Euphytica (2022) 218:57 https://doi.org/10.1007/s10681-022-03011-1 Genotype by environment interaction and grain yield stability of drought tolerant cowpea landraces in Ethiopia Tesfaye Walle Mekonnen  · Firew Mekbib  · Berhanu Amsalu · Melaku Gedil  · Maryke Labuschagne Received: 6 November 2021 / Accepted: 24 March 2022 / Published online: 12 April 2022 © The Author(s), under exclusive licence to Springer Nature B.V. 2022 Abstract Cowpea is one of the most important of environment, genotype and GEI as 63.98, 2.66% indigenous food and forage legumes in Africa. It and 16.30% of the total variation for grain yield, serves as a primary source of protein for poor farmers respectively. The IPCA1, IPCA2 and IPCA3 were in drought-prone areas of Ethiopia. The crop is used all significant and explained 45.47%, 28.05% and as a source of food, and insurance crop during the dry 16.59% of the GEI variation, respectively. The results season. Cowpea is adaptable to a wide range of cli- from AMMI, cultivar superior measure, genotype matic conditions. Despite this, the yield of the crop plus genotype-by-environment biplot yield stabil- is generally low due to lack of stable and drought- ity index, and AMMI stability value analyses identi- tolerant varieties. In this study, 25 cowpea genotypes fied NLLP-CPC-07-145-21, NLLP-CPC-103-B and were evaluated in five environments using a lattice NLLP_CPC-07-54 as stable and high yielding geno- design during the 2017 and 2018 main cropping sea- types across environments. Thus, these genotypes sons. The objectives of this study were to estimate the should be recommended for release for production magnitude of genotype by environment interaction in drought-prone areas. NLLP-CPC-07-143, Kan- (GEI) and grain yield stability of selected drought- keti and CP-EXTERETIS were the least stable. The tolerant cowpea genotypes across different environ- AMMI1 biplot showed that Jinka was a high potential ments. The additive main effect and multiplicative and favorable environment while Babile was an unfa- interaction (AMMI) model indicated the contribution vorable environment for cowpea production. Keywords AMMI stability value · Environment · T. W. Mekonnen (*) · M. Labuschagne  GGE biplot · Yield stability index · Yield Department of Plant Sciences, University of the Free State, Bloemfontein 9301, South Africa e-mail: tesfaye.walle@gmail.com Introduction F. Mekbib  School of Plant Sciences, Haramaya University, Cowpea [Vigna unguiculata (L.) Walp.] is one of the Dire Dawa, Ethiopia most important food and forage legumes grown in the B. Amsalu  semi-arid tropics and some temperate regions of the Melkassa Agricultural Research Center, Adama, Ethiopia world (Timko and Singh 2008). Cowpea, at all stages of growth, serves as food source (Ahenkora et  al. M. Gedil  International Institute of Tropical Agriculture (IITA), 1998; Singh et  al. 2003). Poor people in developing Ibadan, Nigeria countries of the tropics derive their protein, animal Vol.: (0123456789) 1 5 7 Page 2 of 13 Euphytica (2022) 218:57 feed and cash income from the production of the crop and yield stability index (YSI) (Tumuhimbise et  al. (Diouf 2011). The young leaves, green pods and green 2014). Understanding the magnitude of effects of grains are used as vegetables, and the dry grains are the environment, genotype and their interaction on used in various preparations for both human food and yield and stability performance of cowpea genotypes livestock feed (Filho et al. 2017; Owade et al. 2020). across environments are important because it reduces The cowpea grain is highly nutritious and contains the efficiency of the genetic gain through the develop- about 22.8–28.9% protein with an average of 25.6% ment of high yielding genotypes with desirable agro- (Weng et  al. 2019). The portion of the cowpea crop nomic traits (Hall et  al. 2003; Simion et  al. 2018). above ground (except for pods) serves as a useful Ethiopia is a victim of repeated droughts that cause source of nutrient-rich fodder for livestock in many partial or total crop failure, and subsequently, famine areas of the world (Singh et  al. 2003; Weng et  al. in the country. Cowpea can be used to reduce the con- 2019). Cowpea can withstand harsh growing condi- sequence of drought. tions, particularly temperature and moisture stress Knowledge on the effect of genotype, environ- in comparison to other crops (Agbicodo et al. 2009; ment, and their interactions on drought tolerant cow- Goufo et al. 2017; Fatokun et al. 2012). pea grain yield is limited in Ethiopia. The objectives Environment, growing season, and rainfall dis- of the present study, therefore, were to estimate the tribution and intensity may have positive or nega- magnitude of GEI and grain yield stability of selected tive impacts on cowpea genotypes (Marina et  al. drought tolerant cowpea genotypes across drought 2017). Plant breeders evaluate genotypes in multi- prone environments. environments, representing favorable and unfavora- ble growing conditions, to estimate and understand the complexity of the genotype across environments (Mohammadi and Amri 2008). Materials and methods Cowpea is greatly influenced by seasonal envi- ronmental fluctuations and shows large genotype by Study area environment interaction (GEI), which is a major chal- lenge in obtaining a full understanding of genetic The field experiments were conducted during the control of varieties when compared across a series of 2017 and 2018 main cropping season at five environ- environments (Kuruma et al. 2019; Olajide and Ilori ments namely Babile, Melkassa, Miesso, Jinka and 2017; Gerrano et al. 2020; Simion 2018; Simion et al. Sirnka (Kobo) (Table 1). 