Genetic gain of maize (Zea mays L.) varieties in Ethiopia over 42 years (1973 - 2015)

Currently under production, thirty-eight Ethiopian maize varieties released majorly for three agro-climatic zones over the past thirty-nine, twenty-nine and twelve years for the high altitudes, mid–altitudes and low altitudes, respectively, were conducted at three different research center’s field trials, using randomized complete block design with three replications in 2015 main cropping season to estimate the genetic gains made on yield and yield related traits. The regression analysis indicated average annual and annual relative genetic gains of 62.3 (0.19%), 59.0 (0.57%) and –2.64 (–0.16%) in kg ha–1 yr–1 for grain yields, respectively, at Ambo Plant Protection Research Center (APPRC), Bako National Maize Research Center (BNMRC) and Melkassa Agricultural Research Center (MARC). Correlational analysis on the field studied traits indicated positively significant associations of grain yields with grain filling rate, ear length, number of kernels per row, number of ears per plant, biomass production rate, biomass yield and harvest index; also, negatively significant associations were shown for days to anthesis and days to silking at APPRC. Grain yield showed positively significant associations with ear length, plant height, grain filling rate, thousand kernel weight, biomass production rate and harvest index at BNMRC, while those only with harvest index were shown at MARC. Relatively considerable genetic gains and inconsiderable genetic reductions due to grain yields, grain yield related traits and grain yields associations with the other studied maize breeding traits had been observed across the released maize varieties from the three agro–ecological zones of Ethiopia. 
 
 Key words: Annual genetic gains, annual relative genetic gains, correlational analysis, highland maize, lowland maize, mid–altitude maize, regression analysis.


INTRODUCTION
Maize (Zea mays L.) arrived in Africa through various introductions as long ago as 500 years (McCann, 2005).
Since its introduction to Africa, maize has thus become the number one crop in Africa both in cultivated area and *Corresponding author. E-mail: kebedeyomichael@gmail.com.
Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License total grain production (FAOSTAT, 2015). It is believed that maize was first introduced to Ethiopia in the 16 th or 17 th century (Haffnagel, 1961). Since its introduction, it has gained importance as a food and feed crop in the country, which has remained being considered as one of the priority crops in an effort to meet the food demand of the country's increasing population.
In Ethiopia, maize grows from moisture stress areas to high rainfall areas and from lowlands to the highlands (Kelemu and Mamo, 2002). Amongst the cultivated major cereal crops of Ethiopia, maize ranks second to teff [Eragrostis tef (Zucc.)] in area and first in production. Maize remained to be the largest and most productive crop, leading the major cereal crops in Ethiopia since the mid-1990s in terms of both crop yield and production. Over the last decades, maize coverage has reached 2.4 million ha from being a mere garden crop to an economical cereal crop in Ethiopia. The trends in national maize productivity levels show a small but consistent increase from about 1.5 t ha -1 in the early 1990s to 2.3 t ha -1 in the late 2000s (CSA, 2015).
Maize research in Ethiopia started in the early 1950s and passed through distinct stages of research and development (Kebede et al., 1993). Since 1973, the maize research program of Ethiopia has been receiving International Maize and Wheat Improvement Center (CIMMYT) germplasms (Mosisa et al., 2001). In the late 1990s, the breeders began to develop inbred lines from different source materials using the pedigree breeding method. Currently, the maize breeding program introduces fixed or intermediate (semi-processed) inbred lines from international research institutes such as the CIMMYT and International Institute of Tropical Agriculture (IITA) (Legesse et al., 2012).
In Ethiopia, right from the beginning of the comprehensive maize breeding program in the early 1980s, the maize breeding program has passed through many distinct stages of research and development (Degene and Habtamu, 1993). Progressively in the 1990s, the multidisciplinary approach was consolidated. Currently, the strategic focus of Ethiopia's public sector maize breeding programs are to develop improved maize varieties and hybrids for three specific types of agroecological zones (low, medium and high altitude maize growing areas of the country); four types of varieties (extra-early, early, intermediate, and late maturing varieties); and four types of attributes (yield improvement, drought tolerance, earliness and disease resistance) (Ethiopian Agricultural Research Organization (EARO), 2000; Ministry of Agriculture and Rural Development (MoARD), 2004-2016Dawit et al., 2010).
In the last four decades, more than 40 improved varieties of maize including hybrids and Open Pollinated Varieties (OPVs) have been developed and released by the Ethiopian Institute of Agricultural Research (EIAR) in collaboration with the CIMMYT (Zeng et al., 2014). Despite maize suited diverse agro-climate subsists, and strategic maize breeding efforts were made, the production of maize in the country remained low; with the estimated national average yield of 3.25 t ha -1 (CSA, 2014), which is low in light of the potential productivity of the world average of 5.64 t ha -1 with a productivity record of 10.73 t ha -1 by the US for the year 2014/15 (United States Agency for International Development-Foreign Agricultural Service (USDA-FAS), 2016).
Quantifying breeding achievements in yield and associated traits and understanding the crucial characteristics of the crop associated with the genetic gains achieved through breeding is an essential step for improving the current knowledge of yield limiting factors and the design for the future breeding strategies. Historical series of maize varieties have been deployed and used to assess the genetic gains achieved during a period of time through breeding in several countries: in Argentina by Echarte et al. (2013), USA by Russell (1985), Duvick (2005a), Jason et al. (2013) and Chen et al. (2016), and Africa by Badu-Apraku et al. (2013, 2014 and Omolaran et al. (2014). However, in Ethiopia there are few and scanty information that exist on the genetic gains in grain yield and other agronomic traits during the maize breeding eras for the released and registered highland, mid-altitude and lowland maize varieties in Ethiopia.
Periodic evaluation of genetic improvement of improved maize varieties will help identify traits of potential value for future breeding enhancement and target them for higher productivity for the majority of subsistence farmers engaged in the production of the maize crop. With these, the objectives of this study were to estimate the genetic gain made over decades and to identify changes in morpho-physiological characters associated with genetic improvements in grain yield potential of maize varieties.

