The use of principal component analysis (PCA) in the selection of bread wheat (Triticum aestivum L.) genotypes under kahramanmaraş conditions

Introduction

Bread wheat, which has strategic importance with a 7.1 million hectare cultivation area and an annual production of 22 million tons in Turkey, is one of the most important cultivated crops forming the foundation of food security (Kahraman et al., 2021) [1] . Evaluated on a regional basis, the Eastern Mediterranean Region continues to be a strategic center, accounting for approximately 9.47% of Turkey's total wheat cultivation area and 12.12% of its production (Ezici and Hızlı, 2024) [2] . The pressure on agricultural productivity due to the growing global population and climate change makes it essential to obtain higher yields per unit area from existing production areas (Çağlar et al., 2011; Yüce et al., 2022) [3,22] . Towards this goal, identifying genotypes with high yield potential and strong adaptation capabilities to environmental stress conditions in breeding programs constitutes a critical stage for sustainable production (Kurt and Yağdı, 2013) [4] . Grain yield stands out as a multi-factorial trait determined by a complex interaction of genetic capacity, plant morphology, and environmental factors (Bilgin and Korkut, 2014) [5] . In this context, analyzing the complex relationships among yield components using multivariate statistical methods directly affects the success rate of breeders in the selection process (Sönmez and Olgun, 2020) [6] . In particular, Principal Component Analysis (PCA) and biplot techniques allow for the visualization of large amounts of agricultural data and reduce the complexity of relationships among genotypes, enabling the development of a clearer selection strategy (Bayram et al., 2017; Güngör et al., 2022) [7,8] . Visualizing the effects of the investigated traits on genotypic profiles, these methods provide a critical mechanism that supports decision-making processes in selecting lines demonstrating high yield potential among breeding materials (Akçura, 2011; Demirel et al., 2021) [9, 10] . Various studies conducting genotypic evaluations aim to explain the variation in yield components and identify lines sensitive to genotype-environment interactions (Aktaş et al., 2017; Kadıoğlu et al., 2024) [11,12] . The data obtained through such studies reveal the fundamental parameters required to accelerate modern breeding strategies by selecting highyielding genotypes adapted to the regional conditions where the studies are conducted (Bayhan, 2022; Yorulmaz et al., 2022) [13,14] . On the other hand, the application of multivariate statistical approaches aims to increase selection success in breeding programs by more accurately revealing the relationships between the yield capacity and morphological features of genotypes (Kendal, 2026; Stojšin et al., 2026) [15,16] . As frequently emphasized in the international literature, advanced statistical models establish a reliable framework for identifying the most stable and high-yielding genotypes under different environmental conditions by clarifying complex interactions among variables during the selection process (Kendal, 2026; Stojšin et al., 2026) [15] . Elevating the multifarious relationships between yield and yield components of the investigated genotypes provides breeders with a crucial roadmap in selecting lines with high regional adaptation. Such multivariate approaches enhance the efficiency and selection precision of breeding programs by more accurately defining genotypic variation (Janmohammadi et al., 2014; Ali et al., 2020) [17] . In this study, it was aimed to evaluate the agricultural traits of 50 different bread wheat genotypes through multivariate analyses; it was anticipated that the findings would identify the lines most suitable for regional conditions and transfer them to future generation breeding studies.

Materials and Methods

Plant Material and Experimental Design

In the study, 1 registered bread wheat cultivar (Germenicia) widely grown in the region and 49 advanced bread wheat lines originating from CIMMYT (International Maize and Wheat Improvement Center, Mexico) were used as plant material (Table 1). The field experiment was established during the 2024-2025 growing season at the experimental area of the Eastern Mediterranean Transitional Zone Agricultural Research Institute in Kahramanmaraş, Turkey, using a Randomized Complete Block Design (RCBD). Following land preparation, sowing was carried out with a seed drill at the optimal seeding rate; fertilization and agricultural maintenance practices suitable for local cultivation standards were fully applied throughout the growing season.

