In Anyang (36°06′ N and 114°21′ E), Henan, China, Chinese Academy of Agricultural Sciences’ Institute of Cotton Research carried out a two-year pool-culture study from 2019 to 2020. Each pool was 3.6-m-long, 4-m-wide and 1.5-m-high. The experimental soil type is classified as Inceptisols (USDA Soil Taxonomy). Clay loam was the soil used in the experiment (Li et al., 2017). Among soil layers within 0−20 cm, the organic matter, total nitrogen, and available nitrogen, phosphorus, and potassium were respectively 12.9 g kg⁻¹, 0.86 mg kg⁻¹, 64.4 mg kg⁻¹, 3.1 mg kg⁻¹, and 163.6 mg kg⁻¹ in 2019 and 13.1 g kg⁻¹, 0.85 mg kg⁻¹, 63.3 mg kg⁻¹, 3.0 mg kg⁻¹, and 180.4 mg kg⁻¹ in 2020. Meteorological data of the cotton growing seasons in 2019 and 2020 are presented in Table 1 .

The research used the split-plot design, with the main plot consisting of the low-P-tolerant cotton cultivar CCRI-79 and the low-P-sensitive cultivar SCRC-28, selected in the previous study (Sun et al., 2022). The subplots contained three soil AP levels; 3 ± 0.5 mg kg^-1 (P₀, extreme soil AP deficiency), 6 ± 0.5 mg kg^-1 (P₁, moderate soil AP deficiency), and 15 ± 0.5 mg kg^-1 (P₂, control). The soils from 20–40 cm depth in the field were chosen to develop P-deficiency in pool soil. Soil AP levels were regulated using triple superphosphate (44% P₂O₅) (Sun et al., 2022). The P fertilizer amount of 0, 50.6, and 202.4 g pool⁻¹ was applied in P₀, P₁, and P₂ during both years. The nitrogen fertilizer used was 225 kg N ha^-1 (urea, 46% N), and 50% of the fertilizer was used for basal application before sowing (April 23, 2019, and April 15, 2020), while the other 50% for topdressing in the early flowering stage (July 22, 2019, and July 6, 2020). Furthermore, the potassium fertilizer used was 150 kg K₂O ha^-1 (potassium sulfate, 51% K₂O) and was applied as basal fertilizer (Li et al., 2017). We applied the base fertilizer 7 days before sowing, and we watered the soil to dissolve it. An analysis of soil samples at the 0−20 cm soil layer was conducted one month after the application of base fertilizer at the surface level of the soil, after which the samples were dried and sieved at a fineness of 1 mm. The soil AP contents were determined by the Olsen-P method (Yang and Jacobsen, 1990). Thereafter, cotton seeds were planted through manual drilling (April 30, 2019, and April 24, 2020). Five rows of 80 cm spacing were used in each experimental plot with an area of 14.4 m² and a density of 52,500 plants per hectare. Every treatment was conducted in three replicates, and field management was in accordance with the management measures for high-yielding cotton cultivation.

During the flowering period of cotton plants, 8−10 cotton bolls of the first fruit node in the lower fruiting branches (FB_2-3), middle fruiting branches (FB_6-7), and upper fruiting branches (FB_10-11) were collected at 5, 10, 15, 17, 24, 31, 38, and 45 DPA (8:00-9:00 a.m.). The harvested cotton bolls were placed on ice, and the cotton fibers were peeled off from the cotton seeds within one hour. Thereafter, we analyzed the malate content and enzyme activity of one third (1/3) of the cotton fibers that were frozen in liquid nitrogen and held at -80°C. Another 1/3 was dried at 70°C to a constant weight to determine K⁺ and carbohydrate contents. The remaining 1/3 was used for fiber length measurement.

After crushing with a disintegrator, dried cotton fiber samples were sieved at 0.5 mm. Thereafter, K⁺ was extracted from the samples using the H₂SO₄-H₂O₂ digestion method, and its content was determined by an atomic absorption spectrometer (NOVAA400P, Analytik Jena GmbH, Jena, Germany) at 769.9 nm.

