50% of the spikes in a plot had lost all green color. Time from sowing to phenological stages was converted from calendar days to GDD by means of the following formula:

where Tmax and Tmin are the maximum and minimum air temperatures (°C), respectively, and Tb is the base temperature (°C). A Tb of 5°C was used. These primary data were used to calculate grain-filling duration (time to maturity less time to heading) and phase index (the ratio of grain-filling duration to time to maturity). All the growth periods were presented in GDD unless otherwise specified.

At maturity, the total plot area was harvested. All stems in a plot were counted and allowed to dry in open air for several weeks to a constant dry weight. Sun-dried, above-ground biomass was recorded, and then the plants were threshed and grain yield was determined. Straw yield was determined as a difference of total above-ground biomass and grain yield. At harvest, 10 spikes from each plot were sampled and threshed separately to determine kernels per spike and kernel weight. Yields of these spikes were later added to the plot yield to determine the final plot yield. No attempt was made to separate stems with and without spikes (if any), and as a result, spikes per square meter actually refers to the stem number per square meter.

Analysis of variance was done on the data by means of the following model:

where T is the observation of the ith variety G in the lth year Y and kth fertilizer level F in the jth replication R within year l; µ is the general mean, e is the variation due to random error or the residual, and YF, GF, GY, GYF, and FR(Y) are the interactions. In the analysis, year, fertilizer, and YF interaction were considered fixed and all other effects were considered random. The fertilizer effect in Model [1] above was partitioned into components as follows:

where T is the observation of the ith variety G in the lth year Y of the uth nitrogen level N and vth phosphorous level P in the jth replication R within year l; µ is the general mean, e is the variation due to random error or the residual, and YN, YP, NP, YNP, GN, GP, GNP, GY, GYN, GYP, GYNP, and NPR(Y) are the interactions. In the analysis, year, N, P, and all possible interactions among these three factors were considered fixed, and all the remaining effects were considered random. Genotypes were considered random because inferences were to be made not about the particular genotypes included in this study but about a wider range of germplasm handled by the national barley breeding program at Holetta, Ethiopia. Littell et al. (1996) argued the conceptual inconsistency of treating random effects, genotype effect in this case, as fixed for obtaining means by PROC GLM, and emphasized the importance of prediction for the nonestimable functions by PROC MIXED in SAS (SAS Institute, 1990). They further asserted that best linear unbiased predictors (BLUPs) from PROC MIXED would give, in this instance, environment-specific genotype trait predictors using information from the entire year-fertilizer combinations. For these reasons, genotype mean trait genotype x year x nitrogen x phosphorous (G x Y x N x P) interaction BLUPs were given instead of genotype trait means. Significance levels for the fixed and random effects were obtained by PROC MIXED and PROC GLM, respectively. Pair-wise comparison of mean genotype trait BLUPs could be made by means of the ESTIMATE statement in PROC MIXED but could not conveniently be presented here. As a result, only the standard errors of trait BLUPs were given.

Data within each of the eight environments (year-fertilizer combination) were standardized by removing the average effect of each environment and the grand mean from the genotype trait G x Y x N x P interaction BLUPs and dividing the resulting value by within-environment standard deviation. Pearson product moment correlation coefficient (r) among the traits for standardized genotype trait G x Y x N x P interaction BLUPs was calculated across all the environments (n = 208). Nonetheless, trait associations were also explored for each of the eight environments (n = 26), for each year, and N and P levels (n = 104) when such analyses were deemed necessary for better understanding of patterns and extents of associations among the traits under different circumstances. Since vegetative duration appeared to influence other associations, and because of the need to making inference for genotypes of different maturity classes or similar areas of differing growing-season length, correlation analysis was repeated among the traits with vegetative duration as a partial variable.

