Influence of leaf inclination angle and tillering on population transpiration, soil evaporation, and yield in winter wheat near-isogenic lines

Experimental site

The experiments were conducted at the Shunyi Scientific Experimental Station, which is located in the Beijing area, the northern part of the North China Plain, affiliated with the Institute of Environment and Sustainable Development in Agriculture, Chinese Academy of Agricultural Sciences (CAAS), in the 2015–2016, 2016–2017, and 2017–2018 growing seasons of winter wheat. The experimental field’s soil was classified as calcareous fluvo-aquic (sand 28.7%, silt 64.2%, and clay 7.1%), exhibited a soil bulk density of 1.40 g cm⁻³, and a pH of approximately 7.7. The soil nutrient levels were as follows: 0.109 g/kg of total nitrogen, 14.4 g/kg of organic matter, 24.5 mg/kg of available phosphorus, and 106 mg/kg of available potassium. The previously planted crop was field peas (Pisum sativum spp. arvense).

Table 1 outlines the monthly mean air temperature and precipitation for the three experimental growing seasons. The highest total precipitation occurred in 2015–2016 (173.4 mm), followed by 2017–2018 (130.1 mm), with 2016–2017 being the driest at 119.6 mm. April, the largest water requirement period of wheat crops in the North China Plain, witnessed meager precipitation in April 2016 (5.6 mm) and 2017 (0.1 mm), but April 2018 recorded 52.7 mm, surpassing long-term average (20.6 mm) by 32.1 mm. The monthly mean air temperatures were relatively higher in 2016–2017, lower in 2017–2018, and moderate in 2015–2016. Across the three years, temperatures were consistently higher from March to June but notably lower in November compared to the long-term average.

Experimental design

Winter wheat cultivars Jinmai 47 and Jing 411, used as the recurrent and donor parents, respectively, along with their six near-isogenic lines (NILs) 908092, 908120, 908188, 908206, 908032, and 908274, were chosen as experimental materials. They were obtained from the Institute of Crop Sciences, CAAS. Jinmai 47 is a loose, flat-leafed type with an LIA of 65°, and Jing 411 is a compact, upright-leafed type with an LIA of 35°, both of which were widely planted in the North China Plain. These parents and their six NILs were selected based on LIA, forming a gradient from 35° in Jing 411 to 65° in Jinmai 47.

The eight genotypes were planted on October 5, 2015, October 3, 2016, and October 3, 2017, and were harvested on June 12, 2016, June 11, 2017, and June 14, 2018, respectively. The seeding rate was 675 seeds m⁻², which is the current conventional and widely adopted sowing density in the target region. Two water treatments were implemented: rain-fed and supplemental irrigation treatments. A 2.5-m-wide isolation belt separated the two water treatments, with winter wheat growing in the 1.5-m-wide middle area. Before sowing, all treatments received 90-mm irrigation was fulfilled for all treatments to ensure uniform soil water conditions and seedling emergence. For the rain-fed treatment, no irrigation was applied during the winter wheat-growing season. For the supplemental irrigation treatment, each plot received 120-mm irrigation via sprinkling water, 60 mm each before jointing and before flowering stages. The eight materials were arranged in a randomized block design with three replicates in each water treatment area, and each plot measured 10.0 × 2.4 m².

LIA and SD

The LIA, representing the angle between the vertical and the leaf plane extension, was measured. The preliminary measurements indicated similarity in the flag leaf LIA between main shoots and tillers in one wheat plant. Hence, LIA measurements were conducted on ten main shoot flag leaves in the central five rows of each plot using a transparent plastic protractor between 10:00 and 11:00 a.m. on windless days at anthesis stage. SD investigation involved manually counting shoots per m² in the central five rows at anthesis stage.

WU traits

Total WU or ET, E, and PT were determined for the period from jointing to early grain filling stage, representing the peak water requirement period for wheat crops, in 2016, 2017, and 2018.

