Effect of irrigation and planting geometry on cotton ber quality and seed composition

Background: Cotton ber quality and seed composition play vital roles in the economics of cotton production systems and cotton seed meal industry. The objective of this research was to examine the effects of different levels of irrigation and planting geometries on ber quality and seed composition of cotton (Gossypium hirsutum L.). A two-year study was conducted in 2018 and 2019 on a Dundee silt loam soil in the Southeast USA with a warm and humid climate. Irrigation treatments were, irrigating every furrow (FI, full irrigation) and alternate furrow (HI, half irrigation), and no irrigation (RF, rainfed), and planting geometries were a single-row (SR) and twin-row (TR) on ridges spaced 102 cm apart. Fiber quality was tested by using the High-Volume Instrument (HVI) and Advanced Fiber Information Systems (AFIS). Seed protein, oil, and ber were estimated using near-infrared reectance spectroscopy. Results: The results showed irrigation and planting treatments played a signicant role in ber quality and seed composition. Across irrigation treatments, signicant differences were seen in ber properties, including micronaire, uniformity, upper half mean length (UHML), strength, yellowness, short ber, upper quartile length (UQL), neness, maturity ratio, and neps. The micronaire was negatively affected by irrigation as FI-SR, FI-TR, HI-SR, and HI-TR had recorded 11-12% over the RF-SR and TR treatments. The planting geometry played a minor role in determining ber quality traits like micronaire and nep count. Irrigation treatments recorded signicantly lower protein content by 3-4 % than rainfed, while oil content signicantly increased by 6-10 %. Conclusions: The results of the study indicate a potential for improving cotton ber and seed qualities by managing irrigation and planting geometries in cotton production systems in the Mississippi Delta region. by irrigation. All three seed composition traits studied here were not inuenced by PG (SR and TR). However, most often, protein and oil accumulation are genotype-dependent. These observations conrm with those of Pettigrew and Dowd (2012) and Bellaloui and Turley (2013). The results from this study demonstrate that (i) environmental conditions such as precipitation and solar radiation during boll development and cracking heavily impact ber quality and seed composition. (ii) Among irrigation and planting geometry, irrigation had a more signicant effect on ber quality and seed composition. (iii) Irrigation may not always result in better ber quality with desirable seed composition. (iv) It appears irrigation had an inverse relationship with protein accumulation while it had a limited positive effect on oil content. Irrigation did not affect seed ber content. Overall, both the ber quality and seed composition parameters are profoundly impacted by the irrigation than PG at critical stages of boll development and seed maturation.

In a recent study conducted in MS Delta, it was demonstrated that TR planting geometry enhanced cotton lint yield by 10.62% in 2018 and 17.62% in 2019 (Pinnamaneni et al., in press). Little is yet known of the impact of such yield enhancement on the lint quality and seed composition under different irrigation levels. When the individual effects of irrigation and planting geometry on cotton yield, ber quality, and seed composition were widely investigated, studies looking at their interactions were lacking. Also, research conducted on planting geometry (PG) or irrigation effects on ber quality tried to dwell on either HVI data or AFIS data, rather than combined analysis. The objectives of this study were to examine the effects of irrigation levels (Rainfed, RF; Full/all row irrigation, FI; half/alternate row irrigation, HI) and two planting geometries (single-row, SR; twin-row, TR) and their interactions on (i) cotton ber quality, (ii) cottonseed quality.