2018). Studying and understanding of GEI is impor- tant to plant breeding programs for improving yield Plant materials and yield components (Yan and Tinker 2006) and is also used for identifying the basic causes of differ- A total of 25 genotypes were used for this study ences between genotypes for yield stability (Yan and (Table 2). Of these, two released varieties were used Tinker 2006). GEI exists when the ranks of geno- as standard checks and 23 genotypes were selected types show an obvious shift from one environment to as drought tolerant from a drought stress experiment, another (Leflon et al. 2008). Measuring GEI is impor- which included 324 genotypes (data not shown). tant to determine the optimum breeding strategy for releasing genotypes with adequate adaptation to tar- get environments (Das et al. 2019; Yan 2016). Experimental Design and Procedures The methods commonly used for quantifying GEI and stability include; principal component analy- The experiments were laid out using a 5 × 5 lattice sis (PCA) (Zobel et  al. 1988), additive main effect design with three replications. The plot size was 4 m and multiplicative interaction (AMMI) (Zobel et  al. long, 0.75 m between rows and 0.2 m between plants. 1988), genotype plus genotype-by-environment Plots consisted of four rows with 20 plants per row. (GGE) biplot analysis (Yan and Tinker 2006), culti- The distance between plots, intra blocks and replica- var superiority measure (Pi) (Lin and Binns 1988), tions was 1 m, 1.5 m and 2 m, respectively. The data AMMI stability value (ASV) (Purchase et  al. 2000) were collected from the middle two rows. Vol:. (1234567890) 1 3 Euphytica (2022) 218:57 Page 3 of 13 57 Table 1 Description of experimental sites Experimental sites Soil type Altitude Geographical location Rainfall Temperature (°C) (m.a.s.l) Latitude Longitude (mm) Min T Max T (N) (E) Babile Sandy 1647 9° 13′ 09′′ 42° 19′ 432 13.85 28.9 Melkassa Andosols 1550 8°3 0′ 39° 24′ 763 15.73 27.31 Miesso Vestisol 1470 9°14′ 40°45′ 787 18.4 30.70 Jinka Vestisol 1383 5° 52′ 36° 38′ 1247.7 17.60 30.00 Sirnka (Kobo) Euric Fluvsol 1470 12° 09′ 39° 28′ 637.3 13.00 34.00 m.a.s.l = meters above sea level, N = North, E = east, mm = millimeter, °C = degree Celsius Min T = minimum temperature, Max T = maximum temperature, Weather data accessed from meteorological stations at each site Table 2 List of genotypes Genotypes Genotype code Source Breeding status used for genotypes by of the genotypes environment interaction NLLP-CPC-07-10 G1 MARC Landrace CP-EXTERETIS G2 MARC Landrace NLLP-CPC-07-169 G3 MARC Landrace Dass 001 G4 MARC Landrace ACC-215-821 G5 MARC Landrace ACC-216-747 G6 MARC Landrace NLLP-CPC-07-139 G7 MARC Landrace NLLP-CPC-07-145-21 G8 MARC Landrace NLLP-CPC-07-143 G9 MARC Landrace ACC-215-762 G10 MARC Landrace NLLP-CPC-103-B G11 MARC Landrace NLLP-CPC-07-156 G12 MARC Landrace NLLP-CPC-07-54 G13 MARC Landrace NLLP-CPC-07-28 G14 MARC Landrace NLLP-CPC-07-36 G15 MARC Landrace NLLP-CPC-07-19 G16 MARC Landrace NLLP-CPC-07-03-B G17 MARC Landrace NLLP-CPC-07-167 G18 MARC Landrace NLLP-CPC-07-166 G19 MARC Landrace NLLP-CPC-07-157 G20 MARC Landrace NLLP-CPC-07-140 G21 MARC Landrace NLLP-CPC-07-57 G22 MARC Landrace ACC-211-490 G23 MARC Landrace MARC = Melkassa Bole (Standard check) G24 MARC Released variety Agricultural Research Kanketi (Standard check) G25 MARC Released variety Center Data Collection and Analyses SAS (SAS 2013) software was used for combined analysis of variance (ANOVA) over environments According to the descriptors of cowpea (IBPGR and seasons. A mixed linear model was used for 1983), yield on a plot basis was collected and ANOVA (Gomez & Gomez 1984). Environment and converted to grain yield per hectare (kg h a−1). season were considered as random and genotypes as Vol.: (0123456789) 1 3 5 7 Page 4 of 13 Euphytica (2022) 218:57 fixed effects (Hartley 1950). F-Max ratio was used YSI = RASV + R to test the homogeneity of error variances before analyzing the combined data. GEA-R was used for where: RASV is the ranking of the AMMI sta- AMMI, GGE biplot, and Pi analysis (Pacheco et al. bility value and R the ranking of genotypes in all 2016). environments. The AMMI model analysis proposed by Zobel Pi measures the deviation from the yield of a et  al., (1988) was used for analyzing GEI. AMMI given genotype in relation to the maximum in each partitions the sum of squares into interaction princi- environment. The significant difference of Pi was pal component (IPC) axes. compared by computing a cutoff point for each GGE biplots were constructed from the data (Yan value. Even though distributional properties of the et al. 2000, 2007; Yan and Rajcan 2002; Yan 2011). cultivar superiority measure is not exactly known, The GGE biplot has many visual interpretations, the cut-off point was calculated by multiplying the which the AMMI does not have. It also allows visu- 5% or 1% significant F-values for Pi at environment alization of crossover GEI (Yan et al. 2007). Moreo- (E) and genotype (E-2) degrees of freedom by the ver, the GGE biplot is more logical for biological deviation from regression mean squares, where G objectives in terms of explaining the first PC score, and E denote the number of genotypes and envi- which represents genotypic level rather than addi- ronments respectively (Lin and Binns 1988). The tive level (Yan et al. 2000). The GGE biplot is based Pi measures were calculated using the following on the first two major components of a PCA using formula: the Site Regression (SREG) model. When the first component is highly correlated with the genotype main effect, the proportion of the yield is consid- ered to be due only to the characteristics of the gen- otype. The second component represents the vari- where: Ẋi = is the mean of genotype i in the environ- ′ ation in the yield due to the GEI (Yan 2011). The ments, M = is the genotype with maximum response GGE biplots were generated using a singular value among all genotypes in the jth environment, X ij = is decomposition model of the first two principal com- the response of the ith genotype in the jth environ- ponents (Yan 2002; Yan and Rajcan 2002). ment, Mj = is the genotype with maximum response