After snap-freezing in liquid nitrogen, 50 mg of frozen fiber samples were ground into powder, and a 1-hour extraction at 80°C was performed on the powder using 1.5 ml of the buffer containing 1.2 ml of absolute ethanol, 100 mM Hepes-KOH (pH 7.1), and 20 mM MgCl₂ to obtain the crude extract. We centrifuged the crude extract at 12000×g (5 min), and the supernatant was collected and mixed with 150 μl of active carbon (100 mg ml^-1). The mixture was centrifuged under the same conditions, and the supernatant was collected. Finally, 5 ml of the supernatant was transferred into a 10 ml cuvette, and 1 ml of Tris-HCl (0.2 mol L^-1 Tris, 0.2 mol L^-1 HCl, and pH 5.36) was added and mixed. Incubate for 15 min with distilled water topped up to 25 ml. The malate contents were measured on the spectrophotometer (SPECORD 40, Analytik Jena GmbH, Jena, Germany) at 656 nm using the colorimetric method (Zhang et al., 2017).

Weighed (0.1 g) cotton fibers were placed in a 10 ml centrifuge tube with 5 ml of 80% ethanol (v/v) and incubated at 80°C for carbohydrates extraction (30 min). The extracts were centrifuged (10000×g, 5 min), and a 25 ml volumetric flask was used to collect the supernatant. Repetition of the extraction procedure was performed, and the obtained supernatant was topped up to 25 ml with ethanol (80%, v/v). In accordance with Hendrix et al. (1993), we measured the soluble sugar and sucrose contents.

Hu et al. (2018) method was used to measure PEPC, V-H⁺-ATPase, and V-H⁺-PPase activities. Briefly, the fiber powder was extracted with 5 ml of the buffer containing 30 mM Hepes-Tris (pH 7.4), 250 mM mannitol, 3 mM ethylenediaminetetraacetic acid (EDTA), 1 mM phenylmethylsulfonyl fluoride (PMSF), 1.5% (w/v) polyvinylpyrrolidone (PVP) 4000, 1 mM dithiothreitol (DTT), and 0.1% (w/v) bovine serum albumin (BSA). In a centrifuge operated at 4°C, the homogenate was centrifuged (480×g, 10 min) and the supernatants were stored to measure the enzyme activities. The reaction solution (1100 µl) for PEPC activity assay contained 800 μl of the reaction buffer (30 mM Hepes-Tris (pH 7.5), 10 mM MgCl₂, 10 mM NaHCO₃, and 0.5 mM DTT), 100 μl of crude enzyme solution, 100 µl of malate dehydrogenase (EC 1.1.1.37, 100 U ml^-1, SIGMA), and 100 µl of 30 mM PEP. The PEPC activity was measured using the colorimetric method at 340 nm. For V-H⁺-ATPase activity determination, the reaction solution (500 µL) contained 400 μl of the reaction buffer (30 mM Hepes-Tris (pH 7.5), 3 mM MgSO₄, 50 mM KCl, 0.5 mM NaN₃, 0.125 mM (NH₄)₂ MoO₄, and 0.125 mM Na₃VO₄), 50 μl of crude enzyme solution, and 50 μl of 20 mM ATP-Tris. Half-hour was spent incubating the mixed solution at 30°C. In order to terminate the reaction, 1 ml of the stop solution (5% (NH₄)₂MoO₄: 5 M H₂SO₄: H₂O=1: 1: 3) was added. Finally, 200 μl of chromogenic agent (0.25 g of aminophenol sulfonic acid in 100 ml of 1.5% Na₂SO₃ (pH 5.5) mixed with 0.5 g of Na₂SO₃) was added and the mixture was incubated for 20 min at 37°C. The V-H⁺-PPase activity was also determined using the colorimetric method at 660 nm. Its reaction solution (500 µl) contained 400 μl of the reaction buffer (30 mM Hepes-Tris (pH 7.5), 3 mM MgSO₄, 50 mM KCl, 0.5 mM NaN₃, and 0.125 mM (NH₄)₂ MoO₄), 50 μl of crude enzyme solution, and 50 μl of 20 mM PP-Tris. The following measurement steps were the same as those employed to determine V-H⁺-ATPase activity. PM-H⁺-ATPase was isolated using Hu et al.’s method (2018). In brief, frozen fiber tissues were ground in an ice bath with 5 ml of cold buffer comprising 50 mM Hepes-Tris (pH 7.0), 300 mM sucrose, 8 mM EDTA, 2 mM PMSF, 1.5% (w/v) PVP 4000, 4 mM DTT, and 0.2% (w/v) BSA. In 4°C, the homogenate was centrifuged (10000×g, 20 min). In order to analyze enzyme activity in the supernatants, the supernatants were stored. The reaction solution (500 µL) for PEPC activity assay contained 400 μl of reaction buffer (30 mM Hepes-Tris (pH 6.5), 3 mM MgSO₄, 50 mM KCl, 0.5 mM NaN₃, 0.125 mM (NH₄)₂ MoO₄), 50 μl of crude enzyme solution, and 50 µl of 20 mM ATP-Tris. The subsequent measurement processes were in accordance with those of V-H⁺-ATPase activity determination.