Cultural Practices
A two year  eld experiment with Cotton (cv. FiberMax1944GLB2) was conducted at the USDA-ARS Crop Production Systems Research Unit's research farm, Stoneville, MS, USA (33° 42′ N, 90° 55′ W, elevation: 32 m above mean sea level). The soil was a Dundee silt loam ( ne silty, mixed, active, thermic Typic Endoaqualfs) with 0.87% organic matter, 0.44% carbon, 0.06% nitrogen, 50 mg Kg − 1 P, 220 mg Kg − 1 K, 348 mg Kg − 1 Mg, 2057 mg Kg − 1 Ca, 2.1 mg Kg − 1 Zn, 9.1 mg Kg − 1 S, 16.6 CEC, and 1.28 g cm − 3 bulk density averaged across 60 cm soil depth. The eld saturated hydraulic conductivity (K fs ) of the soil, as measured in this study, ranged from 0.41 to 1.22 cm hr − 1 . Field preparation consisted of sub-soiling, disking, and bedding in the fall. The raisedridge seedbeds were re-furbished in the spring and before planting, tops of the seedbeds were smoothed as needed to plant cotton in SR and TR planting geometries. Glyphosate at 1.12 kg a.i. ha-1 was applied about one month before cotton planting to kill the existing weeds. A 7300 vacuum planter (John Deere, East Moline, IL) was utilized to plant in the SR planting geometry. TR geometry planting was done using a Monosem NG + 3 TR vacuum planter (ATI, Inc. Monosem, Lenexa, KS) that was set to achieve a plant population density of 120,000 plants ha − 1 . Actual plant populations were estimated at harvest by counting plants in 1 m 2 area in the two center rows at three randomly selected locations in each plot. Fertilizer application, weed control, and insect control programs were standard for cotton production. Isolated weed occurrences from time to time were hand hoed as needed.
Cotton cv. FiberMax1944GLB2 -a medium-maturing variety with broad adaptation, excellent yield potential, and outstanding ber quality -was planted on 8 May (2018) and 16 May (2019). The experiment was conducted in a split-plot arrangement of treatments in a randomized complete block design with six replications. The three irrigation regimes FI, HI, and RF are considered main plots while subplots consisted of two planting geometries, SR -rows evenly spaced at 102 cm and TR -two rows spaced at 25 cm apart on 102 cm centered seedbeds. Each plot consisted of four SR or 8 TR and 40 m long. Sensors for measuring soil-matrix water potential (Watermark sensors, Irrometer Company, Inc, Riverside, CA USA) were installed at soil depths of 15, 30, and 60 cm in selected representative plots. Irrigations were scheduled based on a soil matrix potential of about − 90 kPa at 45 cm soil depths, as recommended by Plumblee et al. (2019). The amount of irrigation water applied during each season in each plot was measured using a ow meter. In 2018, a total of 17.5 cm water was applied in the FI treatments in ve irrigation events of 3.5 cm each applied through every furrow on May 15, June 21, June 29, July 6, and August 4, while the HI treatments received the half the amount of water on the same dates applied in every other furrow (skip furrow irrigation). The total amount of water applied in HI was about 8.85 cm against 17.5 cm in the FI. In 2019, total irrigation applied was 15.2 cm in the FI treatment in four irrigation events of 3.8 cm each on May 26, June 29, July 24, and August 6, while in HI treatments, 7.5 cm of water was applied on the same dates. Irrigations were not applied after the rst boll cracking stage of the growth of cotton.
During mid-to-late September each year, cotton was defoliated using a two-step process. Defoliation was initiated when approximately 65% of the bolls had opened in mid-September. In the rst application, a mixture of 0.035 kg thidiazuron ha − 1 and 0.0175 kg diuron ha − 1 was applied on the crop canopy. One week later, a mixture of 0.035 kg thidiazuron ha − 1 , 0.0175 kg diuron ha − 1 , and 1.68 kg ethephon ha − 1 were applied as a second step to complete the defoliation and facilitate the opening of the remaining unopened bolls. Approximately two weeks after the second defoliant application, yield data was collected by handpicking from 1 m 2 section in the two center rows at three randomly selected locations in each plot.

Data Collection
Weather data was collected from the Mid-South Agricultural Weather Service, Delta Research and Extension Center, Stoneville, MS. The growing degree days (GDD) were calculated using a base temperature of 10 o C for cotton growth (Desclaux and Roumet, 1996). After physiological maturity, above-ground biomass was harvested from a 1 m − 2 section of middle two-rows from each plot at three locations, avoiding the row ends -one row sampled for the SR pattern and two rows sampled for the TR pattern. Seed cotton was ginned on a 10-saw laboratory gin (USDA-ARS Cotton Ginning Lab, Stoneville, MS), and the lint yield was calculated on a per hectare basis.