Path coefficients are standardized partial regression coefficients (Dewy and Lu, 1959), and path analysis on the data was done as reported in Dofing (1997) using the REG procedure of SAS (SAS Institute, 1990) with the stepwise regression method. For each of the dependent variables, independent variables were sequentially added and at each step only the independent variables significant at the probability level of 0.05 were retained in the model. An ontogenetic approach as given, for instance, in Hamid and Grafius (1978), García del Moral et al. (1991), and Dofing (1997), whereby traits formed earlier in the ontogeny of the plant determine those traits formed later, but not vice versa, were followed. Two path models (secondary traits and yield component models) were used to describe the effects of yield-related traits on grain yield formation. The secondary traits model describes grain yield as a function of vegetative growth (vegetative shoot height, mature plant height, spikes per square meter, and straw yield) and phenological (vegetative and grain-filling durations) traits. In this model, path coefficients were in turn calculated for each of the remaining traits assuming that each trait is an outcome of earlier formed traits in the ontogeny of the plant. Accordingly, the remaining paths used in the model were spikes per square meter, vegetative shoot height, mature plant height, vegetative duration, grain-filling duration -> straw yield; vegetative shoot height, spikes per square meter, vegetative duration -> mature plant height; vegetative shoot height, spikes per square meter, vegetative duration -> grain-filling duration; and vegetative shoot height, spikes per square meter -> vegetative duration. Vegetative shoot height and spikes per square meter were assumed to have a mutual association and therefore, joined by a double arrow. Spikes per square meter being a yield component was also included in this model because actually it was stem number that was counted, and stem number is determined relatively early in the ontogeny of the plant, for instance, before mature plant height, grain-filling duration, and straw yield. The other path model assumed grain yield as a function of yield components (spikes per square meter, kernels per spike, and kernel weight) only and had the direct effect of the yield components on grain yield calculated. In addition, path coefficients of the independent variables on kernels per spike and kernel weight were calculated assuming spikes per square meter, vegetative shoot height, mature plant height, vegetative duration, grain-filling duration, straw yield -> kernels per spike, and spikes per square meter, vegetative shoot height, mature plant height, vegetative duration, grain-filling duration, straw yield, kernels per spike -> kernel weight. The SAS statistical package version 6.12 was used for all data analyses (SAS Institute, 1990).


   RESULTS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSION
 REFERENCES
 
Average temperatures were consistently higher in 1998 than in 1999 (Fig. 1) . This led to the accumulation of more heat units early in 1998 than in 1999, resulting in significantly fewer number of days required to reach time to heading (83 vs. 91 d) and time to maturity (116 vs. 132 d) in 1998 than in 1999. The longer crop cycle in 1999 probably exposed the late maturing genotypes to end-of-season moisture stress, resulting in significant genotype x year interaction for most of the traits (Table 1). This, as will be seen later, led to differing patterns of some associations among the traits in the two years. Genotype effect was significant for all the traits (Table 1). Genotype trait G x Y x N x P interaction BLUPs closely matched the observed G x Y x N x P interaction trait means with r-values for all the traits ranging from 0.95 to 0.99 (n = 208). Mean genotype grain yield BLUPs ranged from 107 to 268 g m-2, and mean time-to-maturity BLUPs ranged from 106 d (965 GDD) to 137 d (1219 GDD) giving rise to a difference of 31 d (Table 2).



View larger version (61K):
[in this window]
[in a new window]
  		Fig. 1. Decadal rainfall and minimum, maximum, and mean air temperatures for Holetta Research Center in 1998 and 1999.

 

View this table:
[in this window]
[in a new window]
  		Table 1. Significance of variances for grain yield (YLD), straw yield (STR), spikes per square meter (SPK), kernels per spike (KPS), kernel weight (KW), harvest index (HI), vegetative shoot height (VHT), mature plant height (MHT), vegetative duration (HED), grain-filling duration (GFD) and phase index (PHI) in 26 barley genotypes tested at four fertility (N and P) levels in 1998 and 1999 at Holetta, Ethiopia.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Genotype mean trait genotype x year x N x P interaction best linear unbiased predictors for agronomic traits in barley tested at four N and P levels in 1998 and 1999 at Holetta, Ethiopia.

