ASABE Logo

Article Request Page ASABE Journal Article

Irrigation Response, Water Use, and  Lint Yield of Upland Cotton Cultivars

Robert C. Schwartz1,*, Travis W. Witt2, Mauricio Ulloa3,  Paul D. Colaizzi1, R. Louis Baumhardt1

Published in Journal of the ASABE 67(2): 421-437 (doi: 10.13031/ja.15868). 2024 American Society of Agricultural and Biological Engineers.

1    Conservation and Production Research Laboratory, USDA ARS, Bushland, Texas, USA.

2    Grazinglands Research Laboratory, USDA ARS, El Reno, Oklahoma, USA.

3    Cropping Systems Research Laboratory, USDA ARS, Lubbock, Texas, USA.

*    Correspondence: robert.schwartz2@usda.gov

The authors have paid for open access for this article. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License https://creative commons.org/licenses/by-nc-nd/4.0/

Submitted for review on 23 October 2023 as manuscript number NRES 15868; approved for publication as a Research Article by Associate Editor Dr. Gary Marek and Community Editor Dr. Kati Migliaccio of the Natural Resources & Environmental Systems Community of ASABE on 9 January 2024.

Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture.

Highlights

Abstract. Upland cotton (Gossypium hirsutum L.) production requires less irrigation compared with other crops and thus provides an opportunity to reduce risk and maintain profitability in areas where water is limited. Water use, canopy temperature, lint yield, and crop water productivity were evaluated for four early to medium maturity upland cotton cultivars under three levels (100%, 66%, and 33%) of alternate furrow subsurface drip irrigation (SDI) in a thermally limited environment. Crop evapotranspiration (ET) across cultivars and years averaged 627, 547, and 467 mm for the 100%, 66%, and 33% irrigation levels, respectively, and did not differ among cultivars (P > 0.05). Changes in stored soil water within each irrigation level were similar among cultivars, with significant differences occurring infrequently. Measured canopy temperatures from first white flower to two weeks after cutout did not significantly differ among cultivars (P > 0.10) within each irrigation level. Crop water use during boll maturation, as inferred from the developed crop coefficient curve, was considerably less than reported by other studies, signifying that irrigation could be terminated earlier without reducing lint yield. The cultivar effect on lint yield was significant in all study years (P < 0.001), but only at the 66% and 100% irrigation levels, with one cultivar exceeding the average yields of all evaluated cultivars by 13% across the three study years. Medium maturity cultivars usually yielded less than early maturity cultivars, especially for a year with less accumulation of thermal energy. Crop selection and late season irrigation water management were both key to improving cotton water productivity.

Keywords.Keywords. Upland cotton, Subsurface drip irrigation, Water productivity, Root water uptake, Canopy temperature.

The High Plains aquifer is the principal source of water for irrigation in the U.S. Great Plains. In 2005, approximately 97% of the 24 billion m3 of water pumped from the aquifer was used for irrigation (McGuire, 2009). In many areas of western Kansas and the Texas High Plains (THP), withdrawals have greatly surpassed recharge rates that have reduced the saturated thickness to in excess of 46 m (150 ft.) since predevelopment (McGuire, 2017). The associated reductions in well yields in these areas have increased production risks for growers, especially for crops sensitive to water deficits such as maize (Zea mays L.).

Texas accounts for 50% of the total U.S. planted acreage of cotton with the majority being produced in the southern THP, with acreage increasing in the northern THP in recent years (USDA-NASS, 2022). Cotton production in this region can be profitable yet reduce both water consumption and peak water requirements compared with maize (Howell et al., 2004; Bordovsky et al., 2015). Thus, irrigated cotton production provides an opportunity to reduce risk and maintain profitability in areas with diminished well capacities (Baumhardt et al., 2013; Bordovsky et al., 2015; Bordovsky, 2020; Witt et al., 2020). Subsurface drip irrigation (SDI) for cotton production is gaining popularity in this region, especially in areas with limited well yields in southern Texas High Plains, with irrigation capacities under center pivots typically less than 4 mm d-1 (USDA-RMA, 2020). Subsurface drip irrigation can reduce soil water evaporation, which is significant in cotton production because canopy closure does not occur until late in the growing season or not at all. Additionally, greater lint yields have been obtained under SDI compared to low pressure spray applications (Bordovsky, 2019). In this region, drip laterals are typically installed in every other interrow under an “alternate furrow” configuration. This reduces the initial cost of installation and can also maximize the irrigated area under conditions of limited irrigation capacities (USDA-RMA, 2020). However, stand establishment under alternate furrow SDI can be problematic in years with limited precipitation in April and May (Bordovsky and Mustian, 2020).