Fiber Quality Analysis
Ten subsamples were collected after the lint cleaner from each sample for ber quality analysis, ve of them for testing with AFIS, and ve for HVI. All lint samples for HVI were analyzed in the USDA ARS Cotton Ginning Research Unit (CGRU) in Stoneville, MS. At the same time, AFIS analysis was performed at Fiber and Biopolymer Research Institute, Texas Tech University, TX. Fiber quality parameters measured with both AFIS (nep, short ber content -SFC, upper quartile length -UQL, neness, maturity ratio, ber length by number, ber length by weight, trash, dust and HVI instruments (micronaire, ber length, uniformity index, strength, elongation, yellowness, re ectance, upper half mean length (UHML).

Seed Composition Analysis
Mature cotton seeds were collected, acid delinted, and analyzed for protein and oil. Brie y, approximately 25 g seed was ground using a Laboratory Mill 3600 (Perten, Spring eld, IL). Near-infrared re ectance, according to Wilcox and Shibles (2001) and Bellaloui and Turley (2013), using a diode array feed analyzer AD 7200 (Perten, Spring eld, IL) was employed to estimate protein and oil. Perten's Thermo Galactic Grams PLS IQ software was used for calibrations, and the calibration equation was established according to AOAC methods (Association of O cial Analytical Chemists, AOAC, 1990). Cottonseed protein and oil were expressed on a seed dry matter basis (Bellaloui and Turley, 2013).

Statistical Analyses
Statistical analyses were performed by analysis of variance (PROC MIXED;SAS Institute, 1996). Because all irrigation, planting date, treatments remained in their original location each year, years were treated as a repeated measurement when conducting a combined analysis across years. With the year, irrigation, planting geometry, and their interactions as xed effects and replication and whole plot (irrigation) as random effects. Random effects used in this model for the comparison across years were irrigation X year, planting geometry X year, and irrigation X planting geometry X year. Treatment means were separated at the 5% level of signi cance using Fisher's protected least signi cant difference (LSD) test.  (Table 1a) and seed composition s (Table 1b). The average lint yields in the irrigation and planting geometry combinations in this study were 1779 kg ha − 1 in FI -SR, 2029 kg ha − 1 in FI -TR, 1803 kg ha − 1 in HI-SR, 2082 in kg ha − 1 in HI-TR, and 1573 kg ha − 1 in RF-SR, and 1788 kg ha − 1 in RF-TR (Pinnamaneni et al., in press). The average nal plant-stand established in the FI and HI irrigated plots were 10.4 plants m − 2 each in FI and HI irrigated TR plots and 8.6 plants m − 2 in FI with SR and 8.2 plants m − 2 in HI with TR planting geometries. The higher lint yield in HI treatments is probably due to optimum water availability in the active root zone. However, in FI, wherein excess water around the root zone, owing to heavy precipitation events following the irrigation coinciding boll formation and developmental stages in July and August leading to higher vegetative growth, boll drop, and immature boll formation (Letey and Dinar, 1986;Feng et al., 2014) (Fig. 1b). RF with TR planting geometries had 9.34 plants m − 2, and there were 7.6 plants m − 2 in RF with SR at harvesting. The TR planting geometry produced a signi cantly higher number of bolls (75 bolls m − 2 ) than SR (64 bolls m − 2 ) (Pinnamaneni et al., in press). HVI Measurements

Micronaire
Micronaire represents the surface area of lint and is a measure of ber neness and maturity. Both irrigation and planting geometry has signi cantly affected micronaire (Table 1a). RF recorded a higher micronaire than the corresponding HI and FI treatments. As seen in Table 2a, among the two planting geometries, TR has consistently recorded a signi cantly higher micronaire by 3-7% on an average. Although most of the ber quality parameters were signi cantly different for the year, micronaire differences were consistent in 2018-2019. Means within each column followed by the same letter or letters are not statistically different by LSD means (P ≤ 0.05) Means in each column followed by the same letter or letters are not statistically different by LSD means (P ≤ 0.05)