 
Exploratory analysis of the data showed that patterns of trait association among genotype means for the fertility levels within each year were similar but those for the same fertility levels in different years were substantially different. As a result, the Pearson correlation statistic among the traits was presented for the whole data set (n = 208, Table 3) and for each year separately (n = 104, Table 4). Overall (Table 3) and each year (Table 4), grain yield was consistently correlated positively with straw yield, spikes per square meter, kernels per spike, vegetative and mature heights, and grain-filling duration. Grain yield was not correlated with kernel weight (Tables 3 and 4). Grain yield was correlated most strongly with spikes per square meter followed by straw yield and vegetative shoot height in 1998 (Table 4) and across the two years (Table 3). Grain yield was correlated most strongly with vegetative shoot height in 1999 (Table 4). Vegetative shoot height was correlated negatively with vegetative duration but correlated positively with straw yield, spikes per square meter, harvest index, mature plant height, and grain-filling duration (Tables 3 and 4). When partial correlation was analyzed with vegetative duration as a partial variable, grain yield was most strongly correlated with straw yield, followed by spikes per square meter and again vegetative shoot height (Table 5).


View this table:
[in this window]
[in a new window]
  		Table 3. Correlation coefficients among standardized trait genotype x year x nitrogen x phosphorous interaction best linear unbiased predictors for agronomic traits in barley tested at four N and P levels in 1998 and 1999 at Holetta, Ethiopia.**

 

View this table:
[in this window]
[in a new window]
  		Table 4. Correlation coefficients among standardized trait genotype x year x nitrogen x phosphorous interaction best linear unbiased predictors for agronomic traits in barley tested at four N and P levels in 1998 (lower half) and 1999 (upper half) at Holetta, Ethiopia.

 

View this table:
[in this window]
[in a new window]
  		Table 5. Correlation coefficients partialled for vegetative duration among standardized trait genotype x year x nitrogen x phosphorous interaction best linear unbiased predictors for agronomic traits in barley tested at four N and P levels in 1998 and 1999 at Holetta, Ethiopia.

 
Grain yield was correlated significantly negatively with vegetative duration, but correlated significantly positively with harvest index in 1999; grain yield association with each of vegetative duration and harvest index was not significant in 1998 (Table 4). Generally, separate analysis for each of the N or P levels did not alter the sign and significance of correlation of grain yield with other traits (data not shown) from that presented in Table 3. Grain yield was correlated most strongly with spikes per square meter (r = 0.74), followed by straw yield (r = 0.66) and vegetative shoot height (r = 0.61) under N deficiency. These three traits were nearly equally well correlated with grain yield when N was applied with r-values of 0.65 for vegetative shoot height, 0.64 for spikes per square meter, and 0.63 for straw yield. Grain yield association with other traits in each of the two P levels was similar to that presented in Table 3.

Path analysis using the secondary traits model for the whole data set revealed the importance of large straw dry matter accumulation for high grain yield production (Fig. 2) . The path coefficient for the effect of vegetative height on grain yield was positive and significant but less than half of that for straw yield on grain yield (Fig. 2). Grain yield was correlated most strongly with vegetative height (r = 0.74 with vegetative height, 0.71 with spikes per square meter, and 0.65 with straw yield) when the two earliest heading lines (3371-03 and 3371-18) were dropped from the analysis. Also, the direct effect of vegetative height on grain yield was nearly as high as that of straw yield with coefficients of 0.48 and 0.50 when path analysis was made excluding the above two genotypes. The direct effect of vegetative shoot height was negative on growth durations and positive on mature height (Fig. 2). Grain-filling duration had a significant positive effect on grain yield but mature plant height had a significant negative effect (Fig. 2).



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 2. Path coefficients for grain yield, vegetative growth, and growth durations in 26 barley genotypes tested at four N and P levels in 1998 and 1999 at Holetta, Ethiopia. YLD, grain yield; STR, straw yield; SPK, spikes per square meter; VHT, vegetative shoot height; MHT, mature plant height; HED, vegetative duration; GFD, grain-filling duration. Note: VHT, GFD -> STR; SPK, HED -> YLD were not significant.