Investigated traits

In the study, the following traits were measured in the plots reaching harvest maturity (Çölkesen et al., 1999) [18]: Plant height (cm): In 10 randomly selected plants from each plot, the length from the soil surface to the tip of the spike (excluding awns) was measured and averaged.

Measured with a ruler on 10 random spikes, and the average was calculated.

Number of spikelets (number/spike):

The number of spikelets present on the main stem spike was visually counted and averaged.

Number of kernels per spike (number): Kernels from 10 spikes were counted, and their averages were taken.

Grain weight per Spike (g): Kernels from 10 randomly selected spikes were individually weighed, and the average was calculated.

Grain yield (kg/da): Calculated by converting the harvested yield of the experimental plots to yield per decare (1 decare = 1000 m²).

Statistical analyses

Descriptive statistics of the raw data obtained from the research were calculated, and Pearson correlation coefficients were determined to identify the interactions among the variables. Principal Component Analysis (PCA) was performed to optimize the variance matrix and model the twodimensional distribution of the investigated traits and the 50 genotypes. Within the scope of this analysis, factor loadings, eigenvalues, and variance explanation ratios were determined.

Results and Discussion

Descriptive statistics

The minimum, maximum, and overall means of the parameters belonging to the 50 investigated wheat genotypes were calculated. The ranges of variation were determined as 84.00-106.33 cm (mean 95.46 cm) for plant height, 8.78-12.88 cm (mean 11.03 cm) for spike length, 16.0-22.0 (mean 18.90) for the number of spikelets, 30.80-91.40 (mean 53.64) for the number of kernels per spike, and 1.50-2.81 g (mean 2.07 g) for grain weight per spike (Table 2). The average grain yield was 523.38 kg/da, and the yield potential among genotypes varied between 416.42 and 632.49 kg/da, presenting a very broad genetic variation for selection. Line 208, representing the institute's working materials, exhibited a remarkable potential by reaching a capacity of 91.4 kernels per spike. While the grain yield findings obtained in the study demonstrate parallelism with the findings of Yüce et al. (2022) [3] and Ezici and Hızlı (2024) [2] conducted in the same ecology, the high variation we obtained in terms of spike characteristics proves that the genetic material has a rich foundation for selection.

Correlations among traits

According to the results of the Pearson correlation analysis among the investigated traits (Table 3), spike characteristics (spike length, number of spikelets, number of kernels per spike, and grain weight per spike) showed positive and highly significant relationships among themselves (ranging from r= 0.478 to 0.683). On the other hand, the direct relationship between spike characteristics and grain yield per unit area was negative and non-significant for spike length and number of spikelets (r=-0.135 and r=-0.048, respectively), and positive but non-significant for the number of kernels per spike and grain weight per spike (r=0.136 and r=0.174, respectively). The parameter showing a positive and significant correlation with grain yield was plant height (r=0.416). The fact that the relationships between plant height and spike characteristics remained statistically no significant indicates that height increase does not exert a negative pressure on spike architecture; conversely, the positive and significant correlation (r=0.416) between plant height and grain yield proves that lines with strong vegetative habitus (due to stem reserves) have an advantage under terminal stress conditions in the Mediterranean/Transitional zone ecology. Although modern wheat breeding generally targets short stature to prevent lodging, the direct correlation of yield with plant height rather than spike size reflects the stress-based physiological dynamics of the Eastern Mediterranean transitional zone ecology. In Mediterranean climate locations subjected to drought or heat stress during the grain filling period, it is known that taller genotypes with strong vegetative habitus accumulate more water-soluble carbohydrates (WSC) in their stems (del Pozo et al., 2016) [19] . As emphasized in similar studies in the literature, taller genotypes rapidly translocate these reserves accumulated in their stems to the grain during terminal stress when photosynthesis is suppressed, creating a physiological shield and preserving their yield per unit area (Bainsla et al., 2020) [20] . This situation explains why, in the study, the highest yield was obtained not from the largest spiked genotypes, but from vigorous lines that could maximize plant height development.