Fiber Strength
Fiber strength, an important parameter affecting yarn quality, was signi cantly affected by irrigation treatments while the planting geometry did not have any in uence (Table 1a). There were no statistically signi cant differences in the ber strength between the two irrigation treatments (HI and FI) in both the years, but the RF signi cantly decreased the ber strength for both the seasons, as seen in Table 2a.

Uniformity
Irrigation has signi cantly affected uniformity in both the years, while PG does not affect ber uniformity (Table 1a). Among the irrigation treatments, HI has recorded signi cantly higher ber uniformity (HI-SR: 84.3 and HI-TR: 84.2) than FI treatments (FI-SR: 82.8 and FI-TR: 83.2) (Table 2a). This could be a result of excess water in the root zone during boll development and cracking resulting in more vegetative growth and immature bolls and, in some extreme cases, boll drop.

Upper Half Mean Length (UHML)
UHML is a crucial trait affecting the blend properties of yarn; hence the textile industry gives more importance. Irrigation has a signi cant positive impact on UHML (Table 1a). UHML was signi cantly lower in the rainfed treatments by about 9% on an average. However, like uniformity, UHML was not impacted by planting geometry, and the differences among HI and FI treatments were insigni cant (Table 2a).

Re ectance (RD)
This trait is unaffected by neither irrigation nor planting geometry treatments (Table 1a). However, year-wise differences were signi cant. The mean of all the treatments in 2018 was 5.9% higher than that of 2019, probably due differences in air temperature and precipitation pattern

Yellowness (+b)
Yellowness is unaffected by neither irrigation nor planting geometry treatments like re ectance (Table 1a). However, year-wise differences were signi cant. The mean of all the treatments in 2018 was 7.7% higher than that of 2019, probably due to differences in GDD and precipitation patterns (Table 2a).

AFIS measurements
Length by number (Ln) The length by number (Ln) was signi cantly affected by irrigation but not by planting geometry. However, the interaction between planting geometry and irrigation was insigni cant (Table 1a). The Ln was signi cantly higher in both HI and FI treatments over RF in both the years by 9% and 10% in 2018 and 2019, respectively. The average Ln for FI, HI, and RF were 23.5 mm, 23.4 mm, and 21.4 mm, respectively (Table 2b).

Nep Count
Neps are measures of defects in cotton ber. The measurement of nep count (both size and quantity) was commonly used to adjust in the processing machinery to reduce or eliminate the generation of mechanical neps. Nep count represents the number of neps observed in 0.5 g of the cotton ber sample. Irrigation signi cantly contributed to nep count while planting geometry, or its interaction with irrigation did not in uence nep count (Table 2a). On average, nep count in HI and FI was about 65% higher than that of RF cotton.

Short Fiber Content By Number (SFCn)
SFCn is expressed as a percentage of bers which are shorter than 12.7 mm. High SFCn reduces the quality of yarn. There was no signi cant effect of irrigation and planting geometry in both the years (Table 1a). However, irrigated treatments, both HI and FI, recorded numerically higher SFCn values than that of RF treatment in both the years. In a two year study at Lubbock, TX indicated irrigation increased the SFCn, in contrast to the ndings of the current study (Feng et al., 2014). This is probably due to the establishment of a better water balance between plant evapotranspiration demands and irrigation water applied.