Cultivar selection is an important decision affecting yield potential in the Texas High Plains. Results of cultivar trials for irrigated production (Bell et al., 2021) demonstrated that, across seven locations in the THP, the highest yielding cultivar produced 18% greater lint than the test average, with a few cultivars producing above average yield at most locations. Yield differences may have resulted from improved traits associated with partitioning dry matter to lint yield under water stress (Gerik et al., 1996), root proliferation and soil water extraction under deficit irrigation, and maturity classes better adapted to years and locations with suboptimal accumulation of thermal energy. Crop water use was not evaluated in these trials, so it is not clear if improved root water uptake of some cultivars was key in augmenting yield. Because of the difficulty in accurately monitoring soil water and determining crop water use, studies comparing water use among cultivars are scarce. The only such study in the region (Witt et al., 2020) found that, in one of the study years, upland cultivars extracted significantly more water from the soil profile compared with Pima (G. barbadense) lines, resulting in 30 mm greater water use from first square to cutout. Besides rooting traits facilitating greater water uptake, some upland cultivars may be able to maintain a greater transpiration rate and hence photosynthetic capacity under late season water stress. For example, leaf shape and the degree of pubescence may alter the leaf energy balance and hence stomatal aperture, transpiration, leaf temperature, and photosynthetic efficiency (Bednarz and van Iersel, 2001; Mahan et al., 2016; Pettigrew, 2016).

Crop maturity in cotton is evaluated largely by the attainment of ‘cutout’ characterized by the cessation of effective inflorescence, whereby the initiation of new fruit is limited by assimilate production (Bourland et al., 1992; Bange and Milroy, 2000; Gwathmey et al., 2016). Water availability, besides limiting transpiration and assimilate production, will also influence the timing of cutout and maturity (Gwathmey et al., 2011). Although irrigation at deficit levels can reduce lint yield, it can also hasten the attainment of maturity (Pettigrew, 2004; Gwathmey et al., 2016), which may partly explain the greater lint yields under moderate deficit irrigation during growing seasons with limited heat unit accumulations (Howell et al., 2004). Consequently, in thermally limited zones of production, water management requires matching effective precipitation and irrigation to the available heat units.

Weather-based methods can be used to schedule irrigation to meet the water requirements of cotton using reference evapotranspiration and the associated crop coefficients (Allen et al., 1998). For an indeterminate crop such as cotton, the determination of crop growth stages and the respective FAO-56-based crop coefficients (Allen et al., 1998) are problematic because of the influence of water availability on maturity (Farahani et al., 2008). In a 3-year study evaluating cotton water use under surface drip irrigation, Farahani et al. (2008) found that the developed crop coefficient curves varied in each year of the study, especially for late season senescence. Variations in year-to-year water availability later in the growing season after cutout were partly responsible for the inconsistencies in the fitted crop coefficient during this growth stage. The reported mid-season stage FAO-56 crop coefficient for upland cotton ranges from 1.05 to 1.31 (Howell et al., 2004; Farahani et al., 2008; Ko et al., 2009; Hunsaker and Bronson, 2021); however, the timing and duration of this stage varied greatly among these studies. Additionally, the irrigation application method (sprinkler, surface drip, and SDI) likely influenced the magnitude of the estimated crop coefficients. Subsurface drip irrigation applied either equidistant between crop rows in alternate furrows (Bordovsky, 2020) or directly under crop rows (Bordovsky, 2020; Hunsaker and Bronson, 2021) presents unique challenges when relating water depletion to crop water stress because of the spatial variability in water contents caused by irrigation, the location at which soil water contents are measured in reference to the line source, and lateral root proliferation during the growing season (Bufon et al., 2012).

Canopy temperature measurements in conjunction with a time-temperature threshold irrigation schedule for cotton (Wanjura et al., 1992; 2004) avoid some of the difficulties associated with crop coefficients and the spatial variability of water contents generated by drip irrigation. This technology is increasingly being used in the THP in conjunction with cotton irrigation management. Because of the differences in leaf shapes and pubescence, as well as other phenological traits, the canopy temperature response may not be similar among cultivars. Such evaluations have not been made at the field scale to ascertain if the canopy temperatures of certain cultivars respond differently to irrigation.

The objectives of this study were to (1) evaluate differences in water use, lint yield, and crop water productivity among early to medium maturity class upland cotton cultivars with diverse leaf type characteristics across three irrigation levels; (2) determine differences in canopy temperatures for selected hybrids within two irrigation levels; and (3) develop crop coefficients and corresponding water stress adjustments, timing, and duration of crop stages, for a thermally limited environment under alternate furrow SDI. Yield components, boll distributions, and lint quality will be evaluated in a forthcoming manuscript.

Materials and Methods

Field Experiments

The study was initiated in 2014 at the Conservation and Production Research Laboratory in Bushland, TX (35°11'25? N, 102°5'18? W, 3840 m asl). The mean annual precipitation at this location is 475 mm, with 67% received during the growing season (May–September). The experimental field was a Pullman clay loam (fine, mixed, superactive, thermic Torrertic Paleustoll) with < 1% slope and a noncalcareous Ap horizon (0.0–0.15 m). The silty clay to clay Bt horizon at this field site extends to a depth of 1.4 m ± 0.1 m and is underlain by a clay loam Btk horizon with up to 50% calcium carbonates (Unger and Pringle, 1981). The water retention properties of the Pullman clay loam are reported by Schwartz et al. (2020).

