Potato (Solanum tuberosum L.) is one of the vital tuber crops in terms of food security for a growing population and increased hunger rates (CIP, 2018). Since its introduction to Ethiopia by a German immigrant in 1858 Wilhelm Schimper, the crop was cultivated by few number of farmers as a garden crop. The expansion and its adoption by Ethiopian farmers were gradual and stagnant for several decades (Baye and Gebremedhin, 2013). Strategic potato research in Ethiopia began in 1975 with the understanding of the constraints challenging its production and productivity (Baye and Gebremedhin, 2013). The development and dissemination of more than 36 improved varieties, coupled with other technological packages, contributed greatly to the improvement and rapid expansion in potato production (MANR, 2016). The major objective of potato breeding has been to develop potato cultivars that have maximum yield potential, adaptable to wide agro-ecologies and resistant to late blight that has been the most devastating disease throughout  the  dominant  potato  producing highlands of the country (Wassu, 2016). However, considerable number of released varieties became vulnerable to potato devastating disease because of the pathogen (Phytophthora infestans) ability to rapidly overcome resistant genes even though improvements on late blight resistance were made by breeders (Getachew et al., 2016; Wassu, 2017). This indicates the study on genetic progress with identifying the most important traits for yield as well as resistance of the crop for late blight was the missing component except little attempt made by Haramaya University (Wassu, 2017). Hence periodic assessment of genetic progress for these important traits could supplement the breeding program of the crop in the country. Furthermore, estimation of genetic progress of a breeding program and periodic assessment of advance (Wassu, 2017; CGIAR, 2016) in the genetic gain of a crop is vital to understand changes produced by breeding, to assess the past efforts made in genetic yield potential and to put forward future breeding strategy.

The field trial was carried out under rain fed condition at two locations, Holetta Agricultural Research Center and Adea Berga Research station (Table 1) during the main cropping season of 2017 using 20 released varieties and 1 farmers` cultivar of potato (Table 2). The experiment was arranged in a randomized complete block design with three replications. The experimental plot size was 4 rows each 3 m long and 3.6 m wide; plant spacing was 30 cm between plants and 75 cm between rows. Fertilizer  used in the rate of 108.44 kg N, 92.43 kg P and 16.59 kg S per hectare in the form of Urea (143 kg/ha) and blended fertilizer (NPS) (237 kg/ha) as per the recommendation for the study area (MALR, 2017). Other agronomic practices and data collection was conducted based on the recommendations of Holetta Research Centre (Lemaga et al., 1992).

Data analysis

All data were subjected to separate analysis of variance (ANOVA) of individual locations and a combined ANOVA over locations was done using the procedure of SAS software version 9.3 (SAS Institute, 2010) and a general linear model (GLM) for tuber yield related traits (Gomez and Gomez, 1984). The combined analysis of variance over locations were computed after homogeneity test of error variances using F-test as stated by Gomez and Gomez (1984).

Regression analysis was performed to calculate the genetic gain of yield and other yield related traits. The average annual rate of genetic gain for each trait was estimated by regressing of the mean value  of  each  trait  against  the  corresponding  year  of release of each variety (Singh and Chaudhary, 2007).

The regression equation was, Y = a + bx, assuming the linear relationship where Y = the mean value of the dependent variable, x = the mean value of the independent variable, a = the constant value and b = the regression coefficient.

Where, X = the year of variety release, Y = the mean value of each character for each variety, Cov = Covariance and Var = Variance.

Regression analysis was conducted taking tuber yield and other traits dependent and year of release as independent variable. The annual rate of genetic gain achieved over the last 30 years of potato improvement was determined as the ratio of genetic gain to the corresponding mean value of the oldest variety and expressed as percentage.