Experimental design and field management
All the experiments were laid out in a Randomized Complete Block Design ( Planting for the three sets of experiments were undertaken on June 05 and 08, 2015 respectively at BNMRC for the mid-altitude maize varieties and at APPRC for the highland maize varieties; while on July 09, 2015 for the lowland maize varieties at MARC by hand sowings two seeds per hill, which were later thinned to one plant per hill. The same field managements were used for the three sets of experiments, on which pre-emergence herbicides (Atrazine at the rate of 4 L ha -1 for broad leaved weeds and Primagram at the rate of 4 L ha -1 for grass weeds), nitrogen fertilizer in the form of Urea and phosphorus fertilizer in the form of Diammonium Phosphate were applied as per the specific recommendations for the areas. Similarly, hand weeding was done twice at 25 and 45 days after emergence; and weed slashing was done once at the flowering stages.

Statistical analysis
All measured parameter's field data were subjected to an Analysis of Variance (ANOVA) using SAS statistical software version 9.00 (SAS, 2002) to estimate the prevalent variation among the test varieties. Treatments and replications were the class variables, while the response variables were the traits considered for the data collected. The ANOVA Model: m + Where: Y ij = Observed value of genotype i in block j m = Grand mean of the experiment G i = Effect of genotype i B j = Effect of block j e ij = Random error effect of genotype i in block j The test of mean separation was employed depending on the significance of ANOVA. Mean separation was undertaken using Duncan's Multiple Range Test (DMRT) at the 5% level of significance. Correlation among all the traits was calculated using the means of each variety applying the PROC CORR procedure in SAS. Linear regression analysis was used to calculate the genetic gain for each trait considered in the study. The breeding effects were estimated as a genetic gain for grain yield and associated agronomic traits in maize improvement by regressing mean of each character for each variety against the year of release of the variety using the PROC REG procedure in SAS. The relative gain achieved over the year of release periods for each traits under consideration were determined as a ratio of genetic gain to the corresponding mean value of old variety and was expressed as a percentage using software program Microsoft Office (Excel 2010).
The annual rate of gain was calculated as:

( )
Where: Cov = Covariance Var = Variance X = the year of variety release Y = the mean value of each character for each variety The correlation between traits using means of each variety was calculated as: Where: rxy = Correlation coefficient between X and Y Cov (X, Y) = Covariance between X and Y Var (X) = Variance of X Var (Y) = Variance of Y

Analysis of variance of grain yield and other agronomic traits of Ethiopian highland, mid-altitude and lowland maize varieties
Analysis of variance for grain yield traits indicated significant (P≤ 0.05 and P≤ 0.01) differences for number of ears per plant, ear length, number of kernels per row, grain yield, biomass yield, biomass production rate and harvest index. In contrast, non-significant (P> 0.05) differences were observed among the seven highland maize varieties for ear diameter, number of kernel rows per ear and thousand kernel weight at APPRC, while highly significant (P≤ 0.01) differences were observed for all studied yield and productivity traits of the twenty midaltitude maize varieties at BNMRC ( Table 2). Results of the analysis of variance for the studied grain yield traits of the eleven lowland maize varieties at MARC indicated highly significant (P≤ 0.01) differences among varieties for the number of ears per plant, ear length, grain yield, biomass yield, biomass production rate and harvest index while significant (P≤ 0.05) differences among varieties were shown in number of kernels per row. Ear diameter and number of kernel rows per ear showed nonsignificant (P> 0.05) difference amongst the studied grain yield traits ( Table 2). The analysis of variance for the growth and phenological traits of the seven highland maize varieties studied at APPRC showed highly significant (P≤ 0.01) differences that were observed for days to anthesis, days to silking, grain filling rate and ear height; whereas nonsignificant (P> 0.05) differences were observed for days to maturity, grain filling period and plant height. Further, the results of the analysis of variance for all growth and phenological traits of the twenty mid-altitude maize varieties and the eleven lowland maize varieties studied, respectively at BNMRC and MARC, showed highly significant (P≤ 0.01) differences (Table 3).

Genetic gains in grain yield and other agronomic traits of Ethiopian highland, mid-altitude and lowland maize varieties
Regression of the mean values of the highland maize varieties correspondingly with the year of releases over the past 39 years demonstrated positive and nonsignificant (P> 0.05) annual predictive and average relative genetic gain of 62.26 (1.24%) kg ha -1 yr -1 for grain yield and 76.37 (0.37%) kg ha -1 yr -1 for biomass yield at APPRC ( Figure 1A and B).
The regression of the mean values of the mid-altitude maize varieties correspondingly with the year of releases over the past 29 years demonstrated positively nonsignificant (P> 0.05) annual predictive and average relative genetic gain of 58.97 (0.78%) kg ha -1 yr -1 for grain yield and 95.63 (0.45%) kg ha -1 yr -1 for biomass yield at BNMRC (Figure 2A and B).
Regression of the mean values of the lowland maize varieties correspondingly with the year of releases over the past 12 years demonstrated positive and nonsignificant (P>0.05) annual predictive and average relative genetic gain of 32.64 (0.57%) kg ha -1 yr -1 for biomass yield. Differently, demonstrated negative and non-significant predictive average annual rate of decrease was shown by -2.64 (-0.16 %) kg ha -1 yr -1 for grain yield at MARC (Figure 3A and B).