Field plots were irrigated using an SDI system consisting of eight zones, each 242 m long × 9.1 m (12 rows) wide. Driplines (Typhoon 990, Netafim Ltd., Tel Aviv, Israel) were spaced 1.52 m apart at a depth of 0.28 – 0.30 m and centered between alternate 0.76-m crop rows with a 0.31 m (12-inch) spacing between emitters (flow rate of 0.91 l h-1 (0.24 gal h-1) at 68.9 kPa (10 psi). Each zone was furnished with a pressure reducing valve and solenoid (Netafim Ltd., Tel Aviv, Israel) to regulate downstream pressure to 68.9 kPa and a flow meter to quantify the volume of each irrigation application. Experimental plots (24.4 m long × 4.6 m (6 rows) wide) were arranged within a split plot design with four irrigation levels as main plots and six cotton cultivars in each year randomized within subplots and four replications per treatment (96 experimental plots). Each irrigation zone consisted of two main plots, each 73 m long × 9.1 m wide (3 × 2 replicate plots), so that blocking was controlled for the distance of the dripline from the control valve as well as the zone. Irrigation levels included full irrigation (100%), 66%, and 33% levels, as well as dryland, in which irrigation was applied only to ensure crop establishment. In this study, soil water balance was evaluated only for four cotton cultivars in each study year (table 1), three irrigation levels, and three replicates (blocks 1, 2, and 3) because it was impractical to monitor water contents on all plots using the neutron moisture gauge. Water contents were not monitored for the dryland plots, and hence this treatment was omitted in the analysis of the results. Upland cotton cultivars were sown in 2014–2016 and 2018–2020 in a north-south row orientation with a 6-row vacuum planter (Max-Emerge, John Deere) at a seeding rate of 165,000 seeds ha-1 and a row spacing of 0.76 m, with the goal of achieving a stand density of 11–13 plants m-2. Heavy rainfall and soil crusting hampered plant emergence in 2014 and 2015, resulting in inadequate stand densities. Hail also destroyed recently emerged plants in 2015 and 2019. Consequently, a successful stand was established only in the 2016, 2018, and 2020 cropping seasons. Final plant densities at harvest during these years were 11.7, 11.3, and 12.3 plants m-2, respectively. In 2014 and 2015, the cotton was terminated, and the field was fallowed during the summer. Spring oats (Avena sativa, L.) were seeded in 2017 and terminated with glyphosate [N-(phosphonomethyl) glycine] prior to inflorescence. In 2019, soybean (Glycine max, L.) was planted and harvested in September. In each growing season, one of the cotton cultivars was seeded uniformly in all zones at the southern end of the SDI field (96 m of drip line).

Table 1. Commercial cultivars evaluated for water use in the study and associated maturity ratings and phenological characteristics.[a]
CultivarYearsMaturityStatureLeaf Type
Deltapine DP1212 B2RF2016EarlyMedium-shortLight hairy
FiberMax FM2011 GT2016-2020EarlyShortSemi-smooth
FiberMax FM2484 B2F2016-2018MediumMediumSmooth
PhytoGen PHY333 WRF2016-2020EarlyMedium-tallHairy
Deltapine DP1612 B2XF2018-2020EarlyMedium-shortLight hairy
FiberMax FM2334 B2F2020MediumMediumSmooth

    [a]    In 2018 and 2020, the seeds of the initial selected cultivars were unavailable and replaced with similar varieties.

Growing season weeds were controlled using a pre-plant application of paraquat dichloride (1.2 kg a.i. ha-1 only in 2016) and a pre-emergent application of S-metolachlor [2-chloro-N-(2-ethyl-6-methylphenyl)-N-(2-methoxy-1-methylethyl) acetamide] at 1.4 kg a.i. ha-1. Weeds were also controlled during the growing season with glyphosate, a disc cultivator, and by hand hoeing. In the early spring, plots were typically disc-tilled twice for weed control and for seedbed preparation. Glyphosate (0.56 kg a.i. ha–1) was also used for weed control in the spring.

Extractable nutrients were determined in soil samples collected in the spring prior to planting, and lint yield goals were used to determine nitrogen and phosphorus fertilizer application rates. In 2016 and 2018, 38 kg ha-1 P and 30 kg ha-1 N as liquid fertilizer was injected with a knife applicator prior to planting. Phosphorus was not required in 2020. The remaining nitrogen was applied as UAN by fertigation through the dripline between first square and first white flower. Total N applications averaged 82, 102, and 114 kg ha-1 for the 33%, 66%, and 100% irrigation levels, respectively, across the three study years. Applications of N at greater than optimal rates at given irrigation levels were avoided to prevent vegetative growth and diminished availability of assimilates required for fruiting structure development that would reduce yield potential (Main et al., 2013; Pabuayon et al., 2021).

Irrigation applied through the driplines was necessary for seed germination and plant establishment (e.g., Camp, 1998) in all growing seasons because of inadequate precipitation before and after planting. Two to four days of continuous alternate furrow SDI were required in this soil to permit adequate wetting of the seed zone. In 2016, 27 mm of rainfall was received four days after planting that facilitated germination; however, this was insufficient for plant establishment, and an additional 39 and 52 mm of irrigation were applied on day of year (DOY) 166 and 176, respectively, to supply moisture to the rooting zone. In 2018, 119 mm of irrigation was applied two days after planting to all zones, including the dryland treatments, to achieve germination. Also, 28 mm of irrigation was applied to all zones on DOY 173 in 2018. In 2020, 91 mm irrigation was applied to all zones two days after planting, and an additional 49 mm of irrigation was applied to all zones on DOY 153.