Principal Component Analysis (PCA) and Classification of Genotypes

In the PCA test, applied to reduce dimensions and explain traits and genotypes, two "Principal Components" (F1 and F2) with eigenvalues greater than 1 were obtained, and these two components explained 71.35% of the total variance (Table 4, Figure 1). Weight per Spike (0.827), and Spike Length (0.792), respectively (Table 5). Because the loadings belonged entirely to spike architecture, this axis was designated as the "Spike Yield Components Axis" in the study (Figure 2). 2021) [7,1] . [16] , and Çokkızgın et al. (2026) [23] , which reported that genotypic variance in bread wheat primarily derives from spike architecture. The standard cultivar Germenicia (yield 483.47 kg/da) lagged behind most lines, remaining in the minus (negative) region on both component axes. This confirms that the new breeding lines are much more promising than the current standards.

Conclusion

The fundamental philosophy in wheat breeding is to translate the plant's genetic performance into maximum yield, even under challenging environmental conditions. As a result of this research conducted with 50 different bread wheat genotypes at the Eastern Mediterranean Transitional Zone Agricultural Research Institute, it was determined that the advanced lines used harbor high genetic potential/possess a rich genotypic variation in terms of plant habitus and spike architecture.

In light of the analyses, the following conclusions have been reached for future breeding and registration strategies: in genetic studies to be conducted to enlarge spike physiology and increase kernel-holding capacity per spike to record levels in bread wheat, lines 208 and 209 should absolutely be utilized as parents (donors). Lines 244, 225, 233, and 207, which adapted well to the region's ecology and directly increased the final grain yield and vegetative vigor in the field, can be said to be promising for introduction to Regional/National agriculture.