Visible Foreign Matter (VFM)
Both the irrigation and year had a signi cant effect on VFM, while PG did not affect VFM (Table 1a). In 2018, VFM ranged between 2.5 to 3.15% while it ranged from 5.82 to 8.24% in 2019. The RF treatments have signi cantly lower VFM in both the years (Table 3a). This is probably due to diverse weather conditions during boll development, cracking, and harvesting.  Irrigation is detrimental to this measurement, as the higher neness of ber is preferable to the processing industry (Table 1a). However, the interaction of irrigation with planting geometry was insigni cant in both the years. As seen in Table 3b, rainfed cotton has higher ber neness (RF-SR: 183.5 and RF-TR: 186.1 millitex. However, the range for FI and HI treatments were between 170.6 to 175.4 millitex.

Maturity Ratio
Maturity ratio, a key component of ber quality, has a signi cant inverse relationship with irrigation (Feng et al., 2014). Planting geometry and its interaction with irrigation were insigni cant in both the years (Table 1a). The maturity ratio was highest in RF-SR and TR treatments (0.97), while it ranged between 0.94-0.95 in FI and HI treatments (Table 3b). It is believed that irrigation (continuously wet soil, with hardly any water de cits) propels the plant to vegetative growth, limiting the nutrients left in the soil for extraction during active ber growth bolls, resulting in lower cellulose deposition (Letey and Dinar, 1986).

Upper Quartile Length (UQL)
Upper Quartile Length (UQL) denotes the length of the longest 5% of all bers in the sample. The UQL is signi cantly affected by irrigation, while PG and irrigation interaction with PG were insigni cant (Table 1a). As seen in Table 3b, the range for UQL is narrow 1.28 to 1.31, and both HI and FI treatments had signi cantly higher UQL than that of RF cotton (Table 3b).

Seed Composition Parameters
Protein Irrigation had a signi cant negative effect on seed protein accumulation. The protein measured from the cotton seeds in the experiment were across the two years studied (Table 1b). However, the protein levels were not impacted by the PG. Both the RF treatments accumulated signi cantly higher protein accumulation (SR: 24.8 and TR: 23.55) than FI (SR: 23.1 and TR: 22.8) and HI (SR: 23.2 and TR: 22.8), respectively (Fig. 2a).

Oil
Seed oil content was signi cantly affected by irrigation in both the years (Table 1b), while PG had no impact on oil accumulation. The average oil content in rainfed treatments was 2.2% higher in 2018, while 2019 recorded over 5.4% than that of HI and FI treatments (Fig. 2b).

Fiber
The seed ber plays a vital role in meeting the dietary ber requirements of animals as most of the cotton after gossypol extraction is fed as cake to animals. In the current study, neither irrigation nor PG had any impact on seed ber. The seed ber values ranged from 20.2 to 23.1% (Fig. 2c).
The availability of soil water primarily determines the ability of individual cells within a plant to expand, and both root tips growing through the soil as well as bers elongating on seed coats in the bolls are no exception to it. Apart from limiting the plant growth, soil water stress triggers hormonal differences, particularly during reproductive growth resulting in senescence in fruiting bodies such as squares and bolls. Hence, e cient irrigation management involves reducing moisture stress at critical growth stages such that plants have the maximum capacity to initiate, retain, grow, and produce mature bolls, which results in twin objectives of higher ber quality and better seed composition. Pinnamaneni et al. (in press) reported responses of cotton lint and seed yield to PG and irrigations. They demonstrated that all-row irrigation (FI) would not result in any yield advantage over alternate row irrigation: TR planting geometry produced 10.6% in 2018 and 17.6% in 2019. The lack of irrigation response is probably due to the presence of excess water around the root zone, owing to heavy precipitation events following the irrigation. This situation probably resulted in nutrient leaching, lower water uptake resulting in the redistribution of energy allocation within the plant leading to higher vegetative growth at the expense of reproductive growth in humid climates (Letey and Dinar, 1986;Wanjura et al., 2002;Feng et al., 2014). This is understandable as cotton was initially perennial and abiotic stress can trigger more vegetative growth. This also coincided with boll formation and developmental stages in July and August (Fig. 1b). Another study conducted in the MS Delta demonstrated that the growth and development of individual cotton plants would be reduced to some degree under de cit irrigation, and ber and seed composition affected accordingly (Bellaloui et al., 2015).