 
Year-to-year variation on the effects of some traits on grain yield in the secondary traits model was apparent (data not shown). Again each year straw yield had a highly significant direct influence on grain yield. Spikes per square meter had a significant positive influence on grain yield in 1998, whereas vegetative duration had a significant negative influence in 1999. The direct effects of other traits on grain yield in individual years were not significant. Nitrogen also affected the pattern and significance of the direct influences of some traits on grain yield in this path model (Fig. 3 and 4) . Without fertilizer N, straw yield and spikes per square meter influenced grain yield positively, whereas vegetative duration affected negatively (Fig. 3). With N application, vegetative shoot height, straw yield, and vegetative and grain-filling durations affected grain yield positively, whereas mature plant height affected negatively (Fig. 4). In general, in this path model, the negative effect of vegetative duration on grain yield was associated with low N and end-of-season moisture stress environments. Likewise, the positive effect of spikes per square meter on grain yield in the same model was associated with environments in which plant size was small as a result of rapid development or N stress implying negative spike density x plant size interaction for grain yield. Separate path analysis for the two P levels (data not shown) gave path coefficients similar in sign and significance to the overall results given in Fig. 2. The only exceptions are that the direct effect of mature plant height on grain yield with P application, and the direct effects of vegetative height and spikes per square meter on grain-filling duration without P were not significant.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 3. Path coefficients for grain yield, vegetative growth, and growth durations in 26 barley genotypes tested without N application at two P levels in 1998 and 1999 at Holetta, Ethiopia. YLD, grain yield; STR, straw yield; SPK, spikes per square meter; VHT, vegetative shoot height; MHT, mature plant height; HED, vegetative duration; GFD, grain-filling duration. Note: VHT, SPK -> GFD; VHT, GFD -> STR; VHT, GFD, MHT -> YLD were not significant.

 


View larger version (22K):
[in this window]
[in a new window]
  		Fig. 4. Path coefficients for grain yield, vegetative growth, and growth durations in 26 barley genotypes tested with N application at two P levels in 1998 and 1999 at Holetta, Ethiopia. YLD, grain yield; STR, straw yield; SPK, spikes per square meter; VHT, vegetative shoot height; MHT, mature plant height; HED, vegetative duration; GFD, grain-filling duration. Note: SPK -> GFD; VHT, HED -> STR; SPK -> YLD were not significant.

 
In the yield components model, spikes per square meter followed by kernels per spike exerted the greatest effect on grain yield (Fig. 5) . The effect of kernel weight on grain yield was small, nonetheless positive and significant (Fig. 5). The order of importance of the effects of yield components on grain yield was not altered in the two years, N, or P levels (data not shown). Nevertheless, the path coefficient for the effect of spikes per square meter on grain yield was larger than that of the effect of kernels per spike on grain yield by 47% without fertilizer N and by 8% only with N application. The effects of spikes per square meter on both kernels per spike and kernel weight, and the effect of kernels per spike on kernel weight, were negative (Fig. 5).



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 5. Path coefficients for grain yield as affected by yield components, and for KW and KPS as affected by earlier formed traits in 26 barley genotypes tested at four N and P levels in 1998 and 1999 at Holetta, Ethiopia. YLD, grain yield; SPK, spikes per square meter; KPS, kernels per spike; KW, kernel weight; STR, straw yield; VHT, vegetative shoot height; MHT, mature plant height; HED, vegetative duration; GFD, grain-filling duration. Note: HED, GFD -> KPS were not significant.

 

   DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSION
 REFERENCES
 
The use of path analysis to determine the relative contributions of yield components and morpho-phenological traits to grain yield in barley is not new (Hamid and Grafius, 1978; García del Moral et al., 1991; Dofing, 1997). The process of yield formation can, however, vary with the type of environment (Donald and Hamblin, 1976; van Oosterom and Ceccarelli, 1993) and, therefore, the information presented in this study is vital for this (tropical highlands) and similar areas, even though most of the traits have already been studied. From this study, we understood that early shoot height, large spikes per square meter, large straw yield, large plant size, reasonably shorter vegetative duration, and longer grain-filling duration are the traits associated with good performance of low-input tropical highland barley landraces in the environment of their adaptation. In addition, vegetative shoot height data, the earliest trait measured in the ontogeny of the plant, allowed us to predict the other traits, which offered the possibility of screening a large number of materials indirectly for yield potential and early maturity.