Imposed irrigation treatments were initiated between DOY 172 and DOY 183 at approximately seventh node (fig. 1) and were based on the fraction of mean plant available water content within the profile (0–1.4 m) measured using neutron scattering for the FM2011 cultivar at the 100% level. From seventh node to cutout, irrigations were timed to gradually reduce the fraction of plant available water from 0.8 to 0.6 for the 100% irrigation level. From cutout to first open boll, irrigations were timed to maintain the fraction of plant available water above 0.4 for the 100% irrigation level. The final irrigation for the 100% level was applied on 16 September in 2018 and 2020, within a week of first open boll with heat unit accumulations of 1150 and 1180°C since planting. Irrigations were terminated earlier in the season in 2016 because of considerable rainfall received in late August and early September (fig. 1). An irrigation level of 66% was achieved by skipping the first one of every three irrigations at the 100% level. Likewise, an irrigation level of 33% was implemented by skipping the first two of every three irrigations at the 100% level.

The applied depth of water was approximately 25 mm for each scheduled irrigation after first square in 2016. In 2018 and 2020, 25 mm irrigations were applied from first square until flowering. However, subsequent irrigations after first flower in these two years were applied in increments of approximately 15 mm to avoid deviating from the planned depletion to 0.6 fraction of plant available water at cutout in the event of significant precipitation.

Figure 1. Cumulative reference evapotranspiration (ETo), precipitation, gross irrigation during, and growing degree days (GDD) for the three study years. Day of planting (P) and developmental growth stages for seventh node (N7), first square (S1), first white flower (WF), cutout (CO), first open boll (OB), and harvest (H) are indicated for each growing season. Note that ETo and GDD correspond to the right axis.

Instrumentation

Soil water contents were monitored using a neutron moisture gauge (model 503DR, InstroTek, Inc.) from 0.1 to 2.3 m in 0.2-m increments near the center of the experimental plots at 7-d intervals, except in late September and October, when it was monitored biweekly. A depth control stand was used to ensure that the source of the gauge was centered at the intended measurement depths (Evett et al., 2003). The neutron probe used in this study was previously calibrated in 2002 using methods described by Evett and Steiner (1995) with separate equations for the A, Bt, and Btk horizons.

A micrometeorological station was placed in a field planted to a small grain crop and 400 m west of the SDI study plots. In 2020, the weather station was moved to the eastern edge of the study plots. Relative humidity and ambient air temperature were sampled at 2 m with a HC2S3 probe (Rotronic Instrument Corp.) housed within a radiation shield. Solar irradiance was measured at 1 m above the surface using a model CM14 albedometer (Kipp and Zonen, Delft, The Netherlands) in 2016 and 2018 and using a model SPP pyranometer (Eppley Laboratory, Inc., Newport, RI) in 2020. Wind speed was measured at 2 m above the surface using a model 014A cup anemometer (MET-ONE Instruments) and a WindSonic Ultrasonic Wind Sensor (Model 1405-PK-038, Gill Instruments). Precipitation was measured at 0.6 m above the surface with a tipping bucket rain gauge (TE525M, Texas Electronics). Mean sampled data (15 s sampling frequency and 0.25 and 1 h averages) were recorded with a CR3000 micrologger (Campbell Scientific). Reference evapotranspiration (ETo) was calculated using the ASCE standardized reference evapotranspiration equation for a short reference crop at a 1-h time step (Allen et al., 2005) using the wind speed measured with the cup anemometer. Growing degree days were evaluated using a 15.6°C base temperature and no upper temperature threshold using Method 1 of McMaster and Wilhelm (1997).

Type-T infrared thermometers (Exergen IRT/ C.5) with a 5:1 field of view and a spectral response of 6.5 to 14 µm were deployed each growing season to measure the canopy temperatures of two cultivars after first square. The IRT sensors were enclosed in a white polyvinyl chloride housing to protect the sensor against the elements. Portable thermocouple data loggers with a USB interface (USB-501-TC, Measurement Computing) were used to measure the voltages. All thermocouple - data logger pairs were calibrated with a black body radiation source in a temperature-controlled chamber over the range of 15 to 40°C in 5°C increments and at body temperatures of 20 and 30°C. A single calibration was used for all the IRT’s (RMSE = 0.52°C) because it was found that individually calibrated probes resulted in a doubling of the standard deviation in temperature (n = 24) with mean temperature differences of less than 0.1 °C when focused on a common radiometric target in the field. IRT’s were mounted on vertical poles and centered directly above the cotton rows. The IRT’s were oriented in a north or south direction, 1.2 m north and south of neutron probe access tubes, and at a 20-degree zenith angle at a height of 0.17 m above the canopy so that the field of view was comprised entirely of the canopy, thereby reducing as much as possible the background radiation interference from soil. The IRT’s were raised weekly to ensure the height above the canopy was maintained at approximately 0.17 m. The sensor optics were cleaned weekly with forced air and occasionally with a cotton swab and isopropyl alcohol. In 2016, canopy temperatures were acquired every five minutes, and in 2018 and 2020, canopy temperatures were acquired every minute. A total of 24 IRT’s were deployed each year, comprising six sensors for each of two selected cultivars (three replications, each with a north- and south-facing IRT) in the 100% and 33% irrigation levels, respectively. In 2016 and 2018, the selected cultivars were FM2011 and PHY333, and in 2020, the selected cultivars were DP1612 and FM2334. Data quality was checked manually each year by ensuring that canopy temperatures were within reasonable bounds and that predawn temperatures of all IRT’s converged to within approximately 1.5°C of each other. Data streams of sensors that had outliers were removed from the calculation of the 0.25-h means.