Correlation of grain yield and other agronomic traits of Ethiopian highland, mid-altitude and lowland maize varieties
Correlation coefficients for the grain yield among the seven highland maize varieties released over the past 39 years had shown a positive and highly significant (P≤0.01) associations with grain filling rate (r= 0.99**), ear length (r= 0.89**), number of kernels per row (r= 0.94**), biomass production rate (r= 0.90**) and harvest index (r= 0.91**); while grain yield was positive and significantly (P≤0.05) associated with number of ears per plant (r= 0.85*) and biomass yield (r= 0.82*). Differently, highly significant (P≤ 0.01) and negative association for days to silking (r= -0.91**); and significant (P≤0.05) and negative association for days to anthesis (r= -0.82*) were shown with the grain yield at APPRC (Table 5)

BY Predicted BY Linear (BY)
twenty mid-altitude maize varieties released over the past 29 years had shown a positive and highly significant (P≤0.01) associations with grain filling rate (r= 0.97**), biomass yield (r= 0.90**), biomass production rate (r= 0.91**) and harvest index (r= 0.59**), while grain yield was positive and significantly (P≤0.05) associated with plant height (r= 0.52*), ear length (r= 0.56*) and thousand kernel weight (r= 0.48*) at BNMRC (Table 5). Correlation coefficients for the grain yield among the eleven lowland maize varieties released over the past 12 years had shown a positive and significant (P≤ 0.05) association only with harvest index (r= 0.69*) at MARC (Table 5).

Analysis of variance of grain yield and other agronomic traits of maize varieties
The highly significant mean squares observed for grain yield and other measured traits over the breeding period indicate that genetic differences exist among cultivars within each breeding period over Ethiopian released highland, mid-altitude and lowland maize varieties. The analysis of variance for grain yield traits indicated significant (P≤0.01) differences on the number of ears per plant and grain yield among the varieties released in Ethiopia over highland, mid-altitude and lowland maize varieties. These findings were in agreement with the genetic gain study findings of highly significant (P≤ 0.01) differences on the number of ears per plant and grain yield which were indicated both under multiple stress and non-stress environments at Nigeria

