African Journal of
Plant Science

  • Abbreviation: Afr. J. Plant Sci.
  • Language: English
  • ISSN: 1996-0824
  • DOI: 10.5897/AJPS
  • Start Year: 2007
  • Published Articles: 807

Full Length Research Paper

Normalized difference vegetation index as screening trait to complement visual selections of durum wheat drought tolerant genotypes

Alemayehu Zemede Lemma
  • Alemayehu Zemede Lemma
  • Ethiopian Institute of Agricultural Research (EIAR), Addis Ababa, Ethiopia.
  • Google Scholar
Firew Mekbib Hailemariam
  • Firew Mekbib Hailemariam
  • Department of Plant Sciences, Faculty of Agriculture and Environmental Sciences, Haramaya University, P. O. Box 138, Dire-Dawa, Ethiopia.
  • Google Scholar
Kebebew Assefa Abebe
  • Kebebew Assefa Abebe
  • Ethiopian Institute of Agricultural Research (EIAR), Addis Ababa, Ethiopia.
  • Google Scholar
Zewdie Bishaw
  • Zewdie Bishaw
  • International Center for Agricultural Research in Dry Area (ICARDA), P. O. Box 5689, Addis Ababa, Ethiopia.
  • Google Scholar

  •  Received: 12 May 2021
  •  Accepted: 01 September 2021
  •  Published: 31 January 2022


Normalized difference vegetation index (NDVI) is considered as a potential screening tool detecting stay green traits for drought tolerance. The present study aimed to evaluate durum wheat genotypes using NDVI under drought condition and investigate its association to grain yield. NDVI scores taken at different growth stages for 64 durum wheat genotypes were replicated twice in both sandy clay and clay textured soils situated at the same geographic location of Debre-Zeit Experimental Station, Ethiopia during 2016 rainy season; also, Green Seeker (Hand held) was used to measure the quantity of photosynthetically active pigments in leaves. Measurements of NDVI were made four to five times on sunny days between booting to physiological maturity. Analysis of variance showed significant variation among genotypes at 0.01% on NDVI values across all growth stages except at physiological maturity in both sandy clay and clay soils. The NDVI scores were highly significantly associated with yield at grain filling and anthesis stages in both soil environments. Overall, it is possible to suggest that use of NDVI would help complement identification of drought tolerant genotypes on durum wheat.


Key words: Grain yield, Normalized difference vegetation index (NDVI), Triticum durum.


Wheat is the fourth most important cereal crop after maize, tef and sorghum in terms of area coverage and production in Ethiopia. Both bread and durum are grown extensively in the country, although separate area coverage and production is not known. They are cultivated over an area of 1.69 million hectares with annual production of about 4.5 million tons (CSA, 2018). However, average productivity in the country is low (2704 kg/ha). Drought stress globally, under rain-fed based crop production system of Ethiopia is one of the largest causes of wheat yield reduction. Drought has particularly become a common problem in both lowland environment associated with early termination of rainfall and highland vertisols environment which is waterlogged and does not follow optimum planting time which usually results in late planting. Breeding and selection of genotypes for drought tolerance using grain yield alone has a number of constraints. Low heritability and genetic variance are among the major challenges while using grain yield for identification of genotypes under drought stress (Manal, 2009). High genotype × environment is also another challenge for selection of the right genotypes under drought stress (Sangam et al., 2020).


The use of NDVI to measure spectral traits associated with stay green for crop variety evaluation complementing visual selection is suggested to be the simplest and efficient tool for speeding up the selection cycle by reducing the time spent to evaluate germplasm under water stress condition and select genotypes earlier in the season or evaluate large number of genotypes during each round of selection and grain yield prediction before harvesting (Sanchez et al., 2013; Sruthi and Mohammed, 2015). Furthermore, estimating agronomic traits such as yield or drought indirectly using spectral traits measured proximally with handheld devices provides alternative solution to managing the constraints of phenotyping (Araus and Cairns, 2014).


Normalized difference vegetation index (NDVI) has been studied widely on a range of applications on drought (Seyoum et al., 2016), heat stress (Shanahan et al., 2001; Cao et al., 2019) yield prediction and nitrogen management (William et al., 2001; Syeda et al., 2014; Son et al., 2014; Quemada et al., 2019; Hassan et al., 2019), non-destructive biomass estimation (Semeraro, 2019); disease management, (Sheedy and Thompson, 2009) on root lesion nematode and Mahlein et al., (2019) on fusarium head blight. According to Tardieu (2012), genotypes which maintain consistent and high NDVI values across different stages performed better under drought than those with low values, which ultimately resulted in higher biomass and yield under severe to moderate stress. Based on the existing knowledge, the question related to NDVI in breeding to drought could be to what extent the genotypes vary and traits can be associated with the final yield while using this selection tools for early variety selection.


The present study aimed to evaluate and identify durum wheat genotypes using normalized difference vegetation index under drought condition in two soil types of Debre-Zeit Experimental Station and determine its association to grain yield at different growth stages.


Sixty-four durum wheat genotypes of different origin were used in the study. Twenty-one of the genotypes were released cultivars, 21 of the materials were introduced breeding lines obtained from Centro Internacional de mejoramiento de maiz y Trigo (CIMMYT) and the remaining 14 materials ones were landraces (Table 1).


Site description


The genotypes were grown in the field during the main rainy season in 2016 at Debre-Zeit experimental station which is located in 8° 41’ 36” latitude and 39° 03’ 17’ longitude with altitude 1880 m.a.s.l. Soil data analysis were carried out at Agricultural and Nutritional Research Laboratory of Debre-Zeit Agricultural Research Center (MOALR, 2018) and the physical and chemical properties of the soils are presented in Table 2


Experimental design and field management


The genotypes were planted in 8 x 8 simple lattice design field experiments with two replications on plots consisting of two rows 2 m in length and 20 cm between rows spacing. The genotypes were grown late in the season in sandy clay on August 1, 2016 and in clay soil on July 31, and were exposed to terminal moisture stresses uniformly from the date of anthesis which represent lowland wheat growing environment of Ethiopia. The planting date was selected based on the consistency of the incidence of drought and targeted timing to coincide at anthesis stage for the majority of the genotypes included in the study.


In the experiment, 100 kg of Di-ammonium Phosphate was applied at planting and split application of 100 kg of urea where one-third was applied at planting and two-third at time of tillering. The plots were hand weeded and a fungicide, tilt-250 with rate of 150 ml/ac was applied twice in the season to protect the genotypes from stem rust infection.


Data collection


Five measurements of NDVI were recorded at different growth stages of the plants. The measurement were carried out on a fine sunny, wind free days on plant surface, holding the sensor of the handheld green seeker 50 cm above the stand (Govaerts and Verhulst, 2010). The measurements were taken during booting, heading, anthesis, grain filling and physiological maturity (Table 3). Yield data were recorded on plot basis after drying and threshing.


Data analysis


Analysis of variance was performed using R-4.2 and linear correlations were employed to indicate associations between the parameters Minitab version-16.



Analysis of variance


Analysis of variance showed significant variation among genotypes at 0.01% on grain yield and normalized difference vegetation index at all growing stages except at physiological maturity in sandy clay soils (Table 4) indicating that the genotypes respond differently to spectral reflectance. The results also exhibited a large difference on NDVI scores at different growth stages. For instance, the average NDVI record in sandy clay soil was 0.76, 0.63, 0.52, 0.37 and 0.16 at booting, heading, anthesis, grain filling and maturity, respectively (Table 4). In clay soil, the analysis of variance revealed that significant variation among genotypes at 0.01% on NDVI was recorded at different growth stages except at physiological maturity and grain yield. The NDVI recorded was 0.76 at booting, 0.70 at heading, 0.64 at anthesis, 0.52 at grain filling and 0.15 at physiological maturity (Table 5). The finding of this study suggested that the genotypes ability to absorb the reflectance is reduced due to senescence. Several authors (Syeda et al., 2014; Salah et al., 2015; Nasir et al., 2020) reported that NDVI scores reduced from the vegetative to reproductive stages and dropped to lower levels at physiological maturity.



The trends were similar in both soil types although the average values at each developmental stage in light textured soil was low and showed sharp reduction (Figure 1), indicating the existence of soil moisture variations in maintaining and supplying different soil types to the plant. Both sandy clay and clay soil showed a significant effect on the development of photosynthetically active leaf area. Plants raised under sandy clay soils had a mean NDVI value of 0.49, while that grown under clay soil was 0.55. Although the differences were not visible under visual evaluations when the measurements were carried out at booting and 50% heading, the records showed a significant reduction from booting to heading with mean value of 0.76 and 0.63 in sandy clay soil. The trend was similar and decreased from 0.76 at booting to average value of 0.70 at heading in clay soil. This indicates that there would be an effect on photosynthetically active leaf area. Likewise, NDVI recorded 10 days after heading showed a similar trend on the reduction of green area compared to the earlier stages in both soil textures.



The genotypes under sandy clay soil again showed low NDVI value (0.49) compared to clay pelli-vertisols (0.55) across different growth stages. The reduction of NDVI value continued in similar trend when the plants reached grain filling stage ranging from 0.27 to 0.51 and 0.35 to 0.66 in sandy clay and clay vertisols, respectively.


NDVI recorded at different growth stages in two soil textures are illustrated in Figure 1. NDVI value reached a maximum prior to reproductive stages, and a slight decreasing tendency was observed at heading and further slight reduction at anthesis. The decreasing trend in NDVI values was similar and sharp from grain filling stage where maximum variation between genotypes in NDVI scores was observed (Syeda et al., 2014). A soil variation on the values of NDVI was minimized at the time when genotypes reached physiological maturity.


Correlation analysis


Correlation of NDVI traits with performance traits with yield at sandy clay and clay are presented in Table 6. The correlations were weaker (0.33-0.37) in clay soils which were less favorable for grain yield compared to sandy clay soils (0.37-0.42). The association was medium to high strength indicating that there is a significant relationship between NDVI values and the yield. The correlation was smallest at booting stage, increased at heading and anthesis and reached maximum during grain filling stage in sandy clay soil. In clay soil, the correlation was small and non-significant at heading, became significant with a slight strength at anthesis, and had relatively strong association at grain filling stage. It also clearly showed that NDVI scores highly significantly and consistently associated with yield were at grain filling stage in both soil environments, indicating that early selection of better yielding genotypes of wheat under terminal moisture can be identified. Similar finding was reported by Son et al. (2014) that grain filling stages are the right stage to get maximum variability and association with yield could be obtained. NDVI at maturity showed a non-significant and very weak correlation under both soil textures. This could be due to the severity of moisture stress which resulted in the reduction of the NDVI values associated to loosing leaf green pigment, yellowing and presence of low variation between genotypes at maturity in staying green trait. Mean NDVI values generally reached peak at booting; however, values during grain filling period were found to be the most correlated to the final grain yield in both soils, suggesting that it is the right stage for making earlier selections and estimating grain yield. Similar studies on NDVI suggested that grain filling stages display the maximum potentials for estimating yield (Spitko et al., 2016; Shanahan et al., 2001). Similarly, Nieves et al. (2000) indicated that under rain-fed conditions, the spectral reflectance indices measured at any crop stage were positively correlated (P < 0.05) with yield.



The results of the study showed that the durum wheat genotypes varied significantly based on NDVI scores from booting to grain stages at both sandy clay and clay soil types but showed limitation on differentiation of the genotypes at physiological maturity. The results also revealed that normalized difference vegetation index (NDVI) is significantly correlated with grain yield under stress conditions. The grain filling stage was found to be the most appropriate stage to evaluate the genotypes under sandy clay and clay soils using NDVI as a screening tool. Although the correlation values between grain yield and NDVI records at different stages were relatively very strong, it is evident that integrating NDVI in the process will improve the rate of genetic progress and better complement the breeding process or screening of genotypes under drought. NDVI allowed better yielding lines to be identified; for instance, the genotypes 29, 30, 31, 21 and 16 were among the top yielding lines and showed consistently high NDVI values at different growth stages in both types of soils.


The authors have not declared any conflict of interests.


The authors thank the Ethiopian Institute of Agricultural Research (EIAR), Agricultural Growth Program II, (AGP II) and International Center for Agricultural Research Institute (ICARDA) for providing technical and financial supports for the execution of the field experiment. The authors also appreciate the support provided by Wondo-Genet Research Center and Debre-Zeit Research Center. Research and technical staff members of the crop research process of Debre-Zeit Research Center are highly acknowledged for assisting us on field activities.


Alemayehu Z, Firew M, Kebebew A, Zewdie B (2019). Variability in Ethiopian Durum Wheat under Rain-fed Environment Subjected to Drought at Anthesis. Ethiopian Journal Agricultural Sciences 29(2)17-29.


Araus L, Cairns JE (2014). Field high-throughput phenotyping: the new crop breeding frontier. Trends Plant Science 19(1):52-61.


Cao Z, Yao X, Liu H, Cheng T, Tian Y, Cao W, Zhu Y (2019). Comparison of the abilities of vegetation indices and photosynthetic parameters to detect heat stress in wheat. Agricultural and Forest Metrology 265:121-136.


Central Statistical Agency (CSA) (2018). "Agricultural Sample Survey 2016/17, Area and Production for Major crops. Private Peasant Holdings," Vol. I, Statistical Bulletin No 584 (2017), Addis Ababa, Ethiopia.


Govaerts B, Verhulst N (2010). The Normalized Difference Vegetation Index (NDVI) Greenseeker (TM) Handheld Sensor: Toward the Integrated Evaluation of Crop Management. Part A - Concepts and Case Studies. Mexico, CIMMYT pp. 1-16.


Hassan MA, Yang M, Rasheed A, Yang G, Reynolds M (2019). A rapid monitoring of NDVI across the wheat growth cycle for grain yield prediction using a multi-spectral UAV platform. Plant Sciences 282:95-103.


Mahlein AK, Alisaac E , Masri AA, Behmann J, Dehne HW, Oerke EC (2019).Comparison and Combination of Thermal, Fluorescence, and Hyper spectral Imaging for Monitoring Fusarium Head Blight of Wheat on Spikelet Scale. Sensors 19:2281.


Manal H (2009). Estimation of heritability and genetic advance of yield traits in wheat (Triticum aestivum L.) under drought condition. International Journal of Genetics and Molecular Biology 1(7):115-120.


Ministry of Agriculture and Livestock Resource, MoALR (2017). Crop Variety Register, Issue No. 20, Addis Ababa, Ethiopia.


Nasir MA, Raj K, Longchamps L, Subash D (2020). Using NDVI to Differentiate Wheat Genotypes Productivity Under Dry land and Irrigated Conditions. Remote Sense 2(5):824.


Nieves A, Dolors V, Jaume C, Casadesus J, Luis A, Roy C (2000). Spectral Vegetation Indices as Non-destructive Tools for Determining Durum Wheat Yield. Agronomy Journal agro-climatology 92(1):83-91.


Quemada M, Pancorbo JL, Ayonso M, Gabriel JL, Lopez-Herreza J, Perez Martin E (2019). Vegetation indices from remote sensing imagery as proxies for yield and grain N in wheat. In: Precision Agriculture 19, John V. Staffford Published P 1030


Salah E, Al-Suhaibani N, El-Azeem AS, Rehman UR, Schmidhalter URS (2015). Spectral reflectance indices as a rapid and nondestructive phenotyping tool for estimating different morpho-physiological traits of contrasting spring wheat germplasms under arid conditions. Turkish Journal of Agriculture and Forestry 39:1-11.


Sangam LD, Irwin G, Ceccarelli S, Rodomiro O (2020). Advanced Analytics, Phenomics and Biotechnology Approaches to Enhance Genetic Gains in Plant Breeding: in Donald Sparks Advances in Agronomy, 1st edition.


Sanchez A, Michael A, Gore J, Heun T, Thorp KR, Elizabete Carmo-Silva A, Andrew NF, Michael E, White JW (2013). Development and evaluation of a field based high-throughput phenotyping platform. Functional Plant Biology 41(1): 68-79.


Semeraro T, Mastroleo AG, Pomes A, Luvisi E, Gissi Aretano R (2019). Modeling fuzzy combination of remote sensing vegetation index for durum wheat crop analysis. Computers and Elecrtonics in Agriculture 156:684-692.


Seyoum M, Getachew B, Suryabhagavan KV (2016). Evaluation of Vegetation Indices for Agricultural Drought Monitoring in East Amhara, Ethiopia. International Journal of Scientific Research 5(1):535-540.


Shanahan J, Schepers F, James S, Francis D, Varvel D, Gary E, Wilhelm W, Tringe JM, Schlemmer MR, Major DJ (2001). "Use of Remote-Sensing Imagery to Estimate Corn Grain Yield Agronomy and Horticulture. Faculty Publications.9.


Sheedy JG, Thompson JP (2009). Resistance to the root-lesion nematode pratylenchusthornei of Iranian landrace wheat. Australian plant pathology 38(5):478-489.


Son NT, Chen CF, Minh VQ, Trung NH (2014). A comparative analysis of multi-temporal MODIS EVI and NDVI data for large-scale rice yield estimation. Agricultural and Forest Metrology 197:52-64.


Spitko T, Nagy Z, Subori TZ, Szoke TC, Berzy T, Pinter J, Marton CL (2016). Connection between normalized difference vegetation index and yield in maize. Plant Soil Environment 62(7):293-298.


Sruthi S, Mohammed MA (2015).Agricultural Drought Analysis Using the NDVI and Land Surface Temperature Data; a Case Study of Raichur District. Available online at www. Aquatic Procedia 4:1258-1264.


Syeda R, Amjed SA, Ashfaq A, Muhammad M, Zia-Ul-Haq M, Shakeel A, Ercisli S, Jaafar ZE (2014). Normalized Differences Vegetation Index as a Tool for Wheat Yield Estimation: A Case Study from Faisalabad, Pakistan. Scienti?c World Journal 8 p.


Tardieu F (2012). Any trait or trait-related allele can confer drought tolerance: just design the right drought scenario. Journal of Experimental Botany 63(1):25-31.


William RR, John BS, Gordon VJ, Marvin LS, Erna VL, Wade E (2001). In: Thomason, and James S. Schepers In-Season Prediction of Potential Grain Yield in Winter Wheat Using Canopy Reflectance. Agronomy Journal 93:131-138.