Crop Evapotranspiration and Coefficient Curves

Crop evapotranspiration (ETa) was calculated by soil water balance evaluations obtained by summing the change in soil water content within the root zone measured by the neutron probe, accumulated precipitation, gross irrigation, and estimated runoff. For this SDI system, application efficiency was assumed to be 100%. Runoff of precipitation, which by visual observation occurred once in 2016 and once in 2018, was estimated at 100% irrigation level to modify the overestimated crop coefficient, Kc, for that time interval (e.g., in 2016, unadjusted Kc = 1.45) to a value closer to an appropriate value corresponding to FAO-56 (Allen et al., 1998). Based on profile water contents below the maximum rooting depth, drainage was assumed to be negligible in all years. Initial profile water contents (0–1.4 m) at planting were calculated by assuming Kc = 0.35 prior to the first neutron probe measurements. A high value for this coefficient is reasonable since much of the irrigation applied that evaporated at the surface was unavailable to the cotton rooting system during this growth stage.

Crop water use during the growing season was estimated using an FAO-56–based model (Allen et al., 1998):

        (1)

where

ETc = calculated crop water use (mm)

Ks = crop stress coefficient under nonstandard conditions

Kc = crop coefficient for each of the three growth stages (Kc-ini, Kc-mid, and Kc-end) based on crop phenology

ETo = Short grass reference ET (mm).

The initial stage of Kc-ini extended from planting to the appearance of the 7th node. The mid-stage Kc-mid, which represents peak water use, commenced at a fitted offset in days after the appearance of first white flower and concluded at a fitted number of days after cutout of the early maturing cultivars under the 100% irrigation level. The end stage of Kc-end terminates upon harvest, a killing freeze, or defoliation, whichever occurs first. The stress response function p was evaluated as follows:

        (2)

where

p(ETm) = stress response depletion fraction below which water stress occurs for a given ETm

pb = baseline stress response depletion fraction below which water stress occurs when ETm = ETb

ETb = baseline ET for the stress response set equal to 5 mm d-1

ETm = Kc·ETo

d = multiplier that accounts for changes in stress response for differing ETm(d mm-1).

The baseline stress response pb was assumed to be constant throughout the growing season. A nonlinear stress coefficient Ks is calculated using an adaptation of the FAO-56 relationship (Schwartz et al., 2020) and is expressed as follows:

    (3)

where

SPA/Sfc = fraction of stored plant available water (SPA, mm) relative to stored plant available water at field capacity(Sfc, mm) within the rooting zone with depth (zr)

e = coefficient that accounts for the nonlinearity of the water stress relationship

zr = depth of the rooting zone after emergence (m).

The rooting depth was assumed to increase linearly with days after emergence and attain its maximum extent at first flower. Water retention at the permanent wilting point (1.5 MPa) and field capacity (0.033 MPa) for each depth increment of neutron probe measurements are given by Schwartz et al. (2020).

Crop coefficients and coefficients of the stress response function were fitted so as to minimize the sum of squared differences between measured (ETa) and predicted (ETc) crop water use simultaneously for all years and irrigation levels using the gradient reduction method in the Excel solver.

        (4)

where

F(ß, ?) = objective function of the minimization problem

ß = parameter vector of crop and stress response coefficients (eqs. 1 – 3)

? = vector of independent variables (ETo, precipitation, irrigation, etc.)

s2 = variance of the measured crop ET (ETa) for k = 1 to m intervals between measured changes in stored soil water contents

nI = the number of irrigation levels

nY = the number of optimization years.

Essentially, the parameter vector ß comprising the crop and stress response coefficients in equations 1–3 were optimized by minimizing the sum of squared differences of predicted and measured crop ET throughout the growing season for all years and irrigation levels. For these optimizations, the maximum rooting depth was approximated by increasing the depth for stored soil water calculations in increments of 0.2 m until no further increase in the change in stored soil water was obtained from planting to the final neutron probe measurement. The beginning and ending day of the growth stage for peak water use were evaluated as the number of days offset from first white flower and cutout, respectively. These offsets were manually adjusted to minimize the objective function between crop coefficient parameter fits for three iterations to ensure that the mid-season stage was correctly identified. After identifying these offsets, a final Excel solver solution was obtained for the final parameter estimates. Asymmetric confidence intervals of fitted parameters were approximated using Fisher’s F distribution and a graphical procedure described by Kemmer and Keller (2010).