Vegetative shoot height was the best predictor of grain yield performance in the year of end-of-season moisture stress (Table 4) and under high N supply. Also, of all the traits, vegetative shoot height had the largest direct effect on grain yield when N was applied (Fig. 4). These may imply that early shoot vigor is an adaptive strategy of the landraces for acquiring N during its peak availability and an escape mechanism from end-of-season moisture stress. Both Ethiopian barley and the barley ecology in the country are diverse. It is probable that many landraces have survived generations of natural and conscious selection because of traits that bestowed survival strategies under limiting soil nutrient supply and drought. In the area, fertilizer N application for barley production is not common, and therefore the crop relies on soil mineral N, which is relatively abundant at the beginning of the season. The arrival of rainfall and its distribution within the prime period (mid-June to mid-September) is predictable but rainfall at the end of the season is less well so. Adaptation to such an environment requires vigorous crop establishment and rapid early vegetative growth for establishment of the vegetative structures necessary for adequate resource capture and accumulation at times of relative abundance. On top of this, the crop should be able to switch from vegetative to grain-filling phase reasonably early to escape the onset of end-of-season moisture stress. It therefore appears that the high-yielding landrace lines in this study have combined these adaptive attributes.

Overall, only the direct effect of straw yield on grain yield exceeded the direct effect of vegetative height on grain yield in the secondary traits model (Fig. 2). This was not surprising given grain yield and straw yield determinations are closer in time than grain yield and vegetative height determinations. However, correlation of grain yield with vegetative height was the highest, and also the direct effect of vegetative height on grain yield was comparable to the effect of straw yield on grain yield when the two earliest maturing genotypes were dropped from the analyses. The earliness of these genotypes must have exceeded the critical limit required for adequate production of vegetative structures to support optimum grain yield. It might also imply that early shoot height would predict the relative yielding ability of genotypes better if the range of vegetative duration for the genotypes were not very wide.

Trait associations between the years were noticeably different because of differences in temperature. The temperature was cooler and time to maturity was longer in 1999 than in 1998. Since rainfall in this area starts to taper off by mid-September and virtually comes to a halt in October (Fig. 1), the late maturing genotypes faced moisture stress during grain filling and performed relatively poorly, resulting in a negative association of grain yield with vegetative duration in 1999. In other studies, the effect of vegetative duration on grain yield was also negative under stress (Donald and Hamblin, 1976; van Oosterom and Ceccarelli, 1993) and positive (Donald and Hamblin, 1976) or not significant (van Oosterom and Ceccarelli, 1993) under non-stress conditions. Harvest index was correlated positively with grain yield in the cooler year but not in the warmer year. Since crops normally accumulate larger biomass under cooler conditions, perhaps dry matter partitioning to the grain for high grain yield production, particularly in an environment with end-of-season moisture stress, was critical. Warmer temperatures, nevertheless, enhance rapid development and shorten crop cycle at the expense of large dry matter accumulation. Under such a condition, dry matter partitioning to the grain for high grain yield production was not of major importance. In sum total, shorter vegetative duration and higher harvest index for higher grain yield were important in the cooler season but not in the warmer season. The sensitivity of harvest index to environmental influences is a fact well appreciated (Donald and Hamblin, 1976).

As with year, the importance of the direct influences of spikes per square meter and vegetative duration on grain yield varied with N level. The importance of the effect of spikes per square meter on grain yield without N application was probably largely associated with stand loss resulting from N stress. Under N stress, spike density of many of the genotypes was less than the number of seedlings counted after emergence (data not shown). Almost all the genotypes with a positive change in shoot number under N stress were either the highest in grain yield or the earliest in maturity. Selection for high stand survival under N stress is therefore likely to improve grain yield or result in early maturity. The negative spike density x plant size interaction for grain yield indicates the need for targeting high spike density for short-stature genotypes and/or under management conditions that does not increase plant size. In this study, N deficiency delayed time to heading by 7 d and, as with cooler temperatures, resulted in the negative effect of vegetative duration on grain yield, implying that early maturing genotypes are appropriate for low input stressful environments. Since, however, the direct effect of vegetative duration on grain yield under N sufficiency was positive, potentially high-yielding, late-maturing genotypes could be grown under high N management without yield penalty. Late maturity, nevertheless, was associated with large plant size, which had a negative effect on grain yield under high N, prompting the need to consider varietal stature for barley production under high N management.