Discussion
The ber quality parameters, both HVI and AFIS, were signi cantly impacted by irrigation rates (FI, HI, and RF), and only micronaire and nep count were in uenced by planting geometry (SR and TR). For some parameters like UHML, re ectance, yellowness, VFM, and maturity ratio, the year-wise response appears to be inconsistent owing to the variations in GDD, precipitation, and solar radiation, particularly during the months of July-September that coincides with boll development and maturation. In this study, micronaire was 11% higher in 2018 and 12% higher in 2019 than the average of HI and FI treatments. The acceptable level of micronaire was between 3.5 and 4.9%. The best level of micronaire range from 3.7 to 4.2% and the quality goes down when it is > 4.9 or < 3.5. It varied from 3.95 to 4.91 in this study. Similar results were reported by Dagdelen et al. (2009), Feng et al. (2014, and Zhang et al. (2016). TR geometry recorded up to 6% higher micronaire while nep count was lower in TR planting geometry, with some degree of inconsistency. These observations are like the ndings of Reddy et al. (2009), Stephenson et al. (2011), and Feng et al. (2014. Stephenson et al. (2011) showed that the micronaire, strength, and re ectance were higher, and nep count and seed coat nep count were lower for 38-cm twin rows than for 102-cm solid rows based on a two-year study conducted in Louisiana. During ber developmental stages in the cotton plant, many factors, including weather conditions, nutrient and water stresses, defoliant application time, and cultivar, can impact micronaire (Hake et al., 1990). Overall, planting geometry differences were not found for length, uniformity, yellowness, trash by HVI or for upper quartile length, short ber content, neness, immature ber content, or maturity ratio by AFIS, with some inconsistency between years. Fiber length and uniformity were higher in HI and FI over RF, while neness and maturity ratio were higher in RF treatment, which agrees with Feng et al. (2014), Dagdelen et al. 2009, andPettigrew (2004). Irrigated cotton had higher SFC than the RF treatments, contrary to the observations of Sui et al. (2017), who used a cotton picker for sampling vis a vis hand picking in the current study. This can be explained by irrigation and precipitation during boll development maturation drives the plant towards the vegetative phase, resulting in a lower deposition of cellulose in the developing bers. This is further vindicated by the inverse relationship of irrigation with the maturity ratio in this study. Similar results were reported earlier (Wanjura et al., 2002;Dağdelen et al., 2009;Feng et al., 2014). The reduced ber strength in this study under rainfed condition con rms earlier observations of the involvement of carbohydrate and energy metabolisms in ber development and carbon skeletons for the synthesis of cell wall polysaccharides and fatty acids (Yang et al., 2008). Moisture stress impacts negatively the formation of the actin cytoskeleton that triggers the secondary cell wall synthesis, a key component in determining the ber strength (Wang et al., 2010).
The cottonseed and its products have a high demand from the dairy and food-related industries. It has previously been documented to be affected by variety, planting date, and irrigation (Pettigrew and Dowd, 2012). In this study, protein accumulation was negatively affected by irrigation, while it had a signi cant positive impact on seed oil content. Seed ber is not impacted by irrigation. All three seed composition traits studied here were not in uenced by PG (SR and TR). However, most often, protein and oil accumulation are genotype-dependent. These observations con rm with those of Pettigrew and Dowd (2012) and Bellaloui and Turley (2013).

Conclusion
The results from this study demonstrate that (i) environmental conditions such as precipitation and solar radiation during boll development and cracking heavily impact ber quality and seed composition. (ii) Among irrigation and planting geometry, irrigation had a more signi cant effect on ber quality and seed composition. (iii) Irrigation may not always result in better ber quality with desirable seed composition. (iv) It appears irrigation had an inverse relationship with protein accumulation while it had a limited positive effect on oil content. Irrigation did not affect seed ber content. Overall, both the ber quality and seed composition parameters are profoundly impacted by the irrigation than PG at critical stages of boll development and seed maturation.