ET was calculated from the initial soil water content minus final soil water content, precipitation, irrigation, runoff, and drainage using the following equation (Li et al., 2012): ${\displaystyle ET=I+P-R-D-SW}$ where ET is evapotranspiration (mm) from jointing to early grain filling stages; I is irrigation (mm); P is precipitation (mm); R is the surface runoff (mm); and D is downward drainage (mm). R and D were assumed to be 0 due to the low precipitation and deep percolation under the experimental conditions. SW denotes changes in water storage for the 1.6-m soil profile. SW was determined as previously described by Huang et al. (2023). Specifically, the gravimetric soil water contents were measured at 20-cm intervals to a depth of 160 cm by taking soil cores, and then converted into volumetric soil water contents using bulk density data. To mitigate the influence of soil spatial heterogeneity, three sampling points between two rows were randomly arranged in each plot, and samples were taken and fully mixed to obtain a representative sample for each soil layer in each plot. Measurements for gravimetric soil water contents were conducted on April 17 and May 11 in 2016, April 18 and May 12 in 2017, and April 20 and May 11 in 2018.

The E was determined by the micro-lysimeter method as previously described in Huang et al. (2023). Specifically, the micro-lysimeters were made of polyvinyl chloride cylinders (PVC) (15 cm length and 10 cm inner diameter), and slightly larger PVC cylinders with 12 cm inner diameter and the same length with the instrument serving as the sleeves to facilitate insertion and removal of the micro-lysimeters. The sleeve was pushed into the ground, and emptied of soil prior to installing the micro-lysimeter. Subsequently, the micro-lysimeter was pushed into the soil between two rows within the same plot, carefully taken out, and the soil outside it was cleaned, with the bottom being sealed with water-proof adhesive tape, and then weighed and put into the sleeve. Each plot had three micro-lysimeters installed between rows. For water loss determination, the soil inside the micro-lysimeter was weighed daily at dusk using an electronic balance. The soil inside the micro-lysimeter was changed every three days, and immediately after rainfall or irrigation, to maintain similar soil water contents to the surrounding soil. Soil evaporation was assumed negligible on rainy days. Finally, soil evaporation was calculated based on the daily weight change and the micro-lysimeters’ surface area.

PT was finally calculated by subtracting E from ET.

Yield

Following winter wheat maturation, a 1-m section from the center of each plot, spanning five rows, was randomly hand-harvested. The harvested crop was air-dried and then the grain yield (kg ha⁻¹) was calculated.

Statistical analysis

Statistical analysis was performed using SAS procedure (Version 9.2, SAS Institute, Cary, NC, USA). The results for rain-fed and supplemental irrigation treatments were evaluated separately, because the two water treatments were not randomized within the same study, instead, they were separated by a 2.5-m-wide isolation belt, thus constituting distinct experiments. Additionally, the results for each growing season were analyzed separately, because differences in weather data existed between years, also because the initial combined data showed genotype by year interaction (Dong et al., 2006; Steel & Torrie, 1980; Dong et al., 2010). Based on these, a one-way ANOVA (analysis of variance) was carried out in experimental indices (SD, three WU traits, and yield) under each water condition per growing season to primarily evaluate the differences across eight genotypes. The means of the eight genotypes were compared using the least significant differences at a 0.05 probability level. The regression analyses between experimental indices and WU traits, as well as yield were conducted using Pearson correlation analysis and single regression line.

LIA and tillering of different wheat genotypes

In considering flag leaf LIA, the recurrent (Jinmai 47) and donor (Jing 411) parents, along with their six NILs, were categorized into three groups. Group I, comprising Jing 411 and NILs 908274 and 908092, exhibited smaller LIA. In contrast, Group III, including Jinmai 47 and NILs 908188 and 908032, had larger LIA. Group II, with medium LIA, encompassed 908120 and 908206 (Table 2).

Under a consistent sowing density of 6.75 × 10⁶ plants ha⁻¹, the parents Jinmai 47 and Jing 411, along with their six NILs, showed varying tillering, leading to different SD. Jinmai 47 and NILs 908206, 808032, and 908274 each displayed higher SD, whereas Jing 411 and NILs 908092, 908120, and 188 each exhibited lower SD, irrespective of water conditions and growing years (Table 3).