It appears that high straw yield is essential to high grain yield production. In Ethiopia, since barley grain is used as food and straw as livestock feed, a positive association of grain and straw yields facilitates the concurrent improvement of both traits. Despite the negative direct effect of mature plant height on grain yield (Fig. 2 and 4), both traits were correlated positively (Tables 3 and 4). Reduction in plant height and increase in harvest index typify an increase in cereal yields in high input production systems. Shorter heights of modern varieties allows high N application without the lodging problem of taller varieties (Evans, 1993). Since, however, N deficiency in the barley production systems of Ethiopia is ubiquitous, these potentially tall landraces are short, precluding problems with lodging.

Although the direct effects of each of the three yield components on grain yield were significant, the contribution of kernel weight to grain yield was minimal. Kernel weight, therefore, is an unlikely candidate for grain yield improvement in this gene pool. Spikes per square meter was more important in determining grain yield than kernels per spike, both because of its larger path coefficient on grain yield and its positive role on straw production, which is an economic byproduct and a requirement for high grain yield. The strong negative effect of spikes per square meter on kernels per spike, which is a result of compensation between the components, may preclude selection of lines that combine both traits. Rather, positive association of spikes per square meter and kernel weight might enable the selection of lines such as 1829-32 and 1829-40 (Table 2) that combine these two traits.

In the current study, the positive effect of grain-filling duration on grain yield that we found is at variance with other studies (Metzger et al., 1984; Dofing, 1997) and may be attributed to the relatively shorter duration of grain filling in this environment which probably limits assimilation more than in these other environments. According to Metzger et al. (1984), phase index ranged from 0.36 to 0.43 for long-duration and 0.31 to 0.36 for their short-duration genotypes. From the data by Dofing (1997), a phase index of 0.47 can be calculated. In this study, phase index averaged 0.29.

Despite the high correlation of some traits, there was no significant influence on grain yield in a particular season and/or N level. Since path coefficients indicate the relative importance of traits, perhaps only those traits that limited grain yield the most under specific stresses successfully masked the direct effects of other traits. In the path analysis, spikes per square meter was considered an independent variable that affected grain yield in the secondary traits model as well as in the yield components model. Overall, the effect of spikes per square meter on grain yield in the secondary traits model was not significant (Fig. 2) because of the presence of straw yield in the model, and therefore it did not affect the relative effects of the other variables on grain yield. Nonetheless, in this path model, the direct effect of spikes per square meter on grain yield was significant in environments where straw production was constrained by rapid development and/or N stress.


   CONCLUSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
REFERENCES
 
Barley grain yield in Ethiopia and similar areas may be increased by selecting genotypes for early shoot height, high spikes per square meter, high straw yield, and longer grain-filling duration. Selection for shorter vegetative duration and higher harvest index would ensure yield stability over years, but grain yield may decline when shortening exceeds the critical limit required for adequate production of vegetative structures necessary to support high yields. Survivability and early maturity under low N; early shoot vigor, longer growth durations, and shorter mature plant height under high N; and large straw yield under both conditions were identified as traits having direct positive influence on grain yield. Early shoot height is a good predictor of grain and straw yields, spikes per square meter, harvest index, and vegetative and grain-filling durations and can be used for screening large number of materials early in the variety development process.

Nature has programmed early shoot vigor for better resource capture at times of availability and early maturity as a escape mechanism from end-of-season stress into these high-yielding barley landraces. Yield improvement in such a highly adapted genetic background by introducing exotic germplasm may be possible but will require the maintenance of early shoot vigor and early maturity—a challenging task for a relatively new breeding program


   ACKNOWLEDGMENTS
 
My deep appreciation is due to Ao. Prof. Dr. R. Gretzmacher, my academic supervisor; for, inter alia, providing facilities and for support in my obtaining an Austrian Development Cooperation scholarship both of which were vital in accomplishing this work.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
The data herein are from Holetta Agric. Res. Center exp. no. BA/Ag-HO(97-1), and partially fulfills the requirements for a Ph.D. degree at the Univ. of Agricultural Sciences, Vienna. Scholarship and partial research support by Austrian Development Cooperation.

Received for publication May 16, 2001.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Web of Science (9)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sinebo, W.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Sinebo, W.
Agricola
Right arrow Articles by Sinebo, W.
Related Collections
Right arrow Crop Growth and Development
Right arrow Other Crop Management

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome