Response of yield and quality of wheat with different gluten content to altitude above sea level and nitrogen treatment Authors: Yujiao Wang, Xuhong Chang, Demei Wang, Yushuang Yang, Yanjie Wang, Xiwei Liu, Shubing Shi, Guangcai Zhao Published: 2026 ABSTRACT Wheat (Triticum aestivum L.) yield and processing quality varied across environments, especially along altitudinal gradients. This study evaluated the combined effects of altitude above sea level and nitrogen (N) topdressing on yield and quality traits of winter wheat with strong, medium and weak gluten content. Six cultivars were tested for two growing seasons at Renqiu (4 m), Beijing (40 m) and Lhasa (3688 m). Raw-data verification showed that Lhasa reduced mean grain yield by 44.1% compared with Beijing, mainly through reductions in spike number, grains per spike and thousand-grain weight. High altitude increased total grain protein by 17.5% compared with Beijing, especially through increases in albumin and globulin fractions, but this increase did not translate into better processing quality. Strong and medium gluten wheat showed reduced gluten index, shorter dough stability time and higher weakening degree at Lhasa, whereas weak gluten wheat maintained relatively better stability under high-altitude stress. Nitrogen responses differed among gluten types: strong gluten wheat benefited more from 240 kg/ha N, medium gluten wheat performed similarly under 180–240 kg/ha N, and weak gluten wheat showed a more stable response under high-altitude conditions. Multivariate analyses separated the Lhasa environment from Renqiu and Beijing and indicated a clear yield-quality trade-off. These results suggested that cultivar selection and N management should be adjusted jointly according to altitude above sea level and gluten type. Introduction Wheat (Triticum aestivum L.) is one of the world’s major staple crops and provides dietary energy, protein, fibre and mineral nutrients1. Demand for wheat with stable processing quality has increased with the development of the flour and food-processing industries2. Under population growth, limited arable land and climate variability, improving both yield and quality through suitable cultivar deployment and field management has become an important issue for food security3–5. Altitude above sea level integrates multiple environmental gradients, including temperature, radiation, precipitation, atmospheric pressure and soil properties, and therefore affects wheat physiology, yield formation and grain quality6,7. Wheat cultivars with different gluten content may respond differently to such environmental changes because of differences in protein composition, starch properties and end-use requirements8,9. However, previous studies have usually focused on a single environmental factor or a single gluten type. A systematic evaluation of altitude, N management and gluten type is still needed. Altitude above sea level is a useful indicator of environmental heterogeneity in mountainous and plateau regions. With increasing altitude, air pressure decreases, solar radiation usually increases, diurnal temperature range becomes wider, and effective accumulated temperature often decreases. These changes can affect wheat growth and yield formation directly or indirectly10,11. Wheat yield is determined mainly by spike number per unit area, grains per spike and thousand-grain weight, all of which are sensitive to temperature and radiation during vegetative and reproductive growth12,13. Low temperature around jointing can suppress tiller development and reduce effective spike number14, whereas adequate accumulated temperature and assimilate supply before and after flowering support grain filling and thousand-grain weight15. Consequently, the yield response to altitude is not simply linear, and it may differ among gluten types because their phenology and source-sink relationships differ. Grain quality depends on protein concentration, protein composition, gluten strength and starch characteristics, which are jointly regulated by genotype and environment. Altitude above sea level can influence these traits through temperature, radiation and water conditions. Temperature affects assimilate transport, metabolic enzyme activity, cell division and N uptake, thereby influencing N accumulation and grain protein formation16,17. Radiation and sunshine duration regulate photosynthesis and the supply of carbohydrates and proteins to the developing grain18. Water status can also alter gluten content and dough strength19. Previous studies reported that grain protein and lysine contents varied across altitudinal ranges20, suggesting that wheat quality responses to altitude may be strongly genotype dependent. Strong gluten wheat, medium gluten wheat and weak gluten wheat differ in breeding objectives, end use and cultivation requirements. Protein quality formation is especially sensitive to N supply and temperature during grain filling21. Low-temperature events can reduce protein and starch accumulation and alter grain quality formation22. Previous studies have also shown that altitude-related stress affects wheat yield, quality and metabolite profiles23. Therefore, different gluten types may have different adaptive ranges across altitudinal environments. The key knowledge gap is how altitude above sea level, N topdressing and gluten type interact to regulate yield and quality in a unified field experiment. This study aimed to: (i) quantify yield and quality responses of six winter wheat cultivars representing strong gluten wheat (S-W), medium gluten wheat (M-W) and weak gluten wheat (W-W) under two N topdressing rates across three altitudes above sea level over two growing seasons; (ii) identify the main environmental and management factors associated with yield and quality variation; and (iii) clarify how altitude, gluten type and N management jointly affected the yield-quality trade-off. The results provide a basis for regional cultivar deployment and precision N management in wheat production across complex terrains. Materials and methods Plant material Six winter wheat cultivars were used. Strong gluten wheat (S-W) included Zhongmai 578 (ZM-578) and Zhongmai 5051 (ZM-5051), medium gluten wheat (M-W) included Zhongmai 82 (ZM-82) and Zhongmai 89 (ZM-89), and weak gluten wheat (W-W) included Yangmai 15 (YM-15) and Yangmai 24 (YM-24). Seeds were obtained from the wheat breeding and experimental materials maintained by the Institute of Crop Sciences, Chinese Academy of Agricultural Sciences and the collaborating regional experimental stations. Gluten-type classification followed the cultivar release information and the Chinese wheat quality classification criteria, which use grain protein content, wet gluten content and dough properties to distinguish strong, medium and weak gluten wheat. Experimental site Beijing site: Located at low altitude, characterized by a temperate continental climate with dense vegetation and urban infrastructure in the background, providing optimal growing conditions for wheat maturation.Lhasa site: Situated at high altitude (3688 m), featuring a plateau environment with thin air, strong solar radiation, and distant mountain ranges visible; these harsh conditions significantly constrain crop growth duration and biomass accumulation.Seedling stage: Illustrates the uniform planting design across all locations during the early vegetative phase, highlighting the controlled agronomic management prior to divergent responses to altitude and nitrogen stress. These images collectively visualize the contrasting ecological contexts influencing wheat performance in this study. Beijing site: Located at low altitude, characterized by a temperate continental climate with dense vegetation and urban infrastructure in the background, providing optimal growing conditions for wheat maturation. Beijing site: Located at low altitude, characterized by a temperate continental climate with dense vegetation and urban infrastructure in the background, providing optimal growing conditions for wheat maturation. Lhasa site: Situated at high altitude (3688 m), featuring a plateau environment with thin air, strong solar radiation, and distant mountain ranges visible; these harsh conditions significantly constrain crop growth duration and biomass accumulation. Lhasa site: Situated at high altitude (3688 m), featuring a plateau environment with thin air, strong solar radiation, and distant mountain ranges visible; these harsh conditions significantly constrain crop growth duration and biomass accumulation. Seedling stage: Illustrates the uniform planting design across all locations during the early vegetative phase, highlighting the controlled agronomic management prior to divergent responses to altitude and nitrogen stress. These images collectively visualize the contrasting ecological contexts influencing wheat performance in this study. Seedling stage: Illustrates the uniform planting design across all locations during the early vegetative phase, highlighting the controlled agronomic management prior to divergent responses to altitude and nitrogen stress. These images collectively visualize the contrasting ecological contexts influencing wheat performance in this study. Fig. 1Field experiment photos of wheat varieties used in three geographical regions of Beijing, Lhasa, and Renqiu. Field experiment photos of wheat varieties used in three geographical regions of Beijing, Lhasa, and Renqiu. Field experiments were conducted during two winter wheat growing seasons, 2019–2020 and 2020–2021, at three sites: Renqiu experimental station, Cangzhou, Hebei Province; the Institute of Crop Sciences, Chinese Academy of Agricultural Sciences, Beijing; and the Institute of Lhasa Academy of Agricultural and Animal Husbandry Sciences, Tibet. Monthly rainfall, average temperature, sunshine hours and solar radiation from sowing to harvest are shown in Fig. 2. Fig. 2Climatic conditions during the winter wheat growth periods in Renqiu, Beijing and Lhasa during 2019–2020 and 2020–2021. (A) Monthly rainfall (bars) and average temperature (lines). (B) Sunshine hours (bars) and solar radiation (lines). Climatic conditions during the winter wheat growth periods in Renqiu, Beijing and Lhasa during 2019–2020 and 2020–2021. (A) Monthly rainfall (bars) and average temperature (lines). (B) Sunshine hours (bars) and solar radiation (lines). The three experimental sites exhibited distinct climatic characteristics due to differences in geography and elevation above sea level (Table 1). Renqiu (4 m) features a warm temperate subhumid continental monsoon climate, Beijing (40 m) a warm temperate semihumid to semiarid monsoon climate, and Lhasa (3688 m) a plateau temperate semiarid monsoon climate. Across the two growing seasons, Lhasa consistently received much higher solar radiation (4888.5–5238.1 MJ/m²) and sunshine hours (2682.6–2765.2 h) than Renqiu (2937.5–3119.1 MJ/m²; 1563.9–1671.2 h) and Beijing (3031.0–3033.0 MJ/m²; 1728.4–1756.6 h). However, Lhasa also showed a larger average daily temperature range (13.4–14.0 °C) and lower daily minimum temperatures (down to − 11.1 °C), indicating stronger diurnal thermal fluctuations. In contrast, Renqiu and Beijing had higher daily maximum temperatures (up to 37.8 °C and 36.3 °C, respectively) and more days with maximum temperature ≥ 33 °C (up to 14 and 5 days). Annual precipitation varied substantially, with Lhasa receiving 210.9–285.1 mm, Renqiu 73.1–183.2 mm, and Beijing 123.8–168.3 mm. These climatic contrasts—particularly in radiation, temperature range, and thermal extremes—provided the environmental basis for evaluating altitude-related effects on wheat yield and quality. Table 1Climatic characteristics of the three experimental sites during the 2019–2020 and 2020–2021 growing seasons.SiteLatitude and longitudeElevation (m)YearADT(℃)DIT (°C)DXT (°C)ADR (°C)33D (d)Rainfall(mm)Sunshine hours(h)Solar radiation (MJ/m²)Renqiu38.30°N, 116.85°E42019–20209.837.8−11.510.814183.21671.23119.12020–20218.733.6−19.410.9373.11563.92937.5Beijing39.57°N, 116.19°E402019–20208.636.3−14.411.25168.31728.43031.12020–20217.934.9−19.611.11123.81756.63033.0Lhasa29.40°N, 91.20°E36882019–20207.628.2−11.113.40285.12765.24888.52020–20219.328.7−8.914.00210.92682.65238.1ADT average daily temperature, DIT daily maximum temperature, DXT daily minimum temperature, ADR average daily temperature range, 33D number of days with daily maximum temperature ≥ 33 °C, Elevation elevation above sea level. Table 1Climatic characteristics of the three experimental sites during the 2019–2020 and 2020–2021 growing seasons.SiteLatitude and longitudeElevation (m)YearADT(℃)DIT (°C)DXT (°C)ADR (°C)33D (d)Rainfall(mm)Sunshine hours(h)Solar radiation (MJ/m²)Renqiu38.30°N, 116.85°E42019–20209.837.8−11.510.814183.21671.23119.12020–20218.733.6−19.410.9373.11563.92937.5Beijing39.57°N, 116.19°E402019–20208.636.3−14.411.25168.31728.43031.12020–20217.934.9−19.611.11123.81756.63033.0Lhasa29.40°N, 91.20°E36882019–20207.628.2−11.113.40285.12765.24888.52020–20219.328.7−8.914.00210.92682.65238.1ADT average daily temperature, DIT daily maximum temperature, DXT daily minimum temperature, ADR average daily temperature range, 33D number of days with daily maximum temperature ≥ 33 °C, Elevation elevation above sea level. Climatic characteristics of the three experimental sites during the 2019–2020 and 2020–2021 growing seasons. ADT average daily temperature, DIT daily maximum temperature, DXT daily minimum temperature, ADR average daily temperature range, 33D number of days with daily maximum temperature ≥ 33 °C, Elevation elevation above sea level. The pH, organic carbon, total N, alkali-hydrolyzed N, available P, available K and organic matter contents of the 0–20 cm and 20–40 cm soil layers before sowing are shown in Table 2. Before sowing in the two growing seasons, soil properties at Renqiu, Beijing and Lhasa were measured in the 0–20 cm and 20–40 cm layers. The soils at all three sites were slightly alkaline, with pH values ranging from 7.8 to 8.3. Baseline soil fertility differed among sites. Available P was determined using the Olsen sodium bicarbonate extraction method24. Available P was highest at Lhasa (47.19 ± 19.76 mg/kg), followed by Beijing (18.49 ± 6.67 mg/kg) and Renqiu (11.95 ± 8.74 mg/kg). Mean organic matter content was also higher at Beijing and Lhasa than at Renqiu. Soil bulk density and cation exchange capacity of the 0–20 cm layer were measured before sowing to support the interpretation of site differences. Table 2Basic soil fertility of the 0–20 cm and 20–40 cm soil layers at Renqiu, Beijing and Lhasa during the 2019–2020 and 2020–2021 growing seasons.SiteYearSoil depth (cm)pHOrganic carbon (g/kg)Total N (g/kg)Alkali-hydrolyzed N (mg/kg)Available P (mg/kg)Available K (mg/kg)Organic matter (g/kg)Renqiu2019–20200–207.910.181.09118.768.96366.0017.5520–407.95.210.5495.011.44181.008.992020–20210–207.811.991.26109.0921.76207.2720.6720–408.09.881.0270.5115.63184.6917.03Mean7.90 ± 0.08a9.32 ± 2.89b0.98 ± 0.31a98.34 ± 20.96a11.95 ± 8.74b234.74 ± 88.27a16.06 ± 4.98bBeijing2019–20200–207.811.631.30124.3512.52181.0020.0420–407.910.871.23127.1513.28154.0018.742020–20210–208.313.671.2095.3222.10206.7923.5720–408.114.641.27117.3626.07351.9525.24Mean8.03 ± 0.22a12.70 ± 1.75a1.25 ± 0.04a116.05 ± 14.42a18.49 ± 6.67b223.44 ± 88.35a21.90 ± 3.02aLhasa2019–20200–207.813.481.3698.8828.59139.6023.2520–407.810.111.04135.9331.98100.8017.432020–20210–208.213.631.30120.0060.80148.0023.5020–407.814.621.47174.0067.40192.0025.20Mean7.90 ± 0.20a12.96 ± 1.97a1.29 ± 0.18a132.20 ± 31.73a47.19 ± 19.76a145.10 ± 37.42a22.35 ± 3.39aData in “Mean” rows are expressed as Mean ± SD. Different lowercase letters within a column for “Mean” values indicate significant differences among sites at P < 0.05 according to Duncan’s multiple range test. Table 2Basic soil fertility of the 0–20 cm and 20–40 cm soil layers at Renqiu, Beijing and Lhasa during the 2019–2020 and 2020–2021 growing seasons.SiteYearSoil depth (cm)pHOrganic carbon (g/kg)Total N (g/kg)Alkali-hydrolyzed N (mg/kg)Available P (mg/kg)Available K (mg/kg)Organic matter (g/kg)Renqiu2019–20200–207.910.181.09118.768.96366.0017.5520–407.95.210.5495.011.44181.008.992020–20210–207.811.991.26109.0921.76207.2720.6720–408.09.881.0270.5115.63184.6917.03Mean7.90 ± 0.08a9.32 ± 2.89b0.98 ± 0.31a98.34 ± 20.96a11.95 ± 8.74b234.74 ± 88.27a16.06 ± 4.98bBeijing2019–20200–207.811.631.30124.3512.52181.0020.0420–407.910.871.23127.1513.28154.0018.742020–20210–208.313.671.2095.3222.10206.7923.5720–408.114.641.27117.3626.07351.9525.24Mean8.03 ± 0.22a12.70 ± 1.75a1.25 ± 0.04a116.05 ± 14.42a18.49 ± 6.67b223.44 ± 88.35a21.90 ± 3.02aLhasa2019–20200–207.813.481.3698.8828.59139.6023.2520–407.810.111.04135.9331.98100.8017.432020–20210–208.213.631.30120.0060.80148.0023.5020–407.814.621.47174.0067.40192.0025.20Mean7.90 ± 0.20a12.96 ± 1.97a1.29 ± 0.18a132.20 ± 31.73a47.19 ± 19.76a145.10 ± 37.42a22.35 ± 3.39aData in “Mean” rows are expressed as Mean ± SD. Different lowercase letters within a column for “Mean” values indicate significant differences among sites at P < 0.05 according to Duncan’s multiple range test. Basic soil fertility of the 0–20 cm and 20–40 cm soil layers at Renqiu, Beijing and Lhasa during the 2019–2020 and 2020–2021 growing seasons. Data in “Mean” rows are expressed as Mean ± SD. Different lowercase letters within a column for “Mean” values indicate significant differences among sites at P < 0.05 according to Duncan’s multiple range test. Experimental design The experiment compared yield and quality responses of three gluten types of winter wheat under two N topdressing rates at three sites over two years. A randomized complete block design was used at each site, and each treatment had three replicated plots. Urea was used as the N fertilizer. Two N treatments were applied: N240, with 120 kg/ha basal N and 120 kg/ha topdressing N at jointing, and N180, with 120 kg/ha basal N and 60 kg/ha topdressing N at jointing. Basal N was band-applied at 5 cm depth within 24 h after sowing, and topdressing was broadcast and incorporated at the jointing stage (Feekes stage 6, plant height about 30 ± 2 cm). broadcast and incorporated 150 kg P₂O₅/ha and potassium sulfate at 120 kg K₂O/ha before sowing. Each plot was 12 m², with 20 cm row spacing and 50 cm between plots. The target seedling density was 375 plants/m². Irrigation of 60 mm was applied at overwintering, after jointing fertilization and early grain filling, respectively. Field management was standardized across sites. Imidacloprid (75 g of active ingredient/ha) was applied for aphid control at stem elongation, and tebuconazole (125 g of active ingredient/ha) was applied for fungal disease prevention at anthesis. Crop measurement Grain yield and yield components Wheat was harvested at physiological maturity when the grains had turned completely golden. The whole plot was used for management, and a 3.6 m² harvest area in the centre of each plot was used for yield determination to avoid border effects. After seedling emergence, a fixed 1 m row section was selected to count spike number. Ten plants were randomly selected from each plot to measure plant height, spike length, total spikelet number, fertile spikelet number and grains per spike. Thousand-grain weight was determined from 1000 randomly selected grains from each plot. Test weight was measured with a grain test weight meter (with readings recorded to the nearest 0.1 g/L). After harvest, grain samples were stored for three months, milled with a Quadrumat Junior Test Mill (Brabender GmbH & Co. KG, Duisburg, Germany), and then used for flour quality analyses. Nitrogen and protein content determination Mature grains from each plot were mixed and ground with a cyclone mill (FOSS CT293 Cyclone sample mill) to obtain whole wheat flour. The N content of 0.100 ± 0.001 g whole wheat flour was determined using the Kjeldahl method with an Automatic Kjeldahl Nitrogen Analyzer 9860 (Jinan Haineng Instrument Co., Ltd., Shandong, China) following standard procedures25. Whole grain protein content was calculated as N content × 5.7, as commonly used for wheat protein conversion26. Each biological replicate was measured independently, and technical duplicates were used when instrument measurements were required. Grain protein fractionation and physical properties Albumin, globulin, gliadin and glutenin were extracted successively using the classical Osborne fractionation procedure and its later modifications for wheat proteins26,27. Wheat grain hardness was measured using a JYDX100 × 40 Wheat Hardness Tester (Wuxi Xiliang Machinery Co., Ltd., Jiangsu, China). Flour gluten characteristics Total gluten content and dry and wet gluten contents were determined with a Perten GM2200 gluten instrument (Perten Instruments Inc., Hägersten, Sweden) according to AACC Method 38-12.02. Sedimentation value was determined according to AACC Method 56–61.0226. Dough rheological properties Water absorption, dough development time, stability time, weakening degree and farinograph index were determined using a Farinograph-E with a USB port (Brabender GmbH & Co. KG, Duisburg, Germany) according to AACC Method 54-50.0126. Phenological and meteorological analysis The following phenological stages and indices were calculated from field observation dates. M was the whole growth period from seedling emergence to maturity. DVG was the number of days from emergence to regreening, DVR was the number of days from regreening to heading, and DRG was the number of days from heading to maturity. Regreening was defined as the stage when overwintered seedlings resumed visible growth and new green leaves appeared after winter dormancy. RVG was the ratio of vegetative growth duration to the whole growth period, RRG was the ratio of reproductive growth duration to the whole growth period, and RR/V was the ratio of reproductive growth duration to vegetative growth duration. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{RVG }} = {\text{ DVG }}/{\text{ DM }} \times {\text{ 1}}00\%$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{RRG }} = {\text{ DRG }}/{\text{ DM }} \times {\text{ 1}}00\%$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{RR}}/{\text{V }} = {\text{ DRG }}/{\text{ DVG }} \times {\text{ 1}}00\%$$\end{document} DVG, DM and DRG are expressed in days, while RVG, RRG and RR/V are expressed as percentages or ratios. Meteorological data were obtained from the local meteorological stations. Sunshine duration (SM, SVG, SVR and SRG), accumulated temperature (TM, TVG, TVR and TRG), and precipitation (P) were calculated for the whole growth period, vegetative growth period, overlapping vegetative-reproductive period and reproductive growth period. Statistical analysis Statistical analysis was performed using R (version 4.3.0). Data were expressed as mean ± standard deviation (SD). One-way ANOVA followed by Duncan’s multiple range test (P < 0.05) was used to compare initial soil properties and meteorological factors among sites. For yield and quality traits, year, site/elevation above sea level, gluten type, cultivar and N treatment were included according to the experimental design. Biological replicates corresponded to field plots (n = 3 per treatment), and technical replicates were used for laboratory instrument measurements when applicable. PLS-PM quantified the pathways linking environmental drivers to grain yield. Standardized path coefficients (β) were used to compare effect sizes among variables. Reproductive-stage accumulated temperature (TRG) had a direct negative path to thousand-grain weight (β = −0.68, P < 0.01) in the fitted model, while reproductive-stage sunshine duration and precipitation also contributed indirectly through yield components (Fig. 4). For significant interaction terms, post-hoc contrasts were performed to compare the effects of site/elevation above sea level, year, gluten type and N treatment. Because only three altitude levels were included, relationships with altitude were interpreted as site-based descriptive trends rather than as a universal altitude-response model. Figures and graphical summaries were prepared using Microsoft Excel 2019 and Origin 2022. Response surface analysis was used to visualize the joint response of gluten index to altitude above sea level and N rate; because only three altitudes were tested, the fitted surfaces were interpreted descriptively. Descriptive statistics, including maximum, minimum, range, mean, SD and coefficient of variation (CV), were calculated. Pearson correlation analysis was used to assess associations among meteorological factors, yield components, protein fractions and quality parameters. Results Differences in planting sites and the wheat growth cycle Meteorological conditions differed clearly among the three sites (Table 3). Lhasa had the highest sunshine duration and radiation but a lower and more variable thermal environment than the low-altitude sites. Beijing showed relatively favourable accumulated temperature and precipitation during the whole growth period, whereas Renqiu had relatively high reproductive-stage sunshine duration. These site differences provided the environmental basis for comparing altitude-related responses. Across the two growing seasons, Beijing and Renqiu represented low-altitude environments (40 and 4 m), while Lhasa represented a high-altitude plateau environment (3688 m). The meteorological data indicated that altitude above sea level changed the balance among light, temperature and water conditions. In particular, the Lhasa site combined strong radiation with lower effective thermal resources during key growth stages. Because gluten groups differed in phenology, the accumulated sunshine duration, temperature and precipitation experienced by S-W, M-W and W-W also differed. W-W generally had a longer growth period and accumulated more meteorological resources, whereas M-W completed reproductive growth more rapidly. These differences were considered when interpreting yield and quality responses. Table 3Meteorological factors at the three winter wheat growing sites over the entire growth period.SiteYearCultivar groupSM(h)SVG(h)SVR(h)SRG(h)TM(℃)TVG(℃)TVR(℃)TRG(℃) P (mm)Renqiu2019–2020S-W1577.05882.75313.70380.602198.65696.20503.30999.15179.50M-W1550.20875.20306.00369.002104.40681.45492.45930.50179.50W-W1590.45882.75317.90389.802283.20696.20513.401073.60179.50Average1572.57880.23312.53379.802195.42691.28503.051001.08179.502020–2021S-W1517.45912.15226.75378.552010.35544.15491.60974.6073.10M-W1477.15911.55214.05351.551911.75532.75473.05905.9573.10W-W1521.70912.30231.70377.702053.35554.45513.90985.0073.10Average1505.43912.00224.17369.271991.82543.78492.85955.1873.10Mean1539.00 ± 40.27c896.12 ± 17.04c268.35 ± 46.68b374.53 ± 13.46b2093.62 ± 127.61b617.53 ± 77.76b497.95 ± 17.02a978.13 ± 57.18b126.30 ± 55.57bBeijing2019–2020S-W1646.051058.50237.85349.701981.50691.00377.85912.65168.30M-W1619.851062.00202.45355.401918.70686.50344.65887.55167.30W-W1681.101069.80259.70351.602036.60689.60404.05942.95152.80Average1649.001063.43233.33352.231978.93689.03375.52914.38162.802020–2021S-W1694.501150.70185.10358.701924.50671.05398.05855.4087.30M-W1674.751146.65183.20344.901878.65657.25391.90829.5087.30W-W1712.901156.05180.80376.051964.00699.60376.25888.1587.30Average1694.051151.13183.03359.881922.38675.97388.73857.6887.30Mean1671.53 ± 32.50b1107.28 ± 46.09b208.18 ± 31.17c356.06 ± 11.85b1950.66 ± 54.40b682.50 ± 16.03ab382.13 ± 21.95b886.03 ± 39.78c125.05 ± 39.78bLhasa2019–2020S-W2648.101580.10389.90678.102049.85585.95390.801073.10280.60M-W2620.301588.90385.90645.502001.05601.30386.201013.55236.90W-W2658.751576.70403.30678.752061.80573.10406.801081.90281.30Average2642.381581.90393.03667.452037.57586.78394.601056.18266.272020–2021S-W2596.701659.10493.45444.152611.55963.00612.001036.55184.80M-W2573.551659.10461.45453.002536.20963.00552.951020.25154.35W-W2602.851659.10486.20457.552627.60963.00593.301071.30199.70Average2591.031659.10480.37451.572591.78963.00586.081042.70179.62Mean2616.71 ± 531.27a1620.50 ± 40.50a436.70 ± 47.22a559.51 ± 113.61a2314.68 ± 291.73a774.89 ± 196.67a490.34 ± 102.11a1049.44 ± 30.65a222.94 ± 50.50aSM sunshine duration from emergence to maturity, SVG sunshine duration of vegetative growth period, SVR sunshine duration of vegetative growth and reproductive growth period, SRG sunshine duration of reproductive growth period, TM accumulated temperature from emergence to maturity, TVG accumulated temperature of vegetative growth period, TVR accumulated temperature of vegetative growth and reproductive growth period, TRG accumulated temperature of reproductive growth period, P precipitation from emergence to maturity.Cultivar group data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24).Data in “Mean” rows are expressed as Mean ± SD. Different lowercase letters within a column for “Mean” values indicate significant differences among sites at P < 0.05 according to Duncan’s multiple range test. Table 3Meteorological factors at the three winter wheat growing sites over the entire growth period.SiteYearCultivar groupSM(h)SVG(h)SVR(h)SRG(h)TM(℃)TVG(℃)TVR(℃)TRG(℃) P (mm)Renqiu2019–2020S-W1577.05882.75313.70380.602198.65696.20503.30999.15179.50M-W1550.20875.20306.00369.002104.40681.45492.45930.50179.50W-W1590.45882.75317.90389.802283.20696.20513.401073.60179.50Average1572.57880.23312.53379.802195.42691.28503.051001.08179.502020–2021S-W1517.45912.15226.75378.552010.35544.15491.60974.6073.10M-W1477.15911.55214.05351.551911.75532.75473.05905.9573.10W-W1521.70912.30231.70377.702053.35554.45513.90985.0073.10Average1505.43912.00224.17369.271991.82543.78492.85955.1873.10Mean1539.00 ± 40.27c896.12 ± 17.04c268.35 ± 46.68b374.53 ± 13.46b2093.62 ± 127.61b617.53 ± 77.76b497.95 ± 17.02a978.13 ± 57.18b126.30 ± 55.57bBeijing2019–2020S-W1646.051058.50237.85349.701981.50691.00377.85912.65168.30M-W1619.851062.00202.45355.401918.70686.50344.65887.55167.30W-W1681.101069.80259.70351.602036.60689.60404.05942.95152.80Average1649.001063.43233.33352.231978.93689.03375.52914.38162.802020–2021S-W1694.501150.70185.10358.701924.50671.05398.05855.4087.30M-W1674.751146.65183.20344.901878.65657.25391.90829.5087.30W-W1712.901156.05180.80376.051964.00699.60376.25888.1587.30Average1694.051151.13183.03359.881922.38675.97388.73857.6887.30Mean1671.53 ± 32.50b1107.28 ± 46.09b208.18 ± 31.17c356.06 ± 11.85b1950.66 ± 54.40b682.50 ± 16.03ab382.13 ± 21.95b886.03 ± 39.78c125.05 ± 39.78bLhasa2019–2020S-W2648.101580.10389.90678.102049.85585.95390.801073.10280.60M-W2620.301588.90385.90645.502001.05601.30386.201013.55236.90W-W2658.751576.70403.30678.752061.80573.10406.801081.90281.30Average2642.381581.90393.03667.452037.57586.78394.601056.18266.272020–2021S-W2596.701659.10493.45444.152611.55963.00612.001036.55184.80M-W2573.551659.10461.45453.002536.20963.00552.951020.25154.35W-W2602.851659.10486.20457.552627.60963.00593.301071.30199.70Average2591.031659.10480.37451.572591.78963.00586.081042.70179.62Mean2616.71 ± 531.27a1620.50 ± 40.50a436.70 ± 47.22a559.51 ± 113.61a2314.68 ± 291.73a774.89 ± 196.67a490.34 ± 102.11a1049.44 ± 30.65a222.94 ± 50.50aSM sunshine duration from emergence to maturity, SVG sunshine duration of vegetative growth period, SVR sunshine duration of vegetative growth and reproductive growth period, SRG sunshine duration of reproductive growth period, TM accumulated temperature from emergence to maturity, TVG accumulated temperature of vegetative growth period, TVR accumulated temperature of vegetative growth and reproductive growth period, TRG accumulated temperature of reproductive growth period, P precipitation from emergence to maturity.Cultivar group data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24).Data in “Mean” rows are expressed as Mean ± SD. Different lowercase letters within a column for “Mean” values indicate significant differences among sites at P < 0.05 according to Duncan’s multiple range test. Meteorological factors at the three winter wheat growing sites over the entire growth period. SM sunshine duration from emergence to maturity, SVG sunshine duration of vegetative growth period, SVR sunshine duration of vegetative growth and reproductive growth period, SRG sunshine duration of reproductive growth period, TM accumulated temperature from emergence to maturity, TVG accumulated temperature of vegetative growth period, TVR accumulated temperature of vegetative growth and reproductive growth period, TRG accumulated temperature of reproductive growth period, P precipitation from emergence to maturity. Cultivar group data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Data in “Mean” rows are expressed as Mean ± SD. Different lowercase letters within a column for “Mean” values indicate significant differences among sites at P < 0.05 according to Duncan’s multiple range test. Phenological traits differed significantly among the three gluten types and across sites (Table 4). Overall, the high-altitude environment at Lhasa (3688 m) prolonged the whole growth period (DM) compared with Renqiu (4 m) and Beijing (40 m). Mean DM at Lhasa reached 281.4 days, which was 41 days longer than at Renqiu (240.5 days) and 38 days longer than at Beijing (243.2 days). This extension was mainly driven by a substantially longer reproductive period (DRG): Lhasa averaged 60.2 days, whereas Renqiu and Beijing averaged only 45.8 and 43.7 days, respectively. Consequently, the proportion of reproductive growth (RRG) was highest at Lhasa (21. %), followed by Renqiu (19. %) and Beijing (18. %). In contrast, the vegetative phase (DVG) showed smaller differences among sites. Table 4The days and proportion of wheat at different growth stages at different altitudes.Test SitesYearCultivar groupDVG (d)DRG (d)DM (d)RVG (%)RRG (%)Renqiu2019–2020S-W163.544.5243.555.018.3M-W162.542.5240.555.717.7W-W163.547.0246.554.419.12020–2021S-W154.547.5238.050.020.0M-W153.545.5234.550.819.4W-W155.548.0240.049.620.0Mean158.83 ± 4.61b45.83 ± 2.04b240.50 ± 4.01b52.58 ± 2.62b19.05 ± 0.91bBeijing2019–2020S-W173.543.5244.059.017.8M-W172.544.5242.060.018.4W-W174.043.5246.058.117.72020–2021S-W172.543.5242.557.318.0M-W171.542.5240.557.817.7W-W174.544.5244.057.018.2Mean173.08 ± 1.16a43.67 ± 0.89b243.217 ± 1.90b58.19 ± 1.06a17.95 ± 0.35cLhasa2019–2020S-W175.564.5277.547.823.3M-W177.061.0275.048.822.2W-W174.064.5277.547.223.32020–2021S-W174.057.0287.545.219.9M-W174.055.0282.546.019.5W-W174.059.0288.545.120.5Mean174.75 ± 1.22a60.17 ± 3.81a281.42 ± 5.43a46.66 ± 1.42c21.41 ± 1.65aDVG the days number of vegetative growth period, DRG the days number of reproductive growth period, DM the number of days from emergence to maturity, RVG ratio of vegetative growth period, RRG ratio of reproductive growth period.Data in “Mean” rows are expressed as Mean ± SD. Different lowercase letters within a column for “Mean” values indicate significant differences among sites at P < 0.05 according to Duncan’s multiple range test.ZM-578(Zhongmai 578); ZM-5051(Zhongmai 5051); ZM-82(Zhongmai 82); ZM-89(Zhongmai 89); YM-15(Yangmai 15); YM-24(Yangmai 24). Table 4The days and proportion of wheat at different growth stages at different altitudes.Test SitesYearCultivar groupDVG (d)DRG (d)DM (d)RVG (%)RRG (%)Renqiu2019–2020S-W163.544.5243.555.018.3M-W162.542.5240.555.717.7W-W163.547.0246.554.419.12020–2021S-W154.547.5238.050.020.0M-W153.545.5234.550.819.4W-W155.548.0240.049.620.0Mean158.83 ± 4.61b45.83 ± 2.04b240.50 ± 4.01b52.58 ± 2.62b19.05 ± 0.91bBeijing2019–2020S-W173.543.5244.059.017.8M-W172.544.5242.060.018.4W-W174.043.5246.058.117.72020–2021S-W172.543.5242.557.318.0M-W171.542.5240.557.817.7W-W174.544.5244.057.018.2Mean173.08 ± 1.16a43.67 ± 0.89b243.217 ± 1.90b58.19 ± 1.06a17.95 ± 0.35cLhasa2019–2020S-W175.564.5277.547.823.3M-W177.061.0275.048.822.2W-W174.064.5277.547.223.32020–2021S-W174.057.0287.545.219.9M-W174.055.0282.546.019.5W-W174.059.0288.545.120.5Mean174.75 ± 1.22a60.17 ± 3.81a281.42 ± 5.43a46.66 ± 1.42c21.41 ± 1.65aDVG the days number of vegetative growth period, DRG the days number of reproductive growth period, DM the number of days from emergence to maturity, RVG ratio of vegetative growth period, RRG ratio of reproductive growth period.Data in “Mean” rows are expressed as Mean ± SD. Different lowercase letters within a column for “Mean” values indicate significant differences among sites at P < 0.05 according to Duncan’s multiple range test.ZM-578(Zhongmai 578); ZM-5051(Zhongmai 5051); ZM-82(Zhongmai 82); ZM-89(Zhongmai 89); YM-15(Yangmai 15); YM-24(Yangmai 24). The days and proportion of wheat at different growth stages at different altitudes. DVG the days number of vegetative growth period, DRG the days number of reproductive growth period, DM the number of days from emergence to maturity, RVG ratio of vegetative growth period, RRG ratio of reproductive growth period. Data in “Mean” rows are expressed as Mean ± SD. Different lowercase letters within a column for “Mean” values indicate significant differences among sites at P < 0.05 according to Duncan’s multiple range test. ZM-578(Zhongmai 578); ZM-5051(Zhongmai 5051); ZM-82(Zhongmai 82); ZM-89(Zhongmai 89); YM-15(Yangmai 15); YM-24(Yangmai 24). Within each site, the three gluten types exhibited consistent phenological rankings. Weak gluten wheat (W-W) tended to have a relatively longer reproductive period than the other two types, especially at Lhasa. Medium gluten wheat (M-W) generally completed emergence and reproductive growth earlier than S-W and W-W, resulting in a shorter whole growth period. These consistent differences suggest that gluten types possess distinct developmental adaptations to high-altitude conditions. The prolonged reproductive phase at high altitude, particularly in W-W, may represent a compensatory strategy under low-temperature stress, but it did not translate into higher yield, as shown in later sections. ANOVA indicated that planting site had a highly significant effect on the wheat growth cycle (P < 0.01; Table 4). The growth duration at Lhasa (3688 m) was longer than at Beijing (40 m), which was consistent with the lower thermal conditions at the plateau site (Tables 1 and 3). Fig. 3Distribution of yield and quality parameters across Renqiu, Beijing and Lhasa. Boxplots show median, quartiles and range; different lowercase letters indicate significant differences at P < 0.05 according to Duncan’s multiple range test. The phenological indices DE, DVR, DRG, RRG and RR/V were developmental variables rather than cultivar names. These indices consistently indicated delayed development and a longer growth cycle at Lhasa, especially during reproductive growth. This delay was associated with high-altitude thermal constraints rather than with higher yield. Distribution of yield and quality parameters across Renqiu, Beijing and Lhasa. Boxplots show median, quartiles and range; different lowercase letters indicate significant differences at P < 0.05 according to Duncan’s multiple range test. The phenological indices DE, DVR, DRG, RRG and RR/V were developmental variables rather than cultivar names. These indices consistently indicated delayed development and a longer growth cycle at Lhasa, especially during reproductive growth. This delay was associated with high-altitude thermal constraints rather than with higher yield. Interannual differences were present at each site, but they were smaller than the overall contrast between the low-altitude sites and Lhasa. Therefore, subsequent analyses emphasized site/elevation above sea level, gluten type and N treatment trends rather than listing all year-to-year numerical comparisons. Grain yield and yield components Mixed linear model and ANOVA results showed that site/elevation above sea level was a dominant source of variation in grain yield and yield components (Tables 5 and 6; Fig. 4). The effect size of site/elevation above sea level was larger than that of N treatment and several interaction terms, indicating that environmental differences among sites strongly shaped yield formation. Mean grain yield followed the order Beijing (794.9 g/m²) > Renqiu (728.2 g/m²) > Lhasa (444.4 g/m²). Because only three altitudes above sea level were tested, the dashed trend lines in Fig. 3A were interpreted as descriptive trends across sites rather than as formal evidence of a significant linear correlation with altitude. Yield reduction at Lhasa was more evident in S-W and M-W than in W-W, suggesting that high-altitude stress restricted high-yield and high-quality types more strongly. Further analysis of yield components indicated that the yield penalty at Lhasa was caused by simultaneous reductions in spike number, grains per spike and thousand-grain weight. Based on the raw yield dataset, mean grain yield followed Beijing (794.9 g/m²) > Renqiu (728.2 g/m²) > Lhasa (444.4 g/m²). Mean spike number was 559.6, 524.7 and 431.9 spikes/m² in Beijing, Renqiu and Lhasa, respectively. Mean grains per spike were 82.4, 72.2 and 53.0, and mean thousand-grain weight was 34.9, 31.4 and 27.5 g, respectively. The year (Y) × site (S) interaction was significant for several traits, but the overall trend of lower yield components at Lhasa was consistent across years (Tables 5 and 6). Fig. 4Site/elevation above sea level effects on wheat grain yield, thousand-grain weight, grain protein content and harvest index. Points show means for gluten groups, and error bars indicate standard errors. Lines show descriptive site-level trends across the three tested altitudes and should not be interpreted as a universal altitude-response regression. Site/elevation above sea level effects on wheat grain yield, thousand-grain weight, grain protein content and harvest index. Points show means for gluten groups, and error bars indicate standard errors. Lines show descriptive site-level trends across the three tested altitudes and should not be interpreted as a universal altitude-response regression. Analysis of variance (ANOVA) presented in Table 5 revealed that year (Y), cultivar or gluten type (C), site/elevation above sea level (S), and their interactions all exerted highly significant effects on grain yield, yield components, and agronomic traits of winter wheat. Among the three main factors, site (S) and year (Y) generally showed the largest F-values across most traits, indicating that environmental conditions, particularly altitude-associated changes in temperature, radiation, and water availability, were the dominant drivers of yield variation. For example, site effects on yield (F = 741.23), plant height (F = 2367.05), and spike number (F = 173.29**) were exceptionally strong. Cultivar or gluten type (C) also significantly influenced all traits, though with generally smaller F-values than site or year, confirming that genetic background modulates how plants respond to environmental gradients. More importantly, several two-way and three-way interactions were significant. The S × C interaction was significant for all traits (e.g., yield: F = 35.06**), suggesting that the relative performance of strong, medium, and weak gluten wheats changed across sites. The Y × C interaction was significant for most traits except grains per spike, indicating that year-to-year climate variability affected gluten types differently. The three-way interaction (S × Y × C) was significant for yield and most yield components (e.g., yield: F = 36.21**), highlighting that the combined effect of site, year, and gluten type was not simply additive. Collectively, these results demonstrate that spike formation, grain set, and final grain yield are jointly regulated by site-specific environmental conditions and genotype. Therefore, optimizing wheat production across altitudinal gradients requires simultaneous consideration of seasonal weather patterns, cultivar selection based on gluten type, and their complex interactions. Table 5Analysis of variance for yield, yield components, and agronomic traits as affected by site/elevation above sea level, growing season, cultivar type and their interactions (F-values).Influence factorsSpike number (spikes/m²)Thousand grain weight (g)Grains per spikeYield (g/m²)Spike length (cm)Plant height (cm)Total spikelet numberNumber of fertile spikelets per spikeSite (S)173.29**61.28**76.77**741.23**303.19**2367.05**200.26**148.65**Year (Y)179.62**85.99**53.61**1121.55**7.24**213.49**76.64**73.95**Cultivar (C)13.38**19.02**44.42**15.50**38.76**218.55**60.93**71.02**S × Y5.00**2.7425.80**121.25**55.60**173.28**11.23**3.01S × C19.20**7.72**5.50**35.06**11.85**13.80**6.51**7.77**Y × C5.50**1.094.36*19.45**24.11**162.60**35.26**6.69**S × Y × C28.47**0.078.07**36.21**18.83**41.14**11.77**4.02**Years data are from 2019 to 2020 and 2020–2021.Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24).*, and ** indicate significance at the 0.05, and 0.01 probability levels, respectively. Table 5Analysis of variance for yield, yield components, and agronomic traits as affected by site/elevation above sea level, growing season, cultivar type and their interactions (F-values).Influence factorsSpike number (spikes/m²)Thousand grain weight (g)Grains per spikeYield (g/m²)Spike length (cm)Plant height (cm)Total spikelet numberNumber of fertile spikelets per spikeSite (S)173.29**61.28**76.77**741.23**303.19**2367.05**200.26**148.65**Year (Y)179.62**85.99**53.61**1121.55**7.24**213.49**76.64**73.95**Cultivar (C)13.38**19.02**44.42**15.50**38.76**218.55**60.93**71.02**S × Y5.00**2.7425.80**121.25**55.60**173.28**11.23**3.01S × C19.20**7.72**5.50**35.06**11.85**13.80**6.51**7.77**Y × C5.50**1.094.36*19.45**24.11**162.60**35.26**6.69**S × Y × C28.47**0.078.07**36.21**18.83**41.14**11.77**4.02**Years data are from 2019 to 2020 and 2020–2021.Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24).*, and ** indicate significance at the 0.05, and 0.01 probability levels, respectively. Analysis of variance for yield, yield components, and agronomic traits as affected by site/elevation above sea level, growing season, cultivar type and their interactions (F-values). Years data are from 2019 to 2020 and 2020–2021. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). *, and ** indicate significance at the 0.05, and 0.01 probability levels, respectively. Analysis of variance presented in Table 6 revealed that cultivar or gluten type (C), site/elevation above sea level (S), and nitrogen topdressing rate (N) all significantly influenced wheat yield and yield components, though with marked differences in effect sizes. Site (S) exhibited the strongest effects across almost all traits, with F-values for yield (234.48), spike length (438.77), and total spikelet number (116.30) being particularly high. This confirms that environmental conditions associated with altitude were the dominant drivers of yield formation. Cultivar (C) also significantly affected all traits, with F-values ranging from 4.90 for yield to 59.53** for fertile spikelets, indicating that genetic background consistently modulated agronomic performance across environments. Nitrogen treatment (N) significantly increased spike number, grains per spike, thousand-grain weight, and grain yield (F = 48.92**), demonstrating a positive effect of higher N supply on productivity. However, N did not significantly affect spike length, plant height, or spikelet numbers, suggesting that nitrogen primarily influenced sink capacity rather than vegetative or spike architecture. Importantly, the S × C interaction was significant for all traits (e.g., yield: F = 11.09**), indicating that the relative performance of strong, medium, and weak gluten wheats varied across altitudinal environments. In contrast, the S × N interaction was not significant for any trait, and the three-way interaction (S × C × N) was also non-significant. These results indicate that the main effect of site was stronger than the differential nitrogen response among sites, and that nitrogen management could be optimized independently of gluten type across altitudes to a large extent. Table 6Analysis of variance for yield, yield components, and agronomic traits as affected by site/elevation above sea level, cultivar type, N treatment and their interactions (F-values).Influence factorsSpike number (spikes/m²)Thousand grain weight (g)Grain per spikeYield (g/m²)Spike length (cm)Plant height (cm)Total spikelet numberNumber of fertile spikelets per spikeSite (S)103.77**59.99**53.47**234.48**438.77**138.17**116.30**124.61**Cultivar (C)8.01**18.61**31.00**4.90**40.51**17.71**35.35**59.53**Nitrogen (N)12.02**6.22*19.66**48.92**0.081.350.300.59S × C11.50**7.53**3.84**11.09**2.56*5.42**3.78**6.52**C × N0.090.271.391.570.000.320.100.16S × N0.430.362.340.450.720.060.550.41S × C × N0.190.050.300.130.051.140.300.06Data are the mean values from 2019 to 2020 and 2020–2021.Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24).Nitrogen treatment data are from N240 and N180.*, and ** indicate significance at the 0.05, and 0.01 probability levels, respectively. Table 6Analysis of variance for yield, yield components, and agronomic traits as affected by site/elevation above sea level, cultivar type, N treatment and their interactions (F-values).Influence factorsSpike number (spikes/m²)Thousand grain weight (g)Grain per spikeYield (g/m²)Spike length (cm)Plant height (cm)Total spikelet numberNumber of fertile spikelets per spikeSite (S)103.77**59.99**53.47**234.48**438.77**138.17**116.30**124.61**Cultivar (C)8.01**18.61**31.00**4.90**40.51**17.71**35.35**59.53**Nitrogen (N)12.02**6.22*19.66**48.92**0.081.350.300.59S × C11.50**7.53**3.84**11.09**2.56*5.42**3.78**6.52**C × N0.090.271.391.570.000.320.100.16S × N0.430.362.340.450.720.060.550.41S × C × N0.190.050.300.130.051.140.300.06Data are the mean values from 2019 to 2020 and 2020–2021.Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24).Nitrogen treatment data are from N240 and N180.*, and ** indicate significance at the 0.05, and 0.01 probability levels, respectively. Analysis of variance for yield, yield components, and agronomic traits as affected by site/elevation above sea level, cultivar type, N treatment and their interactions (F-values). Data are the mean values from 2019 to 2020 and 2020–2021. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Nitrogen treatment data are from N240 and N180. *, and ** indicate significance at the 0.05, and 0.01 probability levels, respectively. Partial least squares path modeling (PLS-PM) quantified the causal pathways linking environmental drivers to grain yield. Altitude exerted a direct negative effect on thousand-grain weight (β = −0.10), while simultaneously promoting spike number through a positive indirect pathway (β = +0.21). During the reproductive phase, accumulated growing degree days showed a significant negative influence on grains per spike (β = −0.10). Conversely, solar radiation played a crucial role in enhancing spike density with a strong positive effect (β=+0.25). Ultimately, all components contributed to final yield, with grains per spike and spike number showing the strongest positive associations (β=+0.51 and β=+0.41). The model confirms that altitude acts as a primary limiting factor, negatively impacting yield both directly and by reducing thousand-grain weight (Fig. 5). Fig. 5Partial least squares path model linking environmental factors, yield components, and grain yield. Standardized path coefficients (β) are shown on arrows; green arrows indicate positive effects and red arrows indicate negative effects. Partial least squares path model linking environmental factors, yield components, and grain yield. Standardized path coefficients (β) are shown on arrows; green arrows indicate positive effects and red arrows indicate negative effects. Fig. 6Comparison of wheat yield and yield components among gluten types and planting sites. Data are the mean values from 2019 to 2020 and 2020–2021. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Bars show mean values from 2019 to 2020 and 2020–2021; error bars indicate standard errors. Comparison of wheat yield and yield components among gluten types and planting sites. Data are the mean values from 2019 to 2020 and 2020–2021. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Bars show mean values from 2019 to 2020 and 2020–2021; error bars indicate standard errors. Figure 6 showed that the yield penalty at Lhasa was associated with simultaneous reductions in spike number, grains per spike and thousand-grain weight. Across gluten types, W-W showed a smaller yield penalty than S-W and M-W, indicating better stability under high-altitude conditions. Fig. 7Comparison of agronomic traits among gluten types and planting sites. Data are the mean values from 2019 to 2020 and 2020–2021. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Different lower-case letters indicate significance at the probability level of 0.05, respectively. Comparison of agronomic traits among gluten types and planting sites. Data are the mean values from 2019 to 2020 and 2020–2021. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Different lower-case letters indicate significance at the probability level of 0.05, respectively. Agronomic traits were significantly influenced by gluten type and planting site. Average spike length followed Beijing (7.8 cm) > Renqiu (7.3 cm) > Lhasa (6.5 cm), which indicated that high-altitude conditions restricted spike elongation and spikelet formation. Spikelet development was constrained in Lhasa. The average total spikelet number was highest in Beijing and lowest in Lhasa. The decline in fertile spikelet number further contributed to reduced grain number per spike and final yield. Fig. 8Summarizing results of wheat yield responses to site/elevation above sea level, N treatment and gluten type. Summarizing results of wheat yield responses to site/elevation above sea level, N treatment and gluten type. Figure 8 summarizes wheat yield responses to site/elevation above sea level, N treatment and gluten type. Site differences explained most of the yield variation, with Beijing and Renqiu producing higher yields than Lhasa. Nitrogen topdressing improved mean yield, but the location × N interaction shown in Fig. 7D was weak or non-significant, indicating that the direction of the N effect was generally consistent across sites. The multivariate panels in Fig. 7 further showed that the Lhasa yield gap was linked mainly to lower spike number, fewer grains per spike and reduced thousand-grain weight. Because the study included only three altitude levels, the matrix in Fig. 7F was interpreted as a descriptive association among variables rather than as a stand-alone proof of altitude-yield correlation. Altitude × Nitrogen interaction on gluten quality Response surface analysis visualized the joint effect of altitude above sea level and N rate on gluten index in strong gluten wheat. At low altitudes (< 500 m above sea level), N240 tended to increase gluten index compared with N180. At high altitude (> 3000 m above sea level), gluten index remained lower even under N240, indicating that additional N could not fully offset the quality limitation imposed by the Lhasa environment. Fig. 9Response surface analysis of the interactive effects of altitude above sea level and N rate on gluten index in strong gluten wheat. Response surface analysis of the interactive effects of altitude above sea level and N rate on gluten index in strong gluten wheat. As illustrated in Fig. 9, the fitted surface provided a descriptive view of the altitude × N response of gluten index. In low-altitude regions (Renqiu and Beijing), gluten index increased more clearly under higher N, whereas the high-altitude Lhasa site showed a lower gluten index and weaker N response. Figure 10 presents the corresponding two-way interaction pattern. Fig. 10Two-way interaction effects of altitude above sea level and N rate on wheat gluten index. Two-way interaction effects of altitude above sea level and N rate on wheat gluten index. Wheat quality Grain quality results are summarized in Table 7. S-W showed the widest variation in several protein traits, whereas M-W and W-W were more stable across treatments. The sample size in Table 7 was 216 observations, calculated as 3 sites × 2 years × 3 gluten groups × 2 cultivars within each group × 2 N treatments × 3 field replicates. Table 7Statistical characteristics of wheat grain protein and its components under different treatments (n = 216; 3 sites × 2 years × 3 gluten groups × 2 cultivars × 2 N treatments × 3 replicates).Cultivar groupProteins contentTotal (%)Albumin (%)Globulin (%)Gliadin (%)Glutenin (%)Gli./Glu.S-WMax19.374.883.036.746.451.60Min14.191.811.263.934.380.67Range5.183.081.772.812.080.93Mean ± SD16.63 ± 1.683.36 ± 0.782.17 ± 0.465.28 ± 0.965.32 ± 0.581.05 ± 0.28CV(%)10.0923.2521.2818.1710.9426.40M-WMax14.653.582.264.495.651.67Min11.881.721.432.884.000.95Range2.771.860.831.611.650.72Mean ± SD13.43 ± 0.612.60 ± 0.371.81 ± 0.203.73 ± 0.464.83 ± 0.391.31 ± 0.18CV(%)4.5414.2311.0112.317.9813.76W-WMax12.542.681.863.625.502.47Min9.741.481.032.133.871.21Range2.811.200.831.501.641.26Mean ± SD11.15 ± 0.721.83 ± 0.271.32 ± 0.202.81 ± 0.354.70 ± 0.441.69 ± 0.25CV(%)6.4614.9815.1512.369.3714.73Max maximum, Min minimum, SD standard deviation, CV coefficient of variation. Values were calculated from the raw quality dataset; n = 216 was derived from 3 sites × 2 years × 3 gluten groups × 2 cultivars × 2 N treatments × 3 field replicates. Table 7Statistical characteristics of wheat grain protein and its components under different treatments (n = 216; 3 sites × 2 years × 3 gluten groups × 2 cultivars × 2 N treatments × 3 replicates).Cultivar groupProteins contentTotal (%)Albumin (%)Globulin (%)Gliadin (%)Glutenin (%)Gli./Glu.S-WMax19.374.883.036.746.451.60Min14.191.811.263.934.380.67Range5.183.081.772.812.080.93Mean ± SD16.63 ± 1.683.36 ± 0.782.17 ± 0.465.28 ± 0.965.32 ± 0.581.05 ± 0.28CV(%)10.0923.2521.2818.1710.9426.40M-WMax14.653.582.264.495.651.67Min11.881.721.432.884.000.95Range2.771.860.831.611.650.72Mean ± SD13.43 ± 0.612.60 ± 0.371.81 ± 0.203.73 ± 0.464.83 ± 0.391.31 ± 0.18CV(%)4.5414.2311.0112.317.9813.76W-WMax12.542.681.863.625.502.47Min9.741.481.032.133.871.21Range2.811.200.831.501.641.26Mean ± SD11.15 ± 0.721.83 ± 0.271.32 ± 0.202.81 ± 0.354.70 ± 0.441.69 ± 0.25CV(%)6.4614.9815.1512.369.3714.73Max maximum, Min minimum, SD standard deviation, CV coefficient of variation. Values were calculated from the raw quality dataset; n = 216 was derived from 3 sites × 2 years × 3 gluten groups × 2 cultivars × 2 N treatments × 3 field replicates. Statistical characteristics of wheat grain protein and its components under different treatments (n = 216; 3 sites × 2 years × 3 gluten groups × 2 cultivars × 2 N treatments × 3 replicates). Max maximum, Min minimum, SD standard deviation, CV coefficient of variation. Values were calculated from the raw quality dataset; n = 216 was derived from 3 sites × 2 years × 3 gluten groups × 2 cultivars × 2 N treatments × 3 field replicates. Fig. 11Comparison of wheat grain protein and its components under different planting environments and N treatments. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Nitrogen treatment data are from N240 and N180. Bars show mean values from 2019 to 2020 and 2020–2021; error bars indicate standard errors. Comparison of wheat grain protein and its components under different planting environments and N treatments. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Nitrogen treatment data are from N240 and N180. Bars show mean values from 2019 to 2020 and 2020–2021; error bars indicate standard errors. Figure 11 shows that total grain protein generally followed S-W > M-W > W-W across the two years. Across sites, the high-altitude Lhasa environment tended to increase total protein and non-gluten protein fractions, especially albumin and globulin, while gluten-forming fractions responded more variably to site and N treatment. Across the two growing seasons, the relative increase in albumin and globulin at Lhasa was more consistent than the response of gliadin and glutenin. This pattern suggested that altitude affected not only total protein concentration but also protein composition. Nitrogen treatment modified protein fractions within each gluten type, but the direction and magnitude differed among sites. For S-W, N240 generally supported glutenin accumulation better than N180, whereas for W-W the protein response was more stable across N rates. Table 8 shows that gluten type and site/elevation above sea level had significant effects on test weight, glutenin/gliadin ratio, grain hardness and flour extraction rate. N treatment had a significant effect on flour extraction rate, while most higher-order interactions were not significant. Test weight and flour extraction rate varied among gluten types and sites. In general, the low-altitude sites had higher test weight than Lhasa, whereas grain hardness tended to be greater at Lhasa. These trends indicated that high-altitude conditions changed both physical grain traits and milling-related quality. The average test weights of S-W, M-W and W-W were 775.8, 794.7 and 766.8 g/L, respectively. The average glutenin/gliadin ratios were 1.0, 1.3 and 1.7, and the average grain hardness indices were 64.6, 65.1 and 54.4, respectively. These values confirmed clear baseline quality differences among gluten types. Within each gluten type, Lhasa generally had lower test weight and altered glutenin/gliadin ratio compared with Renqiu and Beijing. Flour extraction rate showed smaller differences between M-W and W-W, indicating that altitude effects were more pronounced for grain density and protein balance than for milling percentage. Table 8The impact of nitrogen treatment and planting site on wheat grain quality among different cultivars.CultivarSiteNitrogen treatmentTest weight(g/L)Glutenin/gliadin ratioGrain hardness indexFlour extraction rate(%)S-WRenqiuN240799.54a1.08b62.74b63.94aN180798.13a1.15ab62.24b58.15cBeijingN240803.29a1.26a65.88a62.26abN180805.29a1.23a65.40a64.01aLhasaN240730.88b0.84c65.87a59.85bcN180732.96b0.77c66.07a57.17cCV (%)5.0926.403.7413.23M-WRenqiuN240822.67a1.22c65.14ab61.85aN180822.54a1.44a64.08b59.81aBeijingN240810.67a1.37ab65.50ab60.99aN180813.00a1.40ab64.82ab63.37aLhasaN240754.38b1.18c65.90a61.35aN180757.63b1.29bc65.53ab60.59aCV (%)4.5313.762.8113.87W-WRenqiuN240771.92a1.60bc52.54b61.08aN180772.79a1.84a51.30b59.84aBeijingN240785.13a1.69abc51.49b64.39aN180783.75a1.80ab50.33b63.33aLhasaN240747.33b1.67abc59.43a61.24aN180744.38b1.55c59.70a61.46aCV (%)3.5514.739.698.11Influence factor (F value)Cultivar (C)62.69**533.26**0.6610.06**Site (S)283.47**62.84**5.85**573.14**Nitrogen (N)0.063.192.0933.08**C × S15.33**28.78**1.0811.91**C × N0.170.230.790.86S × N0.030.932.733.60*C × S × N0.110.110.880.21Data are the mean values from 2019 to 2020 and 2020–2021.Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24).Bars show mean values from 2019 to 2020 and 2020–2021; error bars indicate standard errors. Table 8The impact of nitrogen treatment and planting site on wheat grain quality among different cultivars.CultivarSiteNitrogen treatmentTest weight(g/L)Glutenin/gliadin ratioGrain hardness indexFlour extraction rate(%)S-WRenqiuN240799.54a1.08b62.74b63.94aN180798.13a1.15ab62.24b58.15cBeijingN240803.29a1.26a65.88a62.26abN180805.29a1.23a65.40a64.01aLhasaN240730.88b0.84c65.87a59.85bcN180732.96b0.77c66.07a57.17cCV (%)5.0926.403.7413.23M-WRenqiuN240822.67a1.22c65.14ab61.85aN180822.54a1.44a64.08b59.81aBeijingN240810.67a1.37ab65.50ab60.99aN180813.00a1.40ab64.82ab63.37aLhasaN240754.38b1.18c65.90a61.35aN180757.63b1.29bc65.53ab60.59aCV (%)4.5313.762.8113.87W-WRenqiuN240771.92a1.60bc52.54b61.08aN180772.79a1.84a51.30b59.84aBeijingN240785.13a1.69abc51.49b64.39aN180783.75a1.80ab50.33b63.33aLhasaN240747.33b1.67abc59.43a61.24aN180744.38b1.55c59.70a61.46aCV (%)3.5514.739.698.11Influence factor (F value)Cultivar (C)62.69**533.26**0.6610.06**Site (S)283.47**62.84**5.85**573.14**Nitrogen (N)0.063.192.0933.08**C × S15.33**28.78**1.0811.91**C × N0.170.230.790.86S × N0.030.932.733.60*C × S × N0.110.110.880.21Data are the mean values from 2019 to 2020 and 2020–2021.Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24).Bars show mean values from 2019 to 2020 and 2020–2021; error bars indicate standard errors. The impact of nitrogen treatment and planting site on wheat grain quality among different cultivars. Data are the mean values from 2019 to 2020 and 2020–2021. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Bars show mean values from 2019 to 2020 and 2020–2021; error bars indicate standard errors. Fig. 12Effects of planting environment and N treatment on flour quality traits of wheat grains. Data are the mean values from 2019 to 2020 and 2020–2021. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Bars show mean values from 2019 to 2020 and 2020–2021; error bars indicate standard errors. Effects of planting environment and N treatment on flour quality traits of wheat grains. Data are the mean values from 2019 to 2020 and 2020–2021. Cultivar data are from S-W (the mean values from ZM-578 and ZM-5051), M-W (the mean values from ZM-82 and ZM-89) and W-W (the mean values from YM-15 and YM-24). Bars show mean values from 2019 to 2020 and 2020–2021; error bars indicate standard errors. Flour quality traits in Fig. 12 differed among gluten types, sites and N treatments. Wet gluten content and dry gluten content were generally higher under N240 than under N180, but gluten index and sedimentation value were more sensitive to site/elevation above sea level than to N rate alone. Under N240, wet gluten content increased in several gluten types, but the magnitude of the response depended on site and cultivar group. The highest wet gluten values were generally observed in S-W, whereas W-W maintained lower gluten content as expected from its end-use classification. Planting site and N treatment both affected wet gluten content. However, the high-altitude Lhasa environment reduced gluten functionality even when total gluten or protein increased, showing that protein composition and dough performance must be considered together. Dough rheological traits in Table 9 were significantly affected by gluten type, site/elevation above sea level and their interaction. N treatment significantly affected stability time, but its effect was smaller than the site and gluten-type effects. All time-related parameters were expressed in minutes (min). Table 9Effects of N treatment and planting site on dough rheological properties of different gluten types.CultivarSiteNitrogen treatment14% water absorption rate(%)Dough formation time(min)Stability time(min)Weakening degree ICC (FU)Farinograph indexS-WRenqiuN24065.02b11.25a9.68b79.58b168.92aN18064.86b10.69a9.04b83.17b161.50aBeijingN24063.75c11.85a12.95a53.75b193.50aN18063.80c9.28a8.09b58.50b176.33aLhasaN24069.85a3.89b2.89c201.50a58.58bN18070.27a3.52b2.35c216.25a51.33bCV (%)4.5866.1763.8168.3963.33M-WRenqiuN24060.68b6.71a8.03a88.33b107.17aN18060.83b6.03ab7.49a93.42b96.33aBeijingN24059.57c4.89ab9.53a57.42c120.67aN18059.54c4.67bc7.66a68.00c104.58aLhasaN24066.15a2.98cd1.99b205.58a41.00bN18066.29a2.68d1.75b210.67a37.50bCV (%)5.0785.54104.2157.9481.16W-WRenqiuN24054.79b4.13a8.77a75.08b98.67aN18054.86b3.80a8.12ab78.25b92.83aBeijingN24053.22c2.48bc7.22bc69.08b74.25bN18052.65c2.37c6.62c72.58b65.75bcLhasaN24061.91a4.23a3.63d156.17a60.25cdN18062.46a3.33ab2.93d166.00a53.17dCV (%)7.4247.9843.5346.4032.76Influence factor (F value)Cultivar (C)1128.60**53.15**3.82*5.76**35.16**Site (S)748.42**27.23**74.33**337.47**56.36**Nitrogen (N)0.182.656.47*2.342.15C × S11.38**13.02**2.62*7.02**9.76**C × N0.030.360.820.020.03S × N0.940.131.840.160.14C × S × N0.360.490.830.100.03 Table 9Effects of N treatment and planting site on dough rheological properties of different gluten types.CultivarSiteNitrogen treatment14% water absorption rate(%)Dough formation time(min)Stability time(min)Weakening degree ICC (FU)Farinograph indexS-WRenqiuN24065.02b11.25a9.68b79.58b168.92aN18064.86b10.69a9.04b83.17b161.50aBeijingN24063.75c11.85a12.95a53.75b193.50aN18063.80c9.28a8.09b58.50b176.33aLhasaN24069.85a3.89b2.89c201.50a58.58bN18070.27a3.52b2.35c216.25a51.33bCV (%)4.5866.1763.8168.3963.33M-WRenqiuN24060.68b6.71a8.03a88.33b107.17aN18060.83b6.03ab7.49a93.42b96.33aBeijingN24059.57c4.89ab9.53a57.42c120.67aN18059.54c4.67bc7.66a68.00c104.58aLhasaN24066.15a2.98cd1.99b205.58a41.00bN18066.29a2.68d1.75b210.67a37.50bCV (%)5.0785.54104.2157.9481.16W-WRenqiuN24054.79b4.13a8.77a75.08b98.67aN18054.86b3.80a8.12ab78.25b92.83aBeijingN24053.22c2.48bc7.22bc69.08b74.25bN18052.65c2.37c6.62c72.58b65.75bcLhasaN24061.91a4.23a3.63d156.17a60.25cdN18062.46a3.33ab2.93d166.00a53.17dCV (%)7.4247.9843.5346.4032.76Influence factor (F value)Cultivar (C)1128.60**53.15**3.82*5.76**35.16**Site (S)748.42**27.23**74.33**337.47**56.36**Nitrogen (N)0.182.656.47*2.342.15C × S11.38**13.02**2.62*7.02**9.76**C × N0.030.360.820.020.03S × N0.940.131.840.160.14C × S × N0.360.490.830.100.03 Effects of N treatment and planting site on dough rheological properties of different gluten types. Under N180, dough development time and stability time were shortened for most gluten types, while weakening degree increased and farinograph index decreased. These responses indicated that lower N supply generally weakened dough strength. Water absorption differed among gluten types. S-W had the narrowest range, whereas W-W showed the widest range across environments. These differences reflected the contrasting protein composition and end-use quality of the three gluten groups. Compared with Beijing, Lhasa markedly shortened dough development time and stability time and increased weakening degree. For example, S-W dough stability time decreased from 10.5 min in Beijing to 2.6 min in Lhasa, while weakening degree increased strongly. These results indicated that high-altitude conditions reduced dough strength even when grain protein concentration increased. In summary, high-altitude planting shortened dough formation time in S-W and M-W, shortened dough stability time in all three gluten types, reduced the farinograph index and increased weakening degree. This confirmed that the high-altitude protein increase did not necessarily improve processing quality. Fig. 13PCA biplot of wheat yield and quality traits across sites/elevations above sea level. PC1 explained 62.1% of the variance and was associated mainly with site/elevation above sea level, TRG and protein-related traits. PC2 explained 15.9% of the variance and was associated mainly with N treatment and TGW. Colors indicate gluten type and shapes indicate site/elevation group. PCA biplot of wheat yield and quality traits across sites/elevations above sea level. PC1 explained 62.1% of the variance and was associated mainly with site/elevation above sea level, TRG and protein-related traits. PC2 explained 15.9% of the variance and was associated mainly with N treatment and TGW. Colors indicate gluten type and shapes indicate