Boiling cotton bolls formed before 30 DPA in 0.1% (v/v) HCl separated fibers from cotton seeds (Schubert et al., 1973). The fibers were stretched through the flowing water method (Thaker et al., 1989), and their lengths were measured using a vernier caliper.

For cotton bolls formed after 30 DPA, the fibers were peeled and incubated at 60°C (30 min) and then at 40°C (2 h). The fibers were finally incubated in a standard assay chamber for 48 h, with temperature and humidity set at 20 ± 2°C and 65 ± 5%, respectively. The fiber length was then determined using a photoelectric stapler (Y-146, Taicang Electron Apparatus Co., Ltd., China) (Yang et al., 2016).

The fibers were obtained by ginning seed cotton from cotton bolls harvested on September 15, 2019, and September 15, 2020. After drying the cotton fiber samples to a constant weight at 35°C, the fiber length was analyzed at the Supervision, Inspection, and Test Center of Cotton Quality, Ministry of Agriculture and Rural Affairs, China.

In order to calculate means, standard errors, and coefficients of variation (CV, %), Microsoft Excel 2007 (Microsoft Corp., Redmond, WA, USA) was used. The SPSS statistical software Version 23.0 (IBM Corp., New York, NY, USA) was used to conduct a variance analysis for all treatments at a 5% significance level, applying the least significant difference (LSD). In order to analyze each variable’s specific relevance, Pearson correlation coefficient was used.

Low soil AP levels (P₀ and P₁) decreased the fiber K⁺ contents ( Figure 1 ). Under P₁ and P₀, the average fiber K⁺ contents reduced by 7.3−8.8% and 10.3−12.3% in CCRI-79 and 10.6−22.7% and 20.8−33.1% in SCRC-28, respectively, at the 3 fruiting branch positions (FBPs) compared to P₂. Low soil AP had less effect at the FB_10–11 compared to FB_2–3 and FB_6–7 ( Figure 1 ). For all soil AP levels, CCRI-79 and SCRC-28 had average fiber K⁺ contents of 11.2-12.7 mg g^-1 and 9.5-12.9 mg g^-1, respectively. However, SCRC-28 had a higher CV (15.3%) than CCRI-79 (6.3%). CCRI-79 had fiber K⁺ contents 6.9 and 18.0% higher than SCRC-28 in P₁ and P₀, but 1.5% lower in P₂ ( Figure 1 ).