References

1. Kahraman T, Güngör H, Öztürk İ, Yüce İ, Dumlupınar Z. Ekmeklik buğday (Triticum aestivum L.) genotiplerinde genotip ve çevrenin tane verimi ve bazı kalite özellikleri üzerine etkisinin temel bileşen ve GGE biplot analizleri ile değerlendirilmesi. Tarim ve Doga Dergisi. 2021;24(5):992.
2. Ezici AA, Hızlı H. Doğu Akdeniz Bölgesinde Bazı İleri Kademe Ekmeklik Buğday (Triticum aestivum L.) İleri Hatlarının Verim Performanslarında Parametrik Stabilite Analizi ve Genetik Parametrelerinin Belirlenmesi. International Journal of Eastern Mediterranean Agricultural Research. 2024;7(1):22-35.
3. Yüce İ, Başkonuş T, Dokuyucu T, Akkaya A, Güngör H, Dumlupınar Z. Evaluation of quality and some agronomic traits of bread wheat (Triticum aestivum L.) lines and cultivars under Kahramanmaraş ecological conditions. Yuzuncu Yıl University Journal of Agricultural Sciences. 2022;32(2):362-371.
4. Kurt Ö, Yağdı K. Bazı ileri ekmeklik buğday (Triticum aestivum L.) hatlarının Bursa koşullarında verim özellikleri yönünden performansının araştırılması. Uludağ Üniversitesi Ziraat Fakültesi Dergisi. 2013;27(2):19-32.
5. Bilgin O, Korkut KZ. Bazı Ekmeklik Buğday (Triticum aestivum L.) Çeşit ve Hatlarının Tane Verimi ve Bazı Fenolojik Özelliklerinin Belirlenmesi. Tekirdağ Ziraat Fakültesi Dergisi. 2014;2(1):58-65.
6. Sonmez AC, Olgun M. Ekmeklik Buğdayda (Triticum aestivum L.) Farklı Ekim Sıklıklarının Verim ve Verim Bileşenleri Üzerine Etkisinin Belirlenmesi. Turkish Journal of Agriculture-Food Science and Technology. 2020;8(2):380-386.
7. Bayram S, Öztürk A, Aydın M. Ekmeklik buğday genotiplerinin Erzurum koşullarında tane verimi ve verim unsurları yönüdn değerlendirilmesi. Yüzüncü Yıl Üniversitesi Tarım Bilimleri Dergisi. 2017;27(4):569-579.
8. Güngör H, Çakır MF, Dumlupınar Z. İleri ekmeklik buğday (Triticum aestivum L.) hatlarının verim, verim unsuru ve kalite özellikleri bakımından değerlendirilmesi. Avrupa Bilim ve Teknoloji Dergisi. 2022;(35):123-127.
9. Akçura M. The relationships of some traits in Turkish winter bread wheat landraces. Turkish Journal of Agriculture and Forestry. 2011;35(2):115-125.
10. Demirel F, Kumlay AM, Yıldırım B. Bazı ekmeklik buğday (Triticum aestivum l.) genotiplerinin agromorfolojik özellikleri bakımından biplot, kümeleme ve path analizi yöntemleri ile değerlendirilmesi. Avrupa Bilim ve Teknoloji Dergisi. 2021;(23):304-311.
11. Aktaş H, Erdemci İ, Karaman M, Kendal E, Tekdal S. Bazı kışlık ekmeklik buğday genotiplerinin tane verimi ve bazı kalite özellikleri bakımından GGE biplot analiz yöntemi ile değerlendirilmesi. Türk Doğa ve Fen Dergisi. 2017;6(1):43-51.
12. Kadıoğlu S, Kadıoğlu B, Taşğın G, Taşcı R. TAGEM Buğday Çeşitlerinin Benimsenme ve Kullanımının İncelenmesi (Ağrı-Erzurum ve Muş İlleri Örneği). Wheat Studies. 2024;13(1):26-36.
13. Bayhan M. GGE Biplot analiz yöntemi ile makarnalık buğday (Triticum durum Desf.) genotiplerinde kalite özelliklerinin değerlendirilmesi. ISPEC Journal of Agricultural Sciences. 2022;6(3):573-582.
14. Yorulmaz L, Bayhan M, Özkan R, Albayrak Ö, Yıldırım M, Öner M. Kurak koşullarda bazı ekmeklik buğday genotiplerinin performanslarının GGE Biplot Analizi Yöntemi ile değerlendirilmesi. Artvin Çoruh Üniversitesi Orman Fakültesi Dergisi. 2022;23(2):88-95.
15. Kendal E. The combinations of genotype, trait and genotype by yield, trait in triticale using biplot methods for define adaptation of varieties to environment. Mustafa Kemal Üniversitesi Tarım Bilimleri Dergisi. 2026;31(1):127-136.
16. Stojšin M, Zečević V, Krstić Đ, Mačkić K, Matković Stojšin M, Bratković K, Knežević D. Multivariate assessment of wheat genotype response to environmental variation. Applied Ecology and Environmental Research. 2026;24(1):315-330.
17. Janmohammadi M, Movahedi Z, Sabaghnia N. Multivariate statistical analysis of some traits of bread wheat for breeding under rainfed conditions. 2014.
18. Çölkesen M, Cesurer L, Yürürdurmaz C, Demirbağ V, Çiçek A, Başgül A, et al. Kahramanmaraş koşullarına uygun yüssek verimli arpa çeşitlerinin belirlenmesi. Türkiye. 1999;3:15-18.
19. Del Pozo A, Yáñez A, Matus IA, Tapia G, Castillo D, Sanchez-Jardón L, et al. Physiological traits associated with wheat yield potential and performance under waterstress in a Mediterranean environment. Frontiers in plant science. 2016;7:987.
20. Bainsla NK, Yadav R, Singh GP, Sharma RK. Additive genetic behavior of stem solidness in wheat (Triticum aestivum L.). Scientific reports. 2020;10(1):7336.
21. Ali N, Hussain I, Ali S, Khan NU, Hussain I. Multivariate analysis for various quantitative traits in wheat advanced lines. Saudi Journal of Biological Sciences. 2021;28(1):347-352.
22. Çağlar Ö, Öztürk A, Bulut S. Bazı ekmeklik buğday çeşitlerinin Erzurum ovası koşullarına adaptasyonu. Atatürk Üniversitesi Ziraat Fakültesi Dergisi. 2006;37(1):1-7.
23. Euroasia Matematik, Mühendislik, Doğa ve Tıp Bilimleri Dergisi Medical Sciences. 2026;13(2):258-267.