Cotton Development and Lint Yield

During the growing season, evaluations of the number of nodes and phenological growth stage were completed approximately biweekly from emergence until first open boll for all irrigation levels and cultivars. More frequent phenological observations of the FM2011 cultivar in plots under the 33% and 100% irrigation levels were used to target the timing of the evaluations of all cultivars so that they would occur around the time of first square, first white flower, cutout, and first open boll. Cotton plants were considered in cutout when they averaged five nodes above a first position white flower, with the final node at the top having a leaf diameter of at least 2.5 cm. Seed cotton was picked by hand in 2 m × 2 inner rows of the six-row plots within each replicate plot, with one of the rows centered on the neutron probe access tube for replicates where water contents were monitored. Measurements included plant density and the number of open and immature (unopened) bolls. Harvest aids (boll openers and defoliants) were not used in the study because of the difficulty in timing the application in a field with early and medium maturing cultivars as well as a range of irrigation levels. A 500 g subsample of seed lint cotton was ginned on a 20-saw gin (Compass Systems Model TT 520, Carmel, IN) to determine percent turnout, and fiber yield was calculated by multiplying percent turnout by harvested seed lint yield.

Statistics

Analysis of variance for the main effects of irrigation level and cultivar was evaluated using the GLIMMIX procedure in SAS 9.4 (SAS Institute, 2014) for a split-plot design. Irrigation was on whole-plots, and cultivars were randomized within whole plots. Irrigated whole plots (blocks) were modeled as fixed effects, and blocks × irrigation level was modeled as random effects. Mean separation and determination of least significant differences were evaluated using the Tukey adjustment for comparisons with more than two means. Effects were declared significant at the a = 0.05 probability level, and each year was evaluated separately. The pooled two-sampled t-test, assuming equal variances, was used to assess significant differences between the mean 0.25 h canopy temperatures of the two cultivars at each irrigation level. Effects were declared significant at a higher probability level (a = 0.10) to reduce the type-2 error rate with smaller sample sizes for some of these comparisons.

Results

Growing Season Weather Conditions

In all three study years, seasonal precipitation was less than the long-term mean (264 ± 96 mm s.d. for 1 May–28 Aug). In 2016, 256 mm of precipitation was received between planting and harvest; however, nearly half (113 mm) fell during mid to late August. Although 326 mm was received between planting and harvest in 2018, only 150 mm could be considered effective in increasing cotton yield because the remainder was received from early to mid-October. In contrast to 2016 and 2018, the 183 mm received in 2020 was more uniformly distributed throughout the growing season (fig. 1).

The timing of phenological development of cotton in all years was similar from planting to first flower among all cultivars (fig. 1). However, first open boll, which occurred approximately 118 days after planting in all years, tended to be delayed 2–4 days under the 100% irrigation level compared with the 33% irrigation level. In 2016, the growing degree days required to reach first open boll (1082°C) was less compared with 2018 and 2020 (1150 and 1159°C, respectively). During 2016, the accumulation of growing degree days slowed after cutout (fig. 1) because of the below normal temperatures associated with the precipitation during this period. Cooler weather in addition to elevated water availability during this period in 2016 likely caused excessive regrowth and delayed boll maturation. Cotton was hand harvested on DOY 319, 304, and 296 in 2016, 2018, and 2020, respectively, that occurred prior to killing freezes in each respective year.

Seasonal irrigation amounts and water use varied considerably with year (fig. 1 and table 2), with greater water requirements in 2018 because of reduced precipitation and in 2020 because of both lower precipitation amounts and a greater cumulative reference ETo (fig. 1). Because of the significant precipitation received in August 2016, differences between total irrigation for 100% and 33% levels were narrowed to 105 mm compared to 2018 (188 mm) and 2020 (223 mm) (table 2).

Crop Water Use

Crop water requirements were greatest during the 2020 growing seasons because of the greater cumulative reference evaporation (ETo) of 1134 mm from planting to harvest compared with 2016 (973 mm) and 2018 (960 mm). Cumulative measured crop ET (table 2) across cultivars averaged 627, 547, and 467 mm for the 100%, 66%, and 33% irrigation levels, respectively. Estimated runoff was a minor component of the water balance in all study years (table 2). Changes in stored soil water from planting to the final neutron gauge measurement across all cultivars, ranged from -145 to -30 mm, which reflects a net loss of stored soil water in all study years. In all years, differences between stored soil water (0–2.0 m) from planting to the final neutron probe measurement just prior to harvest did not vary with respect to cultivar (P = 0.177). Change in stored soil water at 0–2.0 m was significantly influenced by irrigation level only in 2018 (P = 0.011; table 2), with approximately 20 mm greater depletion under the 33% irrigation treatment compared with greater irrigation levels.