site/elevation group. The PCA scores plot (Fig. 13) showed that Lhasa (3688 m) was clearly separated from Renqiu (4 m) and Beijing (40 m), while Renqiu and Beijing were closer to each other. This separation confirmed that the plateau environment created a distinct yield-quality pattern. PC2 was mainly related to N amount and thousand-grain weight. The biplot indicated that N management contributed to quality and yield variation, but the separation caused by site/elevation above sea level was stronger. Fig. 14Correlation analysis among meteorological factors, yield components and wheat quality traits. SRG sunshine duration during the reproductive growth period, TRG accumulated temperature during the reproductive growth period, P precipitation from emergence to maturity, RRG ratio of reproductive growth period, TGW thousand-grain weight, PC protein content, Glu glutenin, Gli gliadin, GI gluten index, ST stability time, WD weakening degree. Correlation analysis among meteorological factors, yield components and wheat quality traits. SRG sunshine duration during the reproductive growth period, TRG accumulated temperature during the reproductive growth period, P precipitation from emergence to maturity, RRG ratio of reproductive growth period, TGW thousand-grain weight, PC protein content, Glu glutenin, Gli gliadin, GI gluten index, ST stability time, WD weakening degree. Correlation analysis among meteorological factors, yield traits and quality traits is shown in Fig. 14. Because only three site altitudes were included, these correlations were used to describe associations among measured variables and to guide interpretation, rather than to establish a universal altitude-response function. Total rainfall during the whole growth period was positively associated with several protein fractions, whereas yield components were more closely related to reproductive-stage temperature and sunshine variables. These results suggested that yield and quality were driven by partly different environmental factors. The correlations differed among gluten types, indicating that S-W, M-W and W-W responded differently to reproductive-stage meteorological conditions. These differences supported the genotype-specific interpretation of the altitude response. Spike number and grains per spike were positively associated with grain yield, whereas some dough-related traits were negatively associated with the environmental variables characterizing Lhasa. These associations were consistent with the observed yield-quality trade-off. Based on the correlations in Fig. 14, a structural equation model (SEM; Fig. 15) was constructed to clarify the hierarchical regulation of yield and quality traits. The model indicated that meteorological factors during reproductive growth influenced thousand-grain weight and protein-related quality traits through both direct and indirect pathways. Fig. 15Summarizing model of the effects of altitude above sea level, N treatment and meteorological factors on wheat yield and quality. Summarizing model of the effects of altitude above sea level, N treatment and meteorological factors on wheat yield and quality. The SEM further showed that reproductive-stage sunshine duration, accumulated temperature and precipitation were important drivers of yield-quality variation. The model provided a conceptual summary of how altitude-related environmental conditions affected yield components, protein composition and processing quality. Discussion High-altitude effects on phenology and morphology: beyond simple delay Our findings confirmed that high altitude delayed wheat development and reduced plant stature (Fig. 7; Table 4), consistent with the temperature-limitation hypothesis11. The differential responses among gluten types suggested that W-W had a more stable developmental strategy under plateau conditions. Low temperature and large diurnal temperature fluctuation can influence starch synthesis and grain filling22. Reduced spike number per m² at Lhasa indicated impaired tiller initiation or survival during the jointing stage, whereas reductions in spike length and fertile spikelet number indicated constraints on spike development. Deciphering the yield-quality trade-off: Altitude × nitrogen × genotype The yield penalty at Lhasa was driven by limitations in sink size (spikes/m² and grains per spike) and sink strength (thousand-grain weight). Low temperature can restrict photosynthetic capacity during grain filling and reduce assimilate transport and starch deposition15,28. S-W required higher N input to express its yield and gluten-quality potential, whereas M-W had a wider N optimum. The idea of a gluten-type-specific altitudinal adaptation range was supported by genotype × site interactions, but it should be considered a working hypothesis rather than a fully established concept because this study tested only three altitudes above sea level. The high-altitude protein paradox: Composition matters more than quantity The high-altitude-induced increase in total grain protein (+ 17.5% compared with Beijing, Table 7; Fig. 11) was consistent but potentially misleading. Fractionation showed that the increase was driven mainly by albumin and globulin. Although these fractions contribute to nutritional value, they contribute less to the viscoelastic gluten network required for bread-making than gliadin and glutenin29,30. In this manuscript, the term “high-altitude protein paradox” is used as our descriptive interpretation of this pattern: total protein increased at high altitude, but gluten functionality and dough processing quality decreased. The reduced glutenin/gliadin balance and poor dough rheology at Lhasa showed that protein composition and polymerization quality were more important than total protein content alone. Climate drivers and management implications: Towards precision agriculture The correlation and SEM analyses linked specific climatic variables to phenotypic responses (Figs. 14 and 15). The associations between reproductive-stage meteorological factors and yield components suggested that the high-altitude response was driven by the combined effects of thermal resources, radiation and water conditions. These analyses should be interpreted within the tested sites rather than extrapolated to all altitudinal zones. These findings support the hypothesis of gluten-type-specific altitudinal adaptation. W-W showed better stability under high-altitude stress, whereas the quality advantages of S-W were expressed more fully at low-altitude sites with adequate N input. This genotype-environment-management framework highlights the need to optimize cultivar selection and N management together. Cultivar selection should be prioritized according to altitude above sea level and production target. W-W was more suitable for high-altitude production when stability was emphasized, whereas S-W was more appropriate for low-altitude production aimed at stronger gluten quality. Precision N management is required rather than blanket N recommendations. For high-altitude regions near 3000 m or above, W-W combined with 180 kg/ha N provided a comparatively stable yield-quality balance in this study, with mean yield of approximately 409.8 g/m² and grain protein content of approximately 11.7%. For low-altitude plains, S-W with 240 kg/ha N maintained stronger gluten quality. For M-W, the 180–240 kg/ha range provided more flexibility. High-altitude S-W cultivation is not recommended when the target is strong bread-making quality. Future breeding should target genotypes with enhanced tolerance to low-temperature stress, improved N use efficiency and stable storage protein composition, especially high-molecular-weight glutenin subunits that help maintain dough strength under environmental stress. Recent work on diverse winter wheat evaluated with both traditional measurements and machine-learning-assisted phenotyping also supports the value of integrating field traits, image-based traits and computational analysis for climate-stress adaptation31. These targets may help reduce the risk of high protein but low processing quality under plateau conditions. This study was limited to three altitudes above sea level (4 m, 40 m and 3688 m) and did not include transitional zones between 2000 and 3000 m. Future studies should include more intermediate altitudes, additional years and molecular analyses of gluten protein polymerization to test the proposed adaptation hypothesis more rigorously. Conclusion This study showed that altitude above sea level, N treatment and gluten type jointly regulated wheat yield and quality. The main conclusions are summarized as follows: High altitude prolonged the growth period but inhibited yield formation. High altitude prolonged the growth period but inhibited yield formation. High altitude prolonged the growth period but inhibited yield formation. At Lhasa (3688 m), the whole growth period was prolonged by 13.2–14.6%, while spike number, grains per spike and thousand-grain weight decreased. Mean grain yield at Lhasa was 44.1% lower than at Beijing (444.4 vs. 794.9 g/m²). W-W showed stronger stability and adaptability under high-altitude conditions. (2)Nitrogen requirements differed among gluten types. (2)Nitrogen requirements differed among gluten types. Nitrogen requirements differed among gluten types. S-W achieved better yield and gluten-protein synthesis at 240 kg/ha N. M-W performed similarly within 180–240 kg/ha N. W-W maintained a more stable yield-quality balance at 180 kg/ha N under high-altitude conditions. (3)High altitude caused a “high protein but weak processing quality” pattern. (3)High altitude caused a “high protein but weak processing quality” pattern. High altitude caused a “high protein but weak processing quality” pattern. Although total grain protein increased at high altitude by 17.5% compared with Beijing, the increase was mainly associated with non-gluten protein fractions, especially albumin and globulin. Gluten functionality declined, as shown by lower gluten index, shorter dough stability time, higher weakening degree and poorer processing quality, especially in S-W and M-W.