Considering both LIA and SD, certain genotypes demonstrated similar LIA but significantly different SD, such as the pair 908274 and Jing 411. Conversely, some had statistically similar SD but different LIA, exemplified by the following five pairs: Jinmai 47 and 908274, Jing 411 and 908120, 908188 and 908206, 908092 and 908120, and 908206 and 908032. These valuable materials enabled a separate analysis of the effects of LIA and SD on WU parameters. Specifically, the effect of LIA on WU could be compared under similar SD, and the impact of SD on WU could be assessed with similar LIA.

Population WU traits of different wheat genotypes

Genotypic variations in WU parameters (ET, PT, and E) were primarily linked to SD differences. Generally, lower SD genotypes (908092, Jing 411, and 908120) consistently ranked highest in WU parameters ET, PT, and E, whereas higher SD genotypes (908032, Jinmai 47, and 908274) consistently ranked lowest. Genotypes 908188 and 908206 exhibited moderate rankings (Table 4). In the narrower SD range resulting from tillering-related genotypic difference, ET, PT, and E exhibited a declining trend with increasing SD.

The correlation analysis indicated that extremely significant negative correlations existed between ET and SD, with an exception in the 2016–2017 growing season under rain-fed condition (R = −0.6065). Conversely, no significant correlation existed between ET and LIA (Table 5). Genotypic comparison further highlighted the superior effect of SD and the negligible effect of LIA on WU (Table 6). Specifically, genotypes with similar LIA and SD exhibited similar ET, while those with similar LIA but differing SDs showed different ETs. Notably, three pairs of genotypes with differing LIA but similar SD each showed comparable ET. An exception was observed in the pair of 908206 and 908032, with similar SD but an 11° LIA difference resulting in different ET. PT showed no correlation with LIA over the three years under two water treatments. In contrast, PT was negatively correlated with SD, and these correlations varied across experimental years. In the driest 2016–2017 growing season, the correlations were not significant. However, in the wettest 2017–2018 growing season, the coefficients reached as high as −0.8496 under the rain-fed condition and −0.8602 under the supplemental irrigation condition. It seemed that the effect of SD on PT were more significant when the plentiful rainfall ensured a better growth of wheat crops. However, the coefficients between SD and PT were not higher under the supplemental irrigation condition than under the rain-fed condition.

Compared to PT, E showed closer associations with both LIA and SD. Especially under supplemental irrigation condition, negative correlations existed between SD and E in all three growing seasons and even between LIA and E in two seasons. In the 2015–2016 growing season in particular, the negative correlations between SD and E reached extremely significant levels under different water treatments.

Significant correlation of LIA with yield

The general trend revealed that Group I genotypes with smaller LIA correlated with lower yields, Group III genotypes with larger LIA had higher yields, and Group II genotypes with medium LIA had moderate yields (Table 7). The correlation analysis confirmed the positive correlation between LIA and yield across the genotypes selected based on LIA (Table 8), indicating an influence of LIA on yield. However, SD did not correlate with yield, indicating that genotypic difference in SD related to tillering did not affect yield at the current sowing density.

A strong negative correlation of SD with ET across different tillering genotypes

WU is influenced by diverse factors encompassing crops, soil and atmosphere. A higher SD implies more shoots utilizing water, resulting in increased ET. However, in this study, SD was negatively correlated with ET (P > 0.01) in most trials across three growing seasons under two water treatments. No relevant data on WU correlations with canopy structure factors are currently available. The negative correlation was observed between SD and ET across different genotypes rather than within the same genotype across different SDs. This negative correlation between SD and ET was merely found in such narrow SD range caused by variations in tillering ability among genotypes. However, within a larger range, SD might not negatively correlate with ET for the same genotype. Our separate study, investigating the effects of sowing density on WU, incorporated three SDs in a broader range (Huang et al., 2024). The findings indicted a general increase in WU with increasing SD (see Table S1). The negative correlation, somehow, suggested that a higher SD does not necessarily result in a greater WU. The underlying mechanism, specific SD range for the negative correlation, and its influencing factors need to be further explored by setting up a more detailed SD gradient.