Observed soil water profile depletion by root uptake and replenishment with irrigation and precipitation during each year is illustrated for the FM2011 cultivar (fig. 2). From planting to harvest, change in stored soil water (?S) at all depth increments (0–0.4, 0.4–0.8, 0.8–1.4, and 1.4–2.0-m) was not influenced by cultivar in any study years or irrigation levels (P=0.086) except on one occasion (?S FM2011 < ?S DP1212F and Phy333 at the 33% irrigation level in 2018; P = 0.018), likely because of an initially low water content at planting. A considerable amount of stored soil water (19–55 mm) was extracted from the 0.8–1.4-m depth increment (fig. 2). Across all irrigation levels, this accounted for 9%, 5%, and 7% of the seasonal crop water use in 2016, 2018, and 2020, respectively. In 2018, estimated water extraction at 0.8–1.4 m was likely smaller because of the lower initial soil water contents at less than 1.3 m (fig. 2). Across all irrigation levels and cultivars, change in soil water at 1.4–2.0 m from planting to the final neutron gauge measurement averaged -11, 0, and -11 mm for 2016, 2018, and 2020, respectively.

Table 2. Crop water use and components of the soil water balance (0–2.0 m) from planting to the final neutron gauge measurement at each irrigation level pooled among cultivars for each year of the study. Precipitation during this period was 256, 150, and 183 mm in 2016, 2018, and 2020, respectively.
YearIrrigation
Level		Gross
Irrigation		Estimated
Runoff[a]		?S[b]Water
Use[c]
mmmmmmmm
2016100%223-35-133577
66%170-145536
33%118-138476
Adjusted LSD38
2018100%4220-33605
66%324-30506
33%234-51433
Adjusted LSD14
2020100%461-10-66700
66%354-70598
33%238-82493
Adjusted LSD22

    [a]    Runoff was approximated across all irrigation levels and cultivars for a 61 mm storm in 2016 and a 27 mm storm that occurred during the final hours of an irrigation application in 2020.

    [b]    ?S is the estimated stored soil water at planting subtracted from stored soil water the final neutron gauge measurement.

    [c]    Least Significant Differences (LSD) are based on the Tukey-Kramer adjustment for irrigation level comparisons in each year.

Within Season Soil Water Content Trends

Profile soil water contents throughout each growing season were segregated into three depth increments (0–0.4, 0.4–0.8, and 0.8–1.4 m) to permit interpretation of intra-season cultivar × irrigation level effects using an analysis of variance for each measurement day (figs. 3-5). In 2016 and 2018, irrigation significantly influenced soil water contents on only four and two days, respectively, suggesting that the additional water being applied at the 100% level was being utilized by the crops. However, in 2020, a significant irrigation effect was observed on nearly every day of the period, starting on DOY 174 immediately after the initiation of the irrigation treatment and extending to three weeks after cut-out. In all years, there was a steep decline in soil water for all cultivars at the 0.4–0.8 depth increment extending from first square to the after cutout. In 2016, water contents at this depth increment began to increase after cutout because of the significant rainfall during this period. Declines in soil water content at the 0.8–1.4 m depth increment were also evident in all years and at all irrigation levels.

Stored soil water contents throughout the 2016 growing season were similar among all cultivars except for the period between first square and cutout at the 0.4–0.8 depth increment, where the FM2011 and Phy333 cultivars tended to have significantly lower water contents compared with the other two cultivars. At first open boll, however, differences in soil water contents among cultivars were not significant. In 2018, a cultivar effect on soil water content was only observed on one day just prior to first square, where the DP1212 cultivar had a lower water content compared with the other cultivars, but only at the 100% irrigation level and at 0–0.4 m. In 2020, the DP1612 cultivar likewise had significantly lower water contents compared to the other cultivars on many days during a period extending from first square to a week after cutout. This was principally observed at the high irrigation level, and these significant cultivar differences, as in 2016, were no longer evident after first open boll. There were no significant differences in soil water contents at the lower depth increment (0.8–1.4 m) in any study year, suggesting similar root extraction rates at these deeper depths. Water content differences due to cultivar at shallower depth increments may have resulted from the better establishment and vigor of certain cultivars, which varied with year.

Figure 2. Mean profile water contents for the FM2011 cultivar at the 100% and 33% irrigation levels in 2016, 2018, and 2020. Dashed lines are water contents at 1.5 and 0.03 MPa.

Canopy Temperatures

From first white flower to two weeks after cutout, canopy temperatures at the 100% irrigation level were maintained at or below the 28–30°C range (figs. 6–8), which is associated with the thermal optimum for enzyme function and biomass accumulation in cotton (Mahan and Upchurch, 1988; Burke and Upchurch, 1989; Conaty et al., 2012). At the 33% irrigation level, canopy temperature depressions were smaller than the 100% level and began to exceed midday ambient air temperatures after cutout in 2020 (fig. 8). At first white flower in 2016, canopy temperatures at the 100% and 33% levels were nearly identical (fig. 6), reflecting the small, non-significant differences in soil water content between the two irrigation levels at 0.0–0.4 m (0.018 m3 m-3) and 0.4–0.8 m (0.008 m3 m-3). Canopy temperature depressions consistently increased on the days following irrigation.