Genetic gains in grain yield and other agronomic traits of maize varieties
Maize genetic gains in grain yield and other measured traits for Ethiopian released maize varieties currently under production within breeding periods in the present studies prompted the examination of the archived and predicted genetic gains that the Ethiopian released highland and mid-altitude maize varieties over the past 39 and 29 years demonstrated positive genetic gains for the grain and biomass yields. Comparably numerous estimates of genetic yield gain of maize hybrids have been shown, without exception, that genetic yield gains during the past 70 years have been positive and linear. Estimates of the average annual gain vary but tend to fall in the range of 65-75 kg ha -1 according to Duvick (2005a). This agrees with a recent result from USA by Chen et al. (2016) who evaluated commercial maize hybrids released over 38 years that reported increased breeding progress over the grain yield by an average of 66 kg ha -1 yr -1 . However, the present studies for the Ethiopian released lowland maize varieties during the past 12 years differently demonstrated genetic reduction for grain yield, while only minimal genetic gain for biomass yield were shown.
The highland and mid-altitude maize varieties demonstrated that non-significant and significant genetic gain improvements on duration reductions had been possible for days to anthesis and silking, while nonsignificant genetic gain decrease was made upon duration reduction for days to anthesis and silking for the lowland maize varieties. In the history of the maize breeding programs of some countries, there have been consistent as well as inconsistent trends made possible on reducing the durations of days to anthesis and silking. Many researchers agree for growth and flowering traits that days to silking and anthesis have not significantly changed over time respectively according to (Russell, 1985;Duvick, 1997Duvick, , 2005a. On the contrary, Omolaran et al. (2014) and Badu-Apraku et al. (2014) reported over the three different breeding eras, that days to anthesis were significantly and consistently lowered over the newly released ones than the oldest released ones. This clearly indicates that throughout the history of the maize breeding program there has been a continual trend made possible on reducing the durations of days to anthesis and silking in many countries.
Highly significant genetic improvement was made upon prolonging grain filling period for the mid-altitude maize varieties, while non-significant genetic improvements were made for the highland and lowland maize varieties. The first two shown findings agreed with Campos et al. (2006) who reported for maize that the grain filling period has been non-significantly improved over the past fifty years of breeding in the U.S. corn-belt. Non-significant genetic reduction of grain filling rate was made for the lowland maize varieties, while genetic increases of grain filling rate were made for the highland and mid-altitude maize varieties. The shown genetic gain increases and reduction of grain filling rates were the ones that have played the role for the realized grain yield as well as thousand kernel weight potentials over the Ethiopian released maize varieties. It was obvious that kernel set must be followed by kernel filling to ensure that yield potential is realized. Kernels near the tip of the ear will often abort after several weeks of growth if droughtaffected. Remobilized assimilate stored in the stem prior to and during the flowering period normally plays a role in buffering filling rate only in the last half of filling (Edmeades, 2013).
Non-significant genetic improvements for number of ears per plant over the mid-altitude maize varieties and reduction for number of ears per plant over the lowland maize varieties were shown, while significant genetic improvement was shown for number of ears per plant over the highland maize varieties. Unlike the lowland maize varieties, comparable results on different hybrids maize varieties grown in USA reported that number of ears per plant was found to increase over the decades (Crosbie, 1982;Russell, 1985;Duvick et al., 2004). Similarly, in Nigeria and Benin significant improvement was observed in number of ears per plant for the different maize cultivars of the three breeding periods when grown in Striga-infested and Striga-free. That the genetic gains increase made in the number of ears per plant were 0.006 and 0.002 ear plant -1 yr -1 over the evaluated different maize cultivars respectively, under Strigainfested and Striga-free conditions. Nevertheless, the different maize cultivars evaluated under Striga-free condition, the number of ears per plant were ranged equally from 0.9 ear plant -1 yr -1 for cultivars during the breeding period 1 (1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000) to 0.9 ear plant -1 yr -1 for cultivars during the breeding period 3 (2007)(2008)(2009)(2010), while under Striga-infested conditions the number of ears per plant was ranged from 0.8 to 0.9 ear plant -1 yr -1 over the two similar breeding periods (Badu-Apraku et al., 2013). Badu-Apraku et al. (2014) also reported the number of ears per plant on maize, that the genetic gains were changed significantly by 0.52 and 0.70 ear plant -1 yr -1 during the three breeding eras respectively; under multiple stress and non-stress environments evaluated at 16 and 35 different sites.
Positively non-significant genetic improvements respectively, over the highland and mid-altitude maize varieties, and negatively non-significant genetic reductions over lowland maize varieties were shown for harvest index. The demonstrated research findings for the genetic improvements for harvest index towards the released maize varieties agreed that the harvest index did consistently change over time; and that in Argentina the harvest index over the evaluated Argentinean maize hybrids, have increased from 0.41 to 0.52 kg ha -1 yr -1 on those maize hybrids varieties grown under the optimal conditions over those past 30 year period of 1960-1990(Echarte and Andrade, 2003Echarte et al., 2004). From another study, particularly at higher plant densities in Iowa-USA, the harvest index showed a significant relative improvement of 0.1 kg ha -1 yr -1 over the maize varieties released for the past 61 year period of 1930-1991(Duvick, 1997. On the contrary in Iowa-USA, for the long-term genetic gain in maize yield for the conditions of the U.S. corn-belt, the harvest index have remained constant over maize hybrids released between the 1930s-2000s for the past 70-80 years (Duvick, 2005b;Tollenaar and Lee, 2006). Another recent study on the commercial hybrid maize varieties released in the USA over the eight commercial DeKalb hybrid maize varieties released over 38 year period from 1967-2005 compared at 2 locations, 2 nitrogen fertilizer rates and 3 plant densities, showed that the harvest indices were similar across hybrid maize varieties except for low values with the 1967 and 1975 released hybrid maize varieties at West Lafayette, USA; and with the 1975 and 1982 released hybrid maize varieties at Wanatah, USA (Chen et al., 2016).