Relative annual rate of gain = Cov (X, Y)/Var (X)

Where, X is the year of variety release, Y is the mean value of each trait for each variety; Cov (X,Y) is the covariance of X and Y, and Var (X) is variance of X (year of variety release). Percent genetic gain per year for each variety was calculated as:

Percent Genetic Gain Year^-1 = [(XG-XAL-624)/XAL-624]/YG-YAL-624 × 100

Where, X is the mean value of observations for a given trait, Y is the year of release of each variety (G), AL-624 is the oldest variety of potato released in 1987.

The increment over farmers’ cultivars for each trait was calculated as:

Percent increment of Variety (%) = XG - XFC/XFC × 100

Where, XG is the mean value of each variety for each trait, XFC is the mean value of the farmers’ cultivar (Nech Abeba) in Central highlands of Ethiopia.

In this experiment, Alemaya 624 was considered as the oldest variety since it was the first potato variety released in the country in 1987.

Step-wise regression analysis was computed using the SAS software (version 9.3) (SAS Institute, 2010) to identify the most contributing traits to the variability that exists in the yield related traits were computed (Gomez and Gomez, 1984).

Improvement made on tuber yield and yield related traits

Yield data of each mean values from  combined  analysis of variance were used to regress the breeding progress of varieties for their tuber yield and other important traits. Potato varieties had huge total and marketable tuber yield difference. Trends in genetic progress, percent genetic gain over location and annual relative rate of genetic progress were calculated and presented in Tables 3 to 5.

Total tuber yield of variety Belete showed an increment of 137.39 and 0.851% over Nech Abeba (farmers` cultivar) and the oldest variety (AL-624), respectively (Table 5). The trend also showed total tuber yield increment over the oldest variety was 1.08% (Awash), 0.79% (Gudene and Gabissa) and 0.78% (Bedassa). On the other hand, the recent variety Dagim had a yield progress of -0.596% over the oldest variety. Hence, the yield increment was not constant throughout the breeding period and some varieties had yield performance below the oldest variety AL-624.

In terms of potato tuber yield, the genetic progress made over the last two to three decades since the first improved potato variety AL-624 was released comparing with the maximum yielding variety Belete was 0.009 t ha^-1. The relative annual rate of gain was 1.15 and 1.42% year^-1 for total tuber yield and marketable tuber yield, respectively. Both the average  tuber  number  and  tuber weight showed annual relative gain of 0.01 and 0.79% year^-1. Similarly, the annual rate of genetic progress in the present study was found to be 0.3177 ton year^-1 (20.27%) and 0.3401 ton year^-1 (24.55%) per hectare for total and marketable tuber yield, respectively (Figure 1a). On the other hand, non-significant and positive regression value was recorded for average tuber number and average tuber weight against year of variety release (Table 4).

Improvements on tuber quality traits

Year of variety release had positive and significant regression value with tuber dry matter content, starch content and total starch yield, but positive and statically non-significant positive value for tuber specific gravity (Table 4) indicating that potato breeders made improvements on tuber quality traits through breeding for the last two to three decades. Among quality traits studied, total starch yield had maximum relative gain 1.70% year^-1 followed by starch and dry matter content (0.80 and 0.44% year^-1, respectively). For both tuber dry matter and starch content in percent, the annual rate of genetic progress in the  present  study  was  found  to  be 0.1326 (R²=0.2063) and 0.1633 (R²=0.2077), respectively (Figure 1b and c). Total starch yield also showed positive significant value with year of variety release, whereas tuber specific gravity had positive and statistically non-significant regression value against year of variety release (Table 4).

Genetic progress on growth and Phenological traits

Year   of   variety   release   had  positive  and  significant regression value with days to physiological maturity. Non-significant negative regression value was recorded for days to 50% flowering indicating the breeding made some weak improvements for this trait. Days to physiological maturity showed a significant positive regression value with a certain amount of increase against the year of variety release without affecting yield of potato (Table 4). However, days to 50% flowering had negative and statistically non-significant correlation value with yield of potato varieties indicating that high yielding varieties have a retarded flowering. On the other hand, all the phenological and growth traits except days to physiological maturity demonstrated statically non-significant values with the year of variety release (Table 4). Days to 50% flowering and days to physiological maturity showed annual relative gains of -0.04 and 0.38% year^-1. The other growth traits like number of leaves per hill, plant height and stem number per hill showed relative gains of 0.33, 0.46, and 0.71% year^-1, respectively. The genetic progress for growth traits recorded was 0.29 cm for plant height, 0.15 for number of leaves per plant and 0.03 for stem number per hill.