Negative correlation of SD with PT implied the predominant effect of canopy structure

SD was negatively correlated with PT, indicating an unexpected decrease in PT with higher SD. This negative correlation implies the presence of an alternative factor influencing PT beyond SD. Rather than transpiration of a single leaf or an individual plant, PT is influenced by canopy structure, creating resistance that regulates air-canopy water and heat exchange. A canopy structure with better permeability facilitates increased transpiration in a population (Johnson, Witters & Ciha, 1981). With similar leaf stomatal aperture, genotypic difference in leaf transpiration rate primarily arises from canopy structure-related variations. Thus, the PT genotypic difference is not only affected by leaf stomatal conductance but is strongly affected by canopy structure. SD emerges as a pivotal factor shaping canopy structure, with higher SD resulting in a more complete canopy, augmenting resistance to water and heat dissipation, and consequently suppressing PT. Thus, the negative correlation suggested the substantial impact of canopy structure related to SD.

Antagonism between the SD-E and SD-T correlations

The higher SD results in a more complete canopy, enhancing efficiency in preventing soil surface evaporation. Blum (2009) suggested that a more complete canopy moderated transpiration from the canopy and especially prevented evaporation from the soil surface (Blum, 2009). Studies on clump planting of grain sorghum (Sorghum bicolor L. Moench) and maize demonstrated increased canopy closure and reduced WU at the vegetative growth stage (Kapanigowda et al., 2010). Consistently, E decreased with increasing SD, and thus negatively correlated with E in the present study.

SD–E and SD–PT correlations varied across growing seasons, displaying apparent antagonism. In the 2017–2018 growing season, SD-PT correlation was higher, while SD-E correlation was lower compared to the other two growing seasons. Although PT and E were essentially antagonistic, they both are affected by soil water conditions and underwent increase under better water availability. Higher rainfall in April 2018 ensured increased wheat crops growth, which certainly prevented E and enhanced PT. If the higher SD–PT correlation in 2018 could be solely attributed to improved crop growth, then under the supplemental irrigation condition, one could expect a higher SD–PT correlation and a lower D–E correlation, assuming better crop growth due to enhanced water availability. However, the study revealed a relatively higher SD–E correlation under supplemental irrigation condition, challenging the expectation. Therefore, the improved crop growth resulting from enhanced water availability could not explain the higher SD–PT and lower SD–E correlations. The observed variation patterns in SD–PT and SD–E correlations across growing seasons became a perplexing phenomenon.

Strong correlation of LIA with yield for wheat crops

LIA affects canopy foliation arrangement, influencing its light interception. Some previous studies indicated that the upright-leafed genotypes with smaller LIA facilitated deeper penetration and a more uniform distribution of light within the canopy, enhancing photosynthetic efficiency (Virk, Khush & Peng, 2004; Zheng et al., 2008). In cotton, since the earlier fruits are on the lower parts of the plant, and the ability of older leaves to produce sufficient photosynthates in concert with fruit requirements limits potential yield, adequate light penetration into the crop canopy is particularly important (Wullschleger & Oosterhuis, 1990; Marois et al., 2004). Contrary findings by Herbert (1991) and Herbert (1992) indicated that a more erect upper foliage distribution promoted deeper light penetration but reduced light interception by upper layers. Additionally, Falster & Westoby (2003) utilized digitizing technology and a three-dimensional architecture model to analyze 38 perennial species only to correlate architectural properties (leaf angle and leaf size) with light capture and simulate carbon gain. They found that shallow-angled leaves had greater total daily light interception and potentially greater day-long carbon gain than steeper-angled leaves.

Differing from the lower plant regions where earlier fruits are found in cotton, the uppermost part in wheat is wheat ear. The yield formation in cereals primarily depends on photosynthates from younger leaves after anthesis, contributing to 70 −80% of grain yield (Prasad et al., 2008). In particular, the flag leaf serves as the major photosynthetic apparatus in later developmental stage, with its contribution to grain yield exceeding one-third (Murchie et al., 1999). Considering these findings, wheat genotypes with larger LIA should enhance light interception and carbon gain. In our study, LIA positively correlated with yield within an LIA range of 35° through 65° across the genotypes with similar genetic backgrounds (six NILs and their parents), indicating its potential influence on light interception and, consequently, yield. This implied the potential of LIA trait in wheat yield improvement. However, cautions should be exercised in drawing general conclusions about the impact of LIA on wheat yield, as the results are based on limited genotypes. Therefore, further investigations with a broader range of cultivars featuring diverse LIA values are needed to substantiate these findings.