The fiber malate content increased at 5−10 DPA and decreased at 10−31 DPA ( Figure 2 ). Under the P₁ and P₀ treatments, the malate content decreased by 14.4–17.0% and 20.2–22.5% for CCRI-79, and 20.7–30.0% and 35.4–47.3% for SCRC-28, respectively, across the three FBPs in 2019 and 2020. SCRC-28 had a greater CV of 25.6% compared to CCRI-79’s CV of 6.3%. Additionally, In P₁ and P₀, CCRI-79 had malate contents that were 7.9% and 26.5% higher than SCRC-28, respectively, but in P₂, they were 3.3% lower. At the same soil AP level, fibers from FB_6–7 and FB_10–11 had higher malate content in contrast with FB_2–3 ( Figure 2 ). According to the above data, SCRS-28 (low-P sensitive) had greater reductions of K⁺ and malate contents in fibers than CCRI-79 (low-P tolerant) under low soil AP treatments (P₀ and P₁).

Nevertheless, the soluble sugar content of fibers from the three FBPs declined with the development of cotton bolls, especially at 10–31 DPA compared to 5–10 DPA ( Figure 3 ). Compared to P₂, the soluble sugar content reduced by 9.1−9.6% and 16.5−20.9% for CCRI-79, in P₁ and P₀ at the three FBPs, respectively, over the two years. A similar change pattern was observed for SCRC-28, with an 8.7−12.4% and 16.5−24.5% decrease under the same conditions ( Figure 3 ).

Interestingly, the fiber sucrose content had a similar trend with the soluble sugar content during the fiber elongation progress ( Figure 4 ). During the course of 24 months, under P₁ and P₀, the sucrose content at the three FBPs was reduced by 10.1−14.2% and 19.0−20.4% in CCRI-79 and by 10.1−13.2% and 19.3−22.2% in SCRC-28, respectively. Moreover, the fibers of FB_10–11 registered lower decline rates of the sucrose content than FB_2–3 and FB_6–7 for CCRI-79 but higher decline rates for SCRC-28 in 2019 and 2020 ( Figure 4 ).

Positive relations existed between the osmotically active solutes involved in fiber elongation, especially the K⁺ and soluble sugar contents, and P content of subtending leaves of both cultivars at 10−31 DPA (p<0.05) ( Figure 5 ). Significantly positive correlations were also observed between these contents from 10 to 24 DPA (p<0.05) (except for malate content of CCRI-79 at 24 DPA). The correlation between the P content of the leaves and the sucrose content of the fibers were all significantly positive (p<0.05) at 10 DPA for SCRC-28 and 31 DPA for CCRI-79 ( Figure 5 ).

The PM-H⁺-ATPase ( Figure 6 ), V-H⁺-ATPase ( Figure 7 ), and V-H⁺-PPase ( Figure 8 ) activities of fiber presented similar trends reaching its maximum at 17 DPA and declining with increasing soil AP content. Under P₁ and P₀, the PM-H⁺-ATPase activity increased by 8.4−10.5% and 11.3−14.4% for CCRI-79 and by 14.9−20.5% and 21.8−33.0% for SCRC-28, respectively at all the FBPs in the two years compared to P₂ ( Figure 6 ). As compared to SCRC-28, CCRI-79 exhibited 8.5 and 15.8% higher activity of fiber PM-H⁺-ATPase under P₁ and P₀, but 0.6% lower under P₂ ( Figure 6 ). The V-H⁺-ATPase activities reduced by 7.0−9.8% and 5.9−10.4% for CCRI-79, and by 12.4−18.6% and 23.4−33.8% for SCRC-28 under P₁ and P₀, respectively, in contrast with P₂ treatments throughout the FBPs and growing seasons ( Figure 7 ). Nevertheless, under P₁ and P₀, the V-H⁺-PPase activities decreased by 14.1−19.4% and 17.2−20.5% in CCRI-79, and by 18.3−23.7% and 28.5−38.4% in SCRC-28, respectively, in contrast with P₂ treatments throughout the FBPs and the two years ( Figure 8 ).