The results of stepwise regression analysis between total tuber yield and other traits were presented in Table 6. The traits (predictors) that were retained in the model after regression analysis with the dependent traits were significant (P ≤ 0.05 and P ≤ 0.01) and regarded as the most important traits for predicting the tuber yield. Based on stepwise regression analysis, the best combination of characters that contributed to genetic progresses in tuber yield included were average tuber number per hill, stem number per hill and marketable tuber yield. About 98.2% of the variations on  total  tuber  yield  were  explained  by stem number per hill followed by average tuber number per hill and marketable tuber yield t ha^-1 contributing to 97.8 and 97.2%, respectively. Therefore, it can be considered that changes in the above three traits had possibly contribution to the changes in total tuber yield in the last two to three decades of potato breeding in Ethiopia.

The inconsistency in yield increment among varieties respect with their year of release could be due to variety AL-624 performed well across a wide range of agro ecologie. Attributing on farm productivity to genetic improvement is also more problematic in a vegetatively propagated crop like potatoes than in sexually propagated cereal. The absence of effective seed systems means that a productivity effect from a variety is confounded with the effect of cleaner or physiologically more correct seed (Sarker et al., 2018). The other important reason could be seed degeneration of the varieties due to their different seed sources and recycling the same seed source for subsequent production seasons (Sarker et al., 2018). Seed potato degeneration, the reduction in yield or quality caused by an accumulation of pathogens and pests in planting material due to successive cycles of vegetative propagation had a great influence on the performance of varieties (Thomas-Sharma et al., 2016).

Similar finding was recorded by Wassu (2017) and some recently released varieties showed lower yield performance than the oldest variety (AL-624) in Eastern Ethiopia. The correlation analysis of the potato for total and marketable tuber yield and respective year of variety release had also positive and significant association. Whereas unmarketable tuber yield had negative non-significant correlation value with year of variety release indicating that potato breeders in Ethiopia made some considerable improvement on seed tuber size.

Wassu (2017) recorded the highest  estimates for annual genetic gain of total tuber yield (4.05%) in the locality of Hirna with the potato variety Gera and the lowest at Arbarakete for the potato variety Gorebella (-3.02%). Similarly, the variety Chiro at Haramaya had the lowest (-3.43%) and Gera variety at Hirna had the highest (4.94%) genetic gain for marketable tuber yield relative to the oldest variety AL-624. Similar genetic gains were reported by many researchers due to improvements in varieties and agronomic practices (Tamene et al., 2015; CGIAR, 2016). Studies on crop improvement for important agronomic and quality traits were undertaken by different scholars  using  gene  engineering  and  other breeding methods (Douches et al., 2015; Massa et al., 2015; Liu, 2017).

The study showed that the high annual rates of gain have been achieved in tuber yields and tuber quality related traits through three decades of potato improvement efforts made in the nation though the gain was inconsistent over years of release of varieties and across the centres that developed the varieties. This might be due to different seed generations, variable sources of seeds from different growing and storage conditions and variations in seed physiological status. Therefore, further study is recommended with tissue culture planting materials with no seed degeneration difference among the varieties to isolate the variety effect, tuber seeds of checks has to meet the same health and physiological standards of prospective varieties over major growing areas including all released varieties and farmers cultivars of each growing area, and considering disease reaction and other important agronomic aspects of potato varieties to design appropriate potato improvement strategy in the country.

The authors have not declared any conflict of interests.