GEA-R was used for AMMI, GGE biplot analy- among all genotypes in the jth environment. sis and Pi (Pacheco et al. 2016). According to Pur- chase et al. (2000) and Farshadfar et al. (2011) the ASV would be essential to quantify and rank geno- types according to their yield stability (Farshadfar Results et al. 2011). √ [ ] √ [ ]2 √ IPCA1Sumofsquares ASV = √ (IPCA1scores) + [IPCA 2 2scores] IPCA2Sumofsquares where ASV = AMMI’s stability value, IPCA1 = inter- Analysis of variance for grain yield across action of first principal component, IPCA2 = interac- environments tion of second principal component. YSI incorporates both mean yield and stability The results of the combined ANOVA across the in a single criterion. Low values of both parameters tested genotypes showed that environment (E) show desirable genotypes with high mean yield and and season (Y) main effects, G × E, G × Y, E × Y, stability (Tumuhimbise et  al. 2014). The YSI was G × E × Y were all highly significant (p < 0.0001) calculated using the following formula: for grain yield, and the genotype main effect (G) was significant (p < 0.05) for grain yield Vol:. (1234567890) 1 3 Euphytica (2022) 218:57 Page 5 of 13 57 Table 3 Sum of squares from combined ANOVA for grain (Table  3).*P ≤ 0.05, ** P ≤ 0.01, DF = degree of yield of 25 cowpea genotypes freedom. Source of variation DF Mean square Grain yield performance across the environments Replication 2 71,291.20 Genotype (G) 24 208,390.70* G2, G8, G10, G3 and G 18 had high yield per- Environment (E) 4 30,026,430.10** formance at specific environments Sirnka, Jinka, Season (Y) 1 57,369,811.20** Melkassa, Miesso, and Babile, respectively. However, Genotype* Environment 96 318,675.40** G20, G9, G12, G15, G11, and G9 were the poorest Genotype* Season 24 334,214.70** performers at Sirnka, Jinka, Melkassa, Miesso, and Environment* Season 4 25,866,297.40** Babile, respectively (Table  4). The highest grain Genotype* Environment* Season 96 301,461.90** yielding was recorded at Jinka (2445.7  kg  ha−1), Error 498 121,728.80 while the poorest grain yield was recorded at Babile Table 4 Grain yield performance across the environments N/S G Sirnka G Jinka G Melkassa G Miesso G Babile G GM 1 G2 1804.7A G8 2445.7A G10 1905.4A G3 2252.2A G18 909.77A G8 1641.25 2 G11 1757.7BA G6 2333.3BA G13 1769.3BA G19 2192.7A G5 898.4BA G17 1595.47 3 G24 1693.0BAC G12 2282.7BAC G7 1748.4BA G12 2166.6BA G20 896.94BA G13 1582.33 4 G4 1602.8BDAC G13 2240.2BDAC G8 1696.6BAC G8 2163.4BA G9 852.86BAC G6 1566.2 5 G16 1597.4BDAC G5 2150.6BDAC G17 1694.9BAC G23 2156.2BAC G22 834.36BAC G14 1551.03 6 G13 1584.0BDAC G1 2139.4EBDAC G14 1691.0BAC G21 2049.4BDAC G25 824.45BAC G12 1529.22 7 G10 1555.6EBDAC G15 2109.4EBDACF G21 1656.3BDAC G25 2015.5EBDAC G23 821.78BAC G11 1523.12 8 G17 1549.7EBDAC G22 2106.9EBDACF G16 1544.8BDEC G11 1979.5EBDACF G15 821.02BAC G21 1521.99 9 G7 1536.5EBDAC G14 1973.4EBDAGCF G1 1541.0BDEC G14 1975.4EBDACF G17 817.86BAC G1 1515.8 10 G15 1517.2EBDAC G3 1972.0EBDAGCF G11 1516.2BDEC G17 1966.1EBDACF G21 787.9BDAC G19 1513.18 11 G6 1503.3EBDAC G19 1971.1EBDAGCF G20 1489.4FBDEC G20 1936.9EBDACF G4 783.2BDAC G3 1489.99 12 G12 1449.7EBDACF G17 1948.8EBDAGCF G24 1476.9FBDEC G6 1886.6EBDACF G2 770.67BDAC G15 1486.16 13 G18 1430.0EBDACF G11 1806.3EBDHGCF G15 1466FBDEC G4 1832.8EBDACF G13 770.05BDAC G7 1480.54 14 G3 1404.3EBDCF G4 1786.5EDHGCF G6 1393.1FGDEC G1 1815.9EBDACF G7 756.3BDEC G4 1471.45 15 G21 1401.0EBDCF G16 1778.2EDHGCF G4 1352.0FGDE G10 1804.5EBDACF G12 746.35BDEC G10 1468.25 16 G25 1382.8EBDCF G7 1746.7EDHGF G22 1342.1FGDE G5 1794.5EBDACF G14 734.93DEC G5 1459.13 17 G14 1380.5EBDCF G21 1715.4EDHGF G18 1329.1FGDEH G18 1704.2EBDCF G1 734.75DEC G16 1450.65 18 G19 1349.2EDGCF G18 1615.8EHGF G19 1321.7FGEH G2 1671.3EDCF G19 731.19DEC G23 1432.33 19 G1 1348.0EDGCF G23 1610.8EHGF G9 1281.1FGEH G9 1655.4EDF G10 728.84DEC G18 1397.76 20 G5 1347.3EDGCF G20 1594.5HGF G2 1278.9FGEH G16 1615.2EDF G16 717.69DEC G2 1397.63 21 G23 1316.2EDGCF G24 1594.3HGF G23 1256.7FGEH G7 1614.8EDF G8 717.22DEC G22 1396.82 22 G9 1223.7EDGF G25 1527.7HG G25 1178.8FGH G22 1586.8EDF G6 714.69DEC G24 1390.28 23 G8 1183.4EGF G2 1462.6HG G3 1168.7FGH G24 1577.2EDF G3 652.89FDE G25 1385.84 24 G22 111.4GF G10 1346.9H G5 1104.9GH G13 1548.2EF G24 610.0FE G20 1379.77 25 G20 981.2G G9 1324.9H G12 1000.8H G15 1517.2F G11 555.81F G9 1267.62 GM 1440.52 1863.36 1448.16 1858.13 767.6 1475.75 LSD 380.87 529.94 333.77 488.26 15,248 176.99 R2 59.72 55.53 79.99 88.33 93.34 CV 23.08 24.82 20.12 22.92 17.34 G = genotype, GM = grand mean, LSD = least significant difference, R2 = coefficient of determination, CV = coefficient of variation, LSD = least significant difference. Values in columns followed by different letters are statistically different at P ≤ 0.05 Vol.: (0123456789) 1 3 5 7 Page 6 of 13 Euphytica (2022) 218:57 (555.81  kg  ha−1). The highest yielding genotype (Table 6). The most unstable genotypes were G9, G2, across the environments and seasons was G8 with a and G25 which ranked 25th, 20th and 24th for grain yield of 1641.25 kg ha −1 while the poorest yielding yield. genotype was G9 with a yield of 1267.62 kg ha −1. GGE biplots Additive main effect and multiplicative interaction Which‑won‑where and what analysis (AMMI) The polygon view of the genotypes in the GGE AMMI analysis (Table 5) showed that environment biplot for 25 genotypes is presented in Fig.  1. The and GEI effects were highly significant (p < 0.0001), cumulative variation contributed by PC1 (AXIS 1) and genotype effect was significant (p < 0.05) for and PC2 (AXIS2) was 69.16%, both of which were grain yield. The test environments contributed highly significant. The biplot showed that two envi- 63.98% of the total variation in yield. Genotype and ronments (Babile and Sirnka) grouped in the same GEI accounted for 2.66% and 16.30% of the total mega-environments. The other three environments variation for grain yield, respectively. The ratio of (Jinka, Melkassa, and Miesso) each fell in a different genotype effect to genotype + genotype × environ- mega-environment. The plot showed that G8, G13, ment (G + G × E) was 0.14. The magnitude of the G3, G16 and G4 recorded the highest grain yield in GEI sum of squares was 6.12 times that of the gen- Jinka, Melkassa, Miesso, Sirnka, and Babile. On the otype sum of squares for grain yield. The AMMI other hand, G9, G25, and G23 did not fall in a spe- model extracted three highly significant (p < 0.0001) cific environment and were poor yielders. IPCA’s from the interaction PC axes (Table  5). Those three IPCA’s accounted for a total of 90.11% Ideal genotypes (ranking genotypes) of the observed variation due to GEI. IPCA1, IPCA2, and IPCA3 captured 45.47%, 28.05%, and The first two principal components (PC1 and PC2) 16.59% of the sum of squares, respectively. were highly significant (p < 0.0001) and explained 42.31% and 26.85% of the yield variation among the Lin and Binns Cultivar superiority measure (Pi) genotypes, respectively (Fig.  2). The GGE biplot, which identifies ideal genotypes that are high yield- According to this stability model, the three most sta- ing and stable across the test environments, identified ble genotypes with the lowest Pi values were G8, G17 G8, G6 and G1, which fell close to the center of the and G13, which ranked 1st, 2 nd and 3 rd for grain yield concentric circle as ideal genotypes. Based on the Table 5 AMMI analysis Source of variation DF SS MS Total VE Sum of squares of variance for grain yield −1 explained (%) (kg h a ) of 25 cowpea genotypes evaluated at five GEI E GEI cum environments Treatments 124 77,849,952 627,822 Genotypes 24 2,500,687 104,195* 2.66 Environments 4 60,052,810 15,013,202** 63.98 Block 10 559,100 55,910 Interactions 96 15,296,455 159,338** 16.30 *P ≤ 0.05, ** P ≤ 0.01, IPCA 1 27 6,710,300 248,530** 45.47 45.47 DF = degree of freedom, IPCA 2 25 4,336,683 173,467** 28.05 73.52 SS = sum of squares, MS = mean squares, IPCA 3 23 2,668,520 116,023* 16.59 90.11 Total VE = total variation Residuals 21 1,580,952 75,283 explained, GEI E = GEI Error 240 15,453,907 64,391 explained, GEI cum = GEI Total 374 93,862,959 250,970 cumulative. Vol:. (1234567890) 1 3 Euphytica (2022) 218:57 Page 7 of 13 57 Table 6 Grain yield and cultivar superiority value (Pi) average environment coordination (AEC) method, Genotypes y Mean R2 y Rank Pi Pi Rank G9, G25 and G23 were the most unstable and low yielding genotypes across the environments (Fig. 2). G1 1515.80 0.94 9 65,612.64 20 G2 1397.63 0.61 20 171,567.2 2 G3 1489.99 0.90 11 99,346.36 14 Ideal environment (ranking environment) G4 1471.45 0.95 14 97,339.92 15 G5 1459.13 0.78 16 114,667.1 10 Jinka was close to the concentric circle and provided G6 1566.2 0.91 4 53,755.99 23 the most ideal environment and the most powerful G7 1480.54 0.79 13 101,497.7 12 to discriminate performance of the tested genotypes G8 1641.25 0.88 1 47,458.89 24 (Fig. 3). In contrast, Babile was located far from the G9 1267.61 0.83 25 234,259.9 1 center of the concentric circle, indicating poor dis- G10 1468.25 0.50 15 150,238.8 5 criminating power. G11 1523.12 0.90 7 76,205.07 19 G12 1529.22 0.82 6 100,484 13 Mean vs stability G13 1582.331 0.73 3 62,452.33 21 G14 1551.03 0.96 5 55,618.29 22 The GGE biplot visualizes performance and effec- G15 1486.16 0.79 12 93,679.66 16 tively identifies the best performing genotypes across G16 1450.65 0.88 17 106,111.3 11 environments with the help of the AEC. The mean G17 1595.47 0.98 2 44,651.05 25 of PC1 and PC2 scores of the tested environments is G18 1397.76 0.97 19 146,153.6 7 represented by the arrowhead, and the AEC ordinate G19 1513.18 0.93 10 80,879.75 18 is the line that passes through the biplot origin and it G20 1379.77 0.69 24 167,535.7 3 is perpendicular to the AEC abscissa (Fig. 4). G21 1521.99 0.90 8 81,427.34 17 The length of the abscissa discriminates the grain G22 1396.83 0.76 21 135,756 9 yield of genotypes that are above and below average G23 1432.33 0.80 18 137,337.8 8 yield if right and left of the biplot origin, respectively G24 1390.28 0.76 22 146,642 6 (Fig. 4). The length of the ordinate approximates the G25 1385.85 0.79 23 161,189.5 4 GEI associated with the genotype stability, and a y = grain yield longer ordinate corresponds to higher variability and Miesso G25G23 G3 G12 G23 Miesso G25 G9 G20 G19 G3 G12 G5 G9 G20 G19 G18 Babile G2 G421 G5 G18 Babile G11 G22 G2 GG241 G14 G6 G8 G24 G17 G1 G11 G22 G10 G16 G14 G8 Sirka G1 G6 G7 G15 Jinka G24 G17 G10 SiGrk1a6 Jinka G7 G15 G13 Melkassa Melkassa G13 -500 0 500 1000 1500 AXIS1 42.31 % -500 0 500 1000 1500 AXIS1 42.31 % Fig. 1 Polygon view of GGE-biplot for 25 genotypes evalu- ated across five environments during 2017 and 2018 Fig. 2 GGE-biplot showing the ideal genotypes based on mean grain yield performance across environments Vol.: (0123456789) 1 3 AXIS2 26.85 % -1000 -500 0 500 AXIS2 26.85 % -1000 -500 0 500 5 7 Page 8 of 13 Euphytica (2022) 218:57 Furthermore, CP-EXTERETIS, Dass 001 (G4), Miesso NLLP-CPC-07-143 (G9), NLLP-CPC-07-166 (G19), NLLP-CPC-07-167 (G18), NLLP-CPC-07-157 (G7), and NLLP-CPC-07-140 () were stable but their yield below average, and NLLP-CPC-07-139 (), ACC-215- Babile 762 (G10), NLLP-CPC-07-19 (G16), ACC-211-490 (G23), Bole (G24) and Kanketi (G25) were not stable and their yield also below the average. Sirka Jinka AMMI stability value (ASV) and yield stability index (YSI) Melkassa Utilizing YSI, the combination of AMMI stability -500 0 500 1000 1500 values and average grain yield was estimated to quan- AXIS1 42.31 % tify and classify the genotype (Table  7). According to the YSI model, G14, G17, and G21 were the most Fig. 3 GGE-biplot showing the ideal environments for the stable genotypes across environments and high grain 2017 and 2018 seasons yielders. On the other hand, genotypes G10, G2, and one check (Bole) were unstable as indicated by high YSI values of 40, 41, and 45 respectively, and those genotypes had poor productivity and lower stability. G2G5 23 Miesso G3 G12 Discussion G9 G20 G19 The majority of cowpea growing environments are G18 G5 G2 GBa24b1ile extremely vulnerable to moisture stress, and the farm- G11 G22 ers use this crop as an insurance crop as they expe- GG1714 G6 G8 G10G24 G1 rience prolonged drought in this area. The higher SGirk1a6 G7 G15 Jinka yield variation contributed by the environment over genotype and GEI, indicates that the test environ- ments were highly variable and had a great impact on Melkassa G13 genotypes. The significant GEI necessitates the need to identify adaptable genotypes with consistent high -500 0 500 1000 1500 grain yield (Yan and Tinker 2006). AXIS1 42.31 % In addition, the response of genotypes varied con- siderably for grain yield due to the genetic makeup Fig. 4 Mean vs. Stability view of GGE biplot showing the of the materials and the interaction between genetic mean performance and stability of 25 genotypes constitution and environmental influences. This is in agreement with Gerrano et  al. (2020) who reported that genotype and environment directly affect the lower stability and vice-versa. NLLPP-CPC-07-10 yield potential of cowpea. From this study, the effect (G1), NLLP-CPC-07-57 (G20), ACC-216-747 (G6), of environment, season, and environment × season NLLP-CPC-07-145-21 (G8), NLLP-CPC-07-28 was high as it was responsible for 33.98%, 16.23%, (G14), NLLP-CPC-07-36 (G15), and NLLP-CPC- and 29.27% of the total variation for grain yield, 103-B (G11) were above average yielders with higher respectively. Thus, the environment and the season stability, however, NLLP-CPC-07-169 (G3), ACC- effects were very high, contributing to diverse cow- 215-821 (G5), NLLP-CPC-07-156 (G12), NLLP- pea grain yield. Cowpea grain yield will therefore CPC-07-54 (G13), and NLLP-CPC-07-03-B (G17) also be largely affected by climate change. A previous had above average yield but with lower stability. study also reported variation in responses of genotype Vol:. (1234567890) 1 3 AXIS2 26.85 % AXIS2 26.85 % -1000 -500 0 500 -1000 -500 0 500 Euphytica (2022) 218:57 Page 9 of 13 57 Table 7 AMMI stability Genotypes IPCA1 IPCA2 ASV rASV(A) YLD r YLD(B) YSI (A + B) value (ASV), yield stability index (YSI), ranks and G1 − 0.41 − 0.36 0.73 13 1515.80 9 22 IPCA scores G2 1.00 0.01 0.92 21 1397.63 20 41 G3 0.33 0.13 0.84 19 1489.99 11 30 G4 − 0.67 0.46 0.34 6 1471.45 14 20 G5 − 0.05 − 0.84 0.80 16 1459.13 16 32 G6 − 0.09 − 0.11 0.81 17 1566.20 4 21 G7 − 0.18 − 0.62 0.74 14 1480.54 13 27 G8 0.16 − 0.41 0.96 22 1641.25 1 23 G9 0.19 − 0.03 0.68 11 1267.62 25 36 G10 0.16 0.03 1.55 25 1468.25 15 40 G11 − 0.45 0.35 0.53 8 1523.12 7 15 G12 0.59 0.10 1.13 24 1529.22 6 30 G13 − 0.10 0.23 0.84 20 1582.33 3 23 G14 0.13 0.15 0.17 1 1551.03 5 6 G15 − 0.48 − 0.39 0.68 10 1486.16 12 22 G16 0.06 0.70 0.48 7 1450.65 17 24 G17 0.66 − 0.15 0.30 5 1595.47 2 7 G18 0.15 0.61 0.25 3 1397.76 19 22 G19 − 0.39 0.59 0.77 15 1513.18 10 25 G20 0.21 0.11 0.27 4 1379.77 24 28 G21 − 0.52 0.01 0.24 2 1521.99 8 10 G22 − 0.50 − 0.24 0.83 18 1396.82 21 39 G23 0.38 − 0.46 0.70 12 1432.33 18 30 G24 − 0.62 − 0.09 1.03 23 1390.28 22 45 G25 0.41 0.25 0.65 9 1385.84 23 32 across environments in different seasons (Kuruma of the highest yielding and adaptable genotypes for et al. 2019). the specific range of environments is important for Generally, the existing variation due to environ- the selection and evaluation of superior genotypes in ment, season, genotype performance and GEI in rela- multi-environment studies and are the main targets of tion to genotype effect suggested to there was the cowpea breeding programs. Kuruma et al. (2019) and possibility of mega-environment effects for different Muranaka et al. (2016) reported that high grain yield genotypes. Therefore, based on the variable response variation could be due to greater differences between of genotypes, it would help to map the mega-environ- the genotypes. ments suitable for the improvement of grain yield to AMMI analysis showed that the environmental combat the rapid climate change. effects accounted for the most (63.98%) of the total Grain yield increment is the goal of cowpea for variation compared to the other components, imply- any stressed environment because yield is governed ing that differential cowpea yield performance was by multi traits with different levels of expression for typically caused by environmental changes. This is in various environments and their interaction. In the agreement with the findings of Gerrano et al. (2020) present study, 52% of the tested genotypes recorded a and Simion (2018) that indicated the environment higher yield than the standard checks (Bole and Kan- made the largest contribution to grain yield variation keti). Genotype NLLP-CPC-07–145-21 (G8) had a in cowpea. The magnitude of the GEI sum of squares yield advantage of 15.26% and 22.25% compared to was 6.12 times that of the genotype sum of squares the worst genotype NLLP-CPC-07–143 (G9), and the for grain yield of cowpea, indicating that there were standard checks, respectively. Therefore identification Vol.: (0123456789) 1 3 5 7 Page 10 of 13 Euphytica (2022) 218:57 considerable differences in genotypic responses not necessarily give the best yield performance. across environments. Therefore, there is a need for approaches that incor- In the present investigation, the three IPCA’s porate both mean yield and stability in a single index. accounted for 90.11% of the interaction sum of The lowest YSI value is considered as the most sta- squares. Zobel et  al. (1988) stated that AMMI with ble, with high grain yield (Bose et al. 2014). NLLP- the first two multiplicative terms was the best predic- CPC-07–28 (G14), NLLP-CPC-07–145-21 (G8) and tive model. In this study, the high (45.47%) and sig- NLLP-CPC-07–54 (G13) were the most stable geno- nificant contribution of IPCA1 to the total variation types with good yield performance. Thus, according across the tested environments implies that IPCA1 to the YSI method, the most desirable genotypes can could identify stable and unstable genotypes based on be considered as widely adapted and with grain yield the value scores or nearest or furthest to zero, which above the grand mean among 25 genotypes. Similarly is in line with the findings of previous investigations Zali et  al. (2012) also indicated that both yield and (Muranaka et  al. 2016; Gerrano et  al. 2020; Simion stability of performance should be considered simul- 2018; Yaw et  al. 2020). The positive and negative taneously to exploit the useful effect of GEI and to IPCA scores of genotypes in AMMI analysis are the select genotypes for diverse environments. best indicators of stability or adaptation over environ- The polygon view of the “which -won-where and ments. High positive interaction of the genotypes like what” GGE-biplot (Fig.  1) showed that genotypes NLLP-CPC-07–169 (G3) in IPCA1 in an environ- NLLP-CPC-07–143 (G9), ACC-215–762 (G10), ment can exploit the agro-ecological conditions of the NLLP-CPC-07–156 (G12), NLLP-CPC-07–54 (G13), specific environment (Sirnka). Therefore, it would be NLLP-CPC-07–145-21 (G8) and ACC-211–490 possible to identify adaptable and suitable genotype/s (G23) were genotype markers located farthest from for the specific environment. Kandus et  al. (2010) the biplot origin in various directions and it shows and Yan et al. (2007) reported that the different high that the genotypes were well adapted to specific envi- yielding genotypes fall in a specific environment, and ronments. If the environment markers fall in different it shows crossover GEI, suggesting that the test envi- sectors it shows that different cultivars won in differ- ronment could be classified into mega-environments. ent environments (Oladosu et al. 2017; Gerrano et al. Using Pythagorean Theorem the distance from the 2020). In addition, because of long environment vec- origin (0:0) in a two-dimensional scattergram indi- tors, the rest of the genotypes scattered around the cates the most stable genotypes (Purchase et al. 2000). biplot origin. This clearly explained that the environ- In the ASV method, a genotype with the lowest ASV ment effect was higher than the genotype effect. This score is the most stable; accordingly, G14, G17, and implies that the genotype had less response for GEI G21 were stable and these genotypes were the highest because of high environment exertion to the GEI. yielding among the tested genotypes, indicating that The GGE-biplot showed that the tested environments the yield performance and stability showed a similar occurred in different sectors, indicating that the par- trend. Oliveira et al. (2014) noted that the dynamics ticular environment had different high yielding geno- of stable genotype and yield response are always par- types for those sectors, this indicating the existence allel to the mean response of the tested environments. of crossover GEI. It also indicated the possibility to Unstable genotypes like Bole, CP-EXTERETIS (G2), classify the environments into mega-environments and ACC-215–762 (G10) had high ASV values, and for cowpea production. Yan and Rajcan (2002) stated they were adapted to a specific favorable environ- that the presence or absence of crossover GEI indi- ment. Likewise, Oladosu et  al. (2017) reported that cates the existence of different mega-environments. the higher the IPCA score, and ASV the more specifi- Generally, the GGE biplot effectively identified the cally adapted a genotype is to a certain environment. best performing genotypes across environments and In this study, crossover stability and yield did not best genotypes for specific environments, whereby have the same trend for all genotypes across season specific genotypes can be recommended for specific and environment. Jadhav et  al. (2019) and Moham- environments and can be used to evaluate the yield madi and Amri (2008) suggested that principle stabil- and stability of genotypes, which is not possible with ity per se should, however, not be the only selection AMMI analysis (Kaya et  al. 2006; Yan and Tinker parameter because the most stable genotypes would 2006). Vol:. (1234567890) 1 3 Euphytica (2022) 218:57 Page 11 of 13 57 GGE biplots help visualize and compare the dis- be used for classifying the environments into mega- tance between each genotype and the ideal geno- environments for cowpea production. In terms of type located at the center of the concentric circle ideal genotypes (having a long vector), good yielding (Yan and Rajcan 2002). The ideal genotypes, based capacity and stability, NLLP-CPC-07-145-21 (G8), on proximity to the center of the concentric circle of NLLP-CPC-103-B (G11) and NLLP_CPC-07-54 the GGE biplot were ACC-216-747 (G6) and NLLP- (G13) were identified as ideal genotypes for drought- CPC-07-145-21 (G8), with high yield and stability prone environments for the country and could be pro- (Fig.  2). In addition, NLLP-CPC-07-10 (G1) was posed for release for production. The results of this located on the next homocentric circle and might be study confirmed that the ideal genotype should have considered as a desirable genotype. In principle, the a long vector, high yield performance, but such geno- ideal genotype should have the longest vector, high- types do not always exist in reality. est mean performance and with zero GEI, and/or it should perform consistently in all environments. Acknowledgements The authors sincerely acknowledge Because of the genetic background or nature of the the contribution of Wolkite University, Melkassa Agricul- tural Research Center (EIAR), and The McKnight Foundation traits (yield) and level of expression, the ideal geno- USA, for providing the research funds to complete this project. type does not always exist in reality. Therefore, such The author acknowledges the staff of national lowland pulse like genotypes the breeders can be used as a reference improvement program of Melkassa Agricultural Research for genotype for further study. Genotypes which were Center (MARC) is highly appreciated for their comprehensive support. high yielding but were not stable across environments could be recommended for a particular environment. Author contributions All authors contributed to the study Yan and Rajcan (2002) and Yan et al. (2007) speci- conception and design. Material preparation, data collection fied that the environments with long vectors (PC1 and analysis were performed by TWM, FM and BA. The first draft of the manuscript was written by TWM and FM and all scores) and relatively small angles or absolute with authors commented on previous versions of the manuscript. All the AEC abscissa are valuable for greater discrimina- authors read and approved the final manuscript for publication. tory capacity (in terms of the genotype main effect) and is representative of the other environments. Funding This study was supported by The McKnight Foun- Therefore, Jinka was in the epicenter of the concen- dation, USA through Collaborative Crop Research Program (CCRP) for Ethiopian Agricultural research Institute (EIAR). tric circle, and it was identified as a highly discrimi- nating environment for these genotypes, thus this Declarations environment is considered as an ideal environment for developing high yielding genotypes, or for iden- Conflict of interest The authors declare that they have no conflict of interest. tify ideal genotypes. Hence, Jinka allowed the geno- type to express genetic potential, minimizing popu- lation development expenses by discriminating the References worst genotypes at an early stage. Agbicodo EM, Fatokun CA, Muranaka S et al (2009) Breeding Conclusions drought tolerant cowpea: constraints, accomplishments, and future prospects. Euphytica 167:353–370. https://d oi. The present study showed that genotype, environ- org/1 0. 1007/ s10681- 009-9 893-8 Ahenkora K, Adu Dapaah HK, Agyemang A (1998) Selected ment, and genotype × environment, genotype × sea- nutritional components and sensory attributes of cowpea son and environment × season interaction effects were (Vigna unguiculata [L.] Walp) leaves. Plant Foods Hum significant for grain yield. The performance of tested Nutr 52:221–229. https:// doi. org/ 10. 1023/A: 10080 19113 genotypes was largely affected by the environment. 245 Bose LK, Jambhulkar NN, Pande K, Singh ON (2014) Use of Jinka was the most ideal environment and it had dis- AMMI and other stability statistics in the simultaneous criminatory power for genotype performance in all selection of rice genotypes for yield and stability under seasons. The AMMI showed that the environmental direct-seeded conditions. Chil J Agric Res 74:1–9. https:// effects accounted for 63.98% of the total yield vari- doi. org/ 10. 4067/ S0718- 58392 01400 010000 1 Das A, Parihar AK, Saxena D, Singh D, Singha KD, Kushwaha ation. The GGE biplot showed that the tested envi- KPS, Chand R, Bal RS, Chandra S, Gupta S (2019) Deci- ronments occurred in different sectors which could phering genotype-by-environment interaction for targeting Vol.: (0123456789) 1 3 5 7 Page 12 of 13 Euphytica (2022) 218:57 test environments and rust resistant genotypes in field pea Kaya Y, Akçura M, Taner S (2006) GGE-Biplot analysis of (Pisum sativum L). Front Plant Sci 10:825. https://d oi.o rg/ multi-environment yield trials in bread wheat. Turkish J 10.3 389/ fpls.2 019.0 0825 Agric for 30:325–337 De OEJ, Paulo J, Freitas XD, Jesus ON (2014) AMMI analy- Kuruma RW, Sheunda P, Muriuki C (2019) Yield stability and sis of the adaptability and yield stability of yellow passion farmer preference of cowpea (Vigna unguiculata) lines in fruit varieties. Sci Agricola 71:139–145 semi-arid eastern Kenya. Afrika Focus 32:65–82 Diouf D (2011) Recent advances in cowpea [Vigna unguicu- Leflon M, Lecomte C, Barbottin A, Jeuffroy MH, Robert N, lata (L.) Walp.] “omics” research for genetic improve- Brancourt-Hulmel M (2008) Characterization of environ- ment. Afr J Plant Biotech 10:2803–2810. https:// doi. org/ ments and genotypes for analyzing genotype × environ- 10.5 897/A JBx10.0 15 ment interaction. J Crop Improv 14:249–298. https://d oi. Farshadfar E, Mahmodi N, Yaghotipoor A (2011) AMMI sta- org/ 10.1 300/ J411v1 4n01 bility value and simultaneous estimation of yield and yield Lin CS, Binns MR (1988) A superiority measure of cultivar stability in bread wheat (Triticum aestivum L.). Aust J performance for cultivar x location data. Can J Plant Sci Crop Sci 5:1837–1844 68:193–198 Fatokun CA, Boukar O, Muranaka S (2012) Evaluation of Marina M, Fernández J, Ochoa J et  al (2017) Genotype by cowpea (Vigna unguiculata ( L.) Walp) germplasm lines environment interactions in cowpea (Vigna unguiculata L. for tolerance to drought. Plant Genet Resour Charact Util Walp) grown in the Iberian Peninsula. Crop Pasture Sci 10:171–176. https:// doi. org/ 10. 1017/S 1479 26211 20002 14 68:924–931 Filho JT, Dos Santos Oliveira CNG, da Silveira LM, de Sousa Mohammadi R, Amri ÆA (2008) Comparison of paramet- Nunes GH, da Silva AJR, da Silva MFN (2017) Genotype ric and non-parametric methods for selecting stable and by environment interaction in green cowpea analyzed via adapted durum wheat genotypes in variable environ- mixed models. Rev Caatinga, Mossoro 30:687–697 ments. Euphytica 159:419–432. https://d oi. org/1 0.1 007/ Gerrano A, Jansen van Rensburg WS, Mathew I, Shayanowako s10681- 007-9 600-6 AIT, Bairu MW, Venter SL, Swart W, Mofokeng A, Mel- Muranaka S, Shono M, Myoda T, Takeuchi J (2016) Genetic lem J, Labuschagne M (2020) Genotype and genotype × diversity of physical, nutritional and functional properties environment interaction effects on the grain yield perfor- of cowpea grain and relationships among the traits. Plant mance of cowpea genotypes in dryland farming system Genet Resour Charact Util 14:67–76. https://d oi. org/1 0. in South Africa. Euphytica 216:1–11. https:// doi. org/ 10. 