Relationship of grain yield and other agronomic traits of maize varieties
Genetic improvements of grain yield in the Ethiopian released highland and mid-altitude maize varieties over the past 39 and 29 years; grain filling rate, ear length, biomass production rate, biomass yield and harvest index were equally amongst the possible contributors oneness associated positively and significantly with the grain yields. Days to anthesis and days to silking were also amongst the possible contributors oneness associated negatively and significantly with the grain yields while, number of ears per plant and number of kernels per row were amongst the possible contributors oneness being positively and significantly associated with the grain yields over the Ethiopian released highland maize varieties. Equally, thousand kernel weights were the other ones amongst the possible contributors being associated positively and significantly with the grain yields over the Ethiopian released mid-altitude maize varieties. While only the harvest index were the ones among the possible contributors being associated positively and significantly with the grain yields over the Ethiopian released lowland maize varieties for the past 12 years.
For maize, grain yield is a function of number of plants per area, the proportion of these plants that produce a harvestable ear, kernel number per ear, and the weight of each individual kernel. Similar findings to the Ethiopian released highland and mid-altitude maize varieties were reported from Nigeria by Omolaran et al. (2014) on maize genetic gains studies under different nitrogen regimes, for highly significant and positive associations of grain yield with the number of kernels per row and thousand kernel weight; while highly significant and negative associations of grain yield with the days to anthesis, days to silking and plant height were identified. Other similar findings from Canada, for a significant and positive association of grain yield with thousand kernel weight (Lee and Tollenaar, 2007), and grain yield with number of kernels (Tollenaar et al., 1992); and from Nigeria and Benin by Badu-Apraku et al. (2013) grain yield with plant height were also reported. Meanwhile, other considered and analyzed success result studies on conventional maize crop improvements over the past 50 years for drought tolerance also indicated the negative association between grain yield and reduced interval between anthesis and silking (Campos et al., 2004).
As regards the Ethiopian released lowland maize varieties, harvest index was shown to be associated positively and significantly with the grain yields, and harvest index trait was also considered as being the ones among the possible contributors towards the grain yield genetic declinations. However, unlike the Ethiopian released lowland maize varieties, comparable results on grain yield improvement in Argentina has been associated with an increase in harvest index trait (Echarte and Andrade, 2003). In contrast to yield improvement in Argentina, previous studies (Crosbie, 1982;Duvick, 1997Duvick, , 2005bTollenaar et al., 1994) have also shown that increase in ERA-hybrid grain yield in the USA can be attributable to changes in light interception due to increased leaf area index and changes in light utilization due to more erect upper leaves, maintenance of green leaf area and leaf photosynthesis during the grain filling period rather than yield per plant and harvest index. Similarly, Tollenaar and Lee (2006) reported from the USA that the yield increase was not associated with a change in maximum harvest index.

Conclusions
We studied the changes in yield gains on a morphophysiological basis with respect to yield and yield component traits for 38 Ethiopian released maize varieties over the past 42-year periods which is currently under production in Ethiopia in the highland, mid-altitude and lowland Ethiopian major maize growing agroecology zones of the regions. The average rate of increase in grain yield corresponding to annual genetic gain was 62.26 (1.24%) kg ha -1 yr -1 over the tested 7 released highland maize varieties and 58.97 (0.78%) kg ha -1 yr -1 over the tested 20 released mid-altitude maize varieties. Differently, the other tested 11 released lowland maize varieties indicated average rate of decreases in grain yield was by -2.64 (-0.16) kg ha -1 yr -1 corresponding to annual genetic gain. Other tested phenological traits, and yield and yield components indicated a significant and positive annual genetic gain increase for number of ears per plant over the released highland maize varieties and grain filling period over the released mid-altitude maize varieties; while significant and negative annual genetic improvement were also observed in shortening the days to anthesis and days to silking over the released mid-altitude maize varieties. However, an average rate of decreases had been indicated in grain yield; a significant and positive annual genetic gain increase was indicated for number of kernel Kebede et al. 429 rows per ear over the released lowland maize varieties. Generally, the results of the present studies indicated that considerable genetic gains over the phenological traits, and inconsiderable genetic reductions over the yield and yield components have been made across the released highland, mid-altitude and lowland maize varieties for the three agro-ecological zones of Ethiopia. Typically, targeting one or few of those identified maize breeding traits relatively contributed to considerable genetic gains and reductions could be used for further improvements in the breeding program.