The fiber PEPC activity presented a single-peaked curve during fiber elongation and reached its climax at 17 DPA ( Figure 9 ). Compared to P₂, the PEPC activity for CCRI-79 reduced by 16.6−21.7% and 22.6−23.4% in P₁ and P₀ at the three FBPs in 2019−2020. Similarly, in P₁ and P₀, the PEPC activity reduced by 20.9−26.5% and 30.0−40.2% for SCRC-28. Low soil AP levels (P₁ and P₀) reduced the PEPC activities of fibers in the FB_10–11 compared to FB_2–3 and FB_6–7 in SCRC-28 but had less effect on the three FBPs of CCRI-79. Obviously, the variations in the four enzymes involved in fiber elongation of SCRC-28 were greater than those of CCRI-79 under low-P-stress.

Among the two cultivars, the enzymes involved in fiber elongation and P content were positively correlated, peculiarly PM-H⁺-ATPase and PEPC activities (p<0.01), during fiber elongation ( Figure 10 ). The positive correlations between P contents, and the V-H⁺-ATPase activities and V-H⁺-PPase were very significant (p<0.01) at 10−31 DPA in SCRC-28, and significant (p<0.05) at 10−17 DPA in CCRI-79 ( Figure 10 ).

The lack of AP in soil seriously hindered the growth and development of cotton plants ( Figure 11A ), thereby affecting fiber length ( Table 2 ). The fiber length changes formed an “s-shape” curve during the development of cotton bolls, which could be fitted using the logistic equation (Yang et al., 2016) as follows:

L means the fiber length (mm), L_max means the theoretical maximum fiber length, and a and b mean parameters.

The maximum velocity of fiber elongation (V_Lmax), the beginning time of rapid elongation (T_B), the finishing time of rapid elongation (T_F), and the rapid elongation duration (T_L=T_F-T_B) of cotton fibers were calculated using formulas (2), (3), and (4), respectively.

The rapid elongation stage of the fibers started from 10 to 24 DPA, after which the fiber length tended to stabilize ( Figure 12 ). Low soil AP levels (P₀ and P₁) decreased the V_Lmax, but increased the T_L, ultimately reducing the fiber length. The fiber length variation was consistent for CCRI-79 and SCRC-28 over the two years ( Table 2 ). V_Lmax had higher sensitivity to soil AP deficiency than T_L. Compared to P₂, the fiber length reduced by 2.2–3.6% and 3.9–6.3% in P₁ and P₀ for CCRI-79, mainly because V_Lmax declined by 3.8–5.9% and 5.9–9.8% in 2019 and 2020 across the FBPs. For SCRC-28, the fiber length reduced by 2.2–3.8% and 6.7–10.2% at P₁ and P₀ in contrast with P₂ due to the fact that V_Lmax reduced by 4.1–6.1% and 12.2–16.3% for both years at the three FBPs. In P₁ and P₀, CCRI-79 had a fiber length that was 2.3 percent and 5.2 percent longer than SCRC-28. However, CCRI-79 fiber length was 2.1% longer than SCRC-28 over the two-year under P₂ treatment. Moreover, there were longer fibers for CCRI-79 than those for SCRS-28 in soil AP deficiency.

The CVs of fiber length were lower in FB_10–11 than in FB_2–3 and FB_6–7 for both cultivars, and the values were even smaller for CCRI-79. Cultivar (C) (p<0.01) and AP (p<0.01) had the greatest impacts on fiber length ( Table 2 ).

We determined the cotton fiber lengths and their relationship with the contents of the osmotically active solutes and the activities of the enzymes engaged in fiber elongation. The results showed that fiber lengths had significant positive correlations (p<0.05) with the contents of all osmotically active solutes (K⁺, malate, soluble sugar, and sucrose) in SCRC-28 but correlated only with malate contents in CCRI-79 (p<0.01) ( Figure 13 ). Moreover, fiber lengths were significantly positively correlated (p<0.01) with the V-H⁺-ATPase and V-H⁺-PPase activities in CCRI-79 ( Figure 13 ). There was a remarkable positive correlation between fiber length and related enzyme activities in SCRC-28 as well (p<0.05) ( Figure 13 ).