1017/ S14792 6211 50000 9X 1007/s 10681-0 20- 02611-z Oladosu Y, Rafii MY, Abdullah N, Magaji U, Miah G, Hus- Gomez KA, Gomez AA (1984) Statistical procedures for agri- sin G, Ramli A (2017) Genotype × Environment interac- cultural research, 2nd edn. John Wiley and Sons Inc., New tion and stability analyses of yield and yield components York, USA of established and mutant rice genotypes tested in multi- Goufo P, Moutinho-pereira JM, Jorge TF, Correia CM (2017) ple locations in Malaysia. Acta Agric Scan 67:590–606. Cowpea (Vigna unguiculata L. Walp.) metabolomics : https:// doi. org/ 10. 1080/ 090647 10.2 017. 13211 38 osmoprotection as a physiological strategy for drought Olajide AA, Ilori CO (2017) Effects of drought on morphologi- stress resistance and improved yield. Front Plant Sci 8:1– cal traits in some cowpea genotypes by evaluating their 22. https:// doi. org/ 10.3 389/ fpls.2 017. 00586 combining abilities. Adv Agric 2017:1–10 Hall AE, Cisse N, Thiaw S et al (2003) Development of cow- Owade JO, Abong G, Okoth M, Mwang AW (2020) A review pea cultivars and germplasm by the Bean/Cowpea CRSP. of the contribution of cowpea leaves to food and nutrition Field Crop Res 82:103–134. https:// doi.o rg/1 0.1 016/ security in East Africa. Food Sci Nutr 8:36–47. https:// S0378-4 290(03) 00033-9 doi.o rg/1 0. 1002/f sn3. 1337 Hartley O (1950) The maximum F-ratio as a short-cut test for Pacheco A, Vargas M, Alvarado G, Rodriguez F, Crossa J, heterogeneity of variance. Biometrika 37:308–312 Burgueno J (2016) User’s Manual GEA-R (Genotype by IBPGR (1983) Cowpea descriptors, 1st edition. international Environment Analysis with R ). GEA-R 1–42 board for plant genetic resources, Rome, Italy Purchase JL, Hatting H, Van DCS (2000) Genotype × environ- SAS II (2013) Statistical Analysis System, Version 9.4 SAS ment interaction of winter wheat (Triticum aestivum L.) institute Inc. 476 in South Africa : I. AMMI analysis of yield performance. Jadhav S, Balakrishnan D, Shankar VG, Beerelli K, Chandu G, South Afr J Plant Soil 17:95–100. https:// doi.o rg/1 0. 1080/ Neelamraju S (2019) Genotype by environment interac- 02571 862.2 000. 106348 77 tion study on yield traits in different maturity groups of Simion T (2018) Adaptability performances of cowpea [Vigna rice. J Crop Sci Biotech 22:425–449. https://d oi. org/1 0. unguiculata ( L.) Walp ] genotypes in Ethiopia. Food Sci 1007/s 12892-0 18- 0082-0 Qual Manag 72:43–47 Kandus M, Almora D, Ronceros RB, Salerno JCS (2010) Simion T, Mohammed W, Fenta BA (2018) Genotype by Statistical models for evaluating the genotype-environ- environment interaction and stability analysis of cowpea ment interaction in Statistical models for evaluating the [Vigna unguiculata (L.) Walp] genotypes for yield in Ethi- genotype-environment interaction in maize (Zea mays opia. J Plant Breed Crop Sci 10:249–257. https:// doi.o rg/ L.) Modelos estadísticos para evaluar la interacción gen- 10.5 897/J PBCS2 018. 0753 otipo-ambiente en maíz (Zea mays). Phyton (B Aires) Singh BB, Ajeigbe HA, Tarawali SA et  al (2003) Improving 79:39–46 the production and utilization of cowpea as food and fod- der. Field Crop Res 84:169–177. https://d oi.o rg/ 10.1 016/ S0378- 4290(03) 00148-5 Vol:. (1234567890) 1 3 Euphytica (2022) 218:57 Page 13 of 13 57 Timko, M.P. and Singh BB (2008) Cowpea, a multifunctional GGE biplot. Crop Sci 40:597–605. https://d oi. org/ 10. legume. In: Paul, H. Moore and Ray M (ed) genomics of 2135/ crops ci2000. 40359 7x tropical crop plants, 1st edn. Springer Science+Business Yan W, Kang MS, Ma B, Woods S, Cornelius PL (2007) GGE Media, LLC, pp 227–258 biplot vs. AMMI analysis of genotype-by-environment Tumuhimbise R, Melis R, Shanahan P, Kawuki R (2014) Gen- data. Crop Sci 47:643–655. https:// doi.o rg/1 0.2 135/ crops otype × environment interaction effects on early fresh ci2006.0 6. 0374 storage root yield and related traits in cassava. Crop J Yaw E, Isaac O, Amegbor K, Mohammed H, Kusi F, Atpkle 2:329–337. https:// doi.o rg/ 10. 1016/j. cj.2 014.0 4. 008 I, Sie EK, Ishahku M, Zakaria M, Iddrisu S, Kendey HA, Weng Y, Jun Q, Eaton S, Yang Y, Ravelombola WS, Shi A Boukar O, Fatokun C, Nutsugah SK (2020) Genotype (2019) Evaluation of seed protein content in USDA cow- × environment interactions of yield of cowpea (Vigna pea germplasm. HortScience 54:814–817 unguiculata ( L.) Walp ) inbred lines in the Guinea and Yan W (2002) Singular-value partitioning in biplot analysis of Sudan Savanna ecologies of Ghana. J Crop Sci Biotechnol multi environment trial data. Agron J 94:990–996. https:// 23:453–460. https:// doi. org/1 0.1 007/s 12892-0 20- 00054-5 doi. org/ 10. 2134/ agron j2002. 0990 Zali H, Farshadfar E, Sabaghpour SH (2012) Evaluation of Yan W (2011) GGE Biplot vs. AMMI Graphs for genotype-by- genotype × environment interaction in chickpea using environment data analysis. Journal Indian Soc Agric Stat measures of stability from AMMI model. Ann Biol Res 65:183–193 3:3126–3136 Yan W (2016) Analysis and handling of G × E in a practical Zobel RW, Wright MJ, Gauch HG (1988) Statistical analysis of breeding program. Crop Sci 56:2106–2118. https:// doi. a yield trial. Agron J 80:388–393. https:// doi.o rg/ 10.2 134/ org/1 0. 2135/ crops ci2015. 06. 0336 agron j1988. 000219 62008 0000 30002x Yan W, Rajcan I (2002) Biplot Analysis of test sites and trait relations of soybean in Ontario. Crop Sci 42:11–20. Publisher’s Note Springer Nature remains neutral with regard https:// doi.o rg/1 0. 2135/ cropsc i2002. 0011 to jurisdictional claims in published maps and institutional Yan W, Tinker NA (2006) Biplot analysis of multi-environment affiliations. trial data: principles and applications. Can J Plant Sci 86:623–645. https:// doi. org/ 10. 4141/ P05- 169 Yan W, Hunt LA, Sheng Q, Szlavnics Z (2000) Cultivar evalu- ation and mega-environment investigation based on the Vol.: (0123456789) 1 3