In 2016 and 2020, significant canopy temperature differences between cultivars at the same irrigation level were not observed (figs. 6 and 8). Variation in canopy temperatures among IRT sensors within the same irrigation level was typically the greatest at midday (±1°C), as exhibited by the larger 90% confidence interval bands. In 2018, significant differences in canopy temperature between the FM2011 and PHY333 cultivars (P < 0.05) were observed at the 100% irrigation level on DOY 202–203 for approximately 3–4 hours after 13:00 hours (fig. 7). This pattern continued up to and including DOY 205 (not shown) with subsequent irrigation and rainfall likely terminating the trend. These results do not reflect the similar mean soil water contents on DOY 201–208 at the 0 to 0.4 m depth increment for PHY333 (0.268 m3 m-3) compared with FM2011 (0.257 m3 m-3) for DOY 201 and 208. These contradictory results may be explained by improved root exploration by the FM2011 cultivar earlier in the growing season, permitting it to access water near drip lines that are outside the measurement volume of the neutron probe.

Figure 3. Measured water contents for three depth increments during the 2016 growing season for each cotton cultivar at the 100% and 33% irrigation level. Asterisks indicate water contents significantly differed (P < .05) for at least one pair of cultivars. Number signs (#) in the 100% irrigation level plots indicate there was a significant irrigation effect on soil water content. Water contents at 1.5 MPa is shown as a dashed horizontal line. Field capacity was approximately 0.33 m3 m-3 for each of the depth increments. The predicted water content threshold for crop water stress [?T = ?pwp + (Sfc / zr) · (1 – p)], where ?pwp is the volumetric water content at wilting point, is shown as a thin, solid line.

Crop Coefficient Optimizations

Initial parameter optimizations for the modified FAO-56 crop coefficient curve included the crop coefficients Kc-ini,Kc-mid, and Kc-end, the parameter delineating the initiation of water stress (pb), and the nonlinear stress exponent (e). In contrast to the development of crop coefficients under solely full irrigation (Howell et al., 2004; Farahani et al., 2008; Ko et al., 2009), this model was fitted to the ETa of the deficit irrigated cotton as well as the 100% irrigation level. Initial fits of pb converged to a fitted parameter value of 0.25, which is considerably smaller than the 0.65 tabulated in FAO-56 (Allen et al., 1998) or 0.55 used by Howell et al. (2004). Although lower values of pb resulted in a lower root mean square error (RMSE) of predicted crop ET, especially at deficit irrigation levels, the predicted Ks at the 100% irrigation level was less than 0.9 from first white flower to cutout in 2018 and 2020, which is unlikely given the high lint yields obtained in these years and canopy temperatures that infrequently exceeded thermal optimums. Additionally, pb was negatively correlated with Kc-mid, which resulted in nonuniqueness and parameter identifiability problems in the optimization. To circumvent these problems, pb was fixed at a value of 0.4, which resulted in reduced RMSE’s compared with higher values proposed by Howell et al. (2004) and Allen et al. (199) yet maintained the simulated stress level Ks = 0.95 at the 100% irrigation level from first white flower to cutout. For this optimization, the crop coefficient Kc-end converged to a value of the lower constraint of zero. Because one would expect some evaporation to occur, Kc-end was set to 0.08, which corresponds to the evaporation rates of a dry soil surface in the late fall for this soil (Schwartz et al., 2019). The fitted values for the first and final day of the peak water use stage associated with Kc-mid were 12 days after first flower and 18 days after the onset of cutout, respectively.

Figure 4. Measured water contents for three depth increments during the 2018 growing season for each cotton cultivar at the 100% and 33% irrigation levels. Asterisks indicate water contents significantly differed (P < .05) for at least one pair of cultivars. Number signs (#) in the 100% irrigation level plots indicate that there was a significant irrigation effect on soil water content. Water contents at 1.5 MPa is shown as a dashed horizontal line. Field capacity was approximately 0.33 m3 m-3 for each of the depth increments. The predicted water content threshold for crop water stress (?T) is shown as a thin solid line.

The final optimizations of the model parameters after fixing the values of pb and Kc-end and establishing the timing of the mid-season stage resulted in a predicted ETc with a RMSE of less than 1 mm d-1 in all three study years (table 3; fig. 9). These coefficients are likely both soil (Pullman clay loam) and environment specific. In contrast to previous crop coefficient curves for an early maturing cultivar (Howell et al., 2004), our fitted mid-season stage was of shorter duration (20 compared with 30 days) and shifted later in the growing season. The fitted timing of peak water use corresponded more closely with the results of Bordovsky (2020). Because of the diminishing soil water availability after the final irrigation, the crop coefficient curve exhibited a relatively steep decline during the final stage that ended at harvest (fig. 9). As with other studies (e.g., Farahani et al., 2008), the steepness of the decline varied with year because of varying rates of senescence combined with precipitation received during this end stage. In this study, predicted ETc was relatively insensitive to Kc-end as indicated by the large 90% confidence interval for this parameter (table 3). In practice, a low value for Kc-end, should be used, even though it may not represent actual crop ET, so as to correctly plan irrigations associated with moderate water stress during boll maturation. The fitted value of the exponent e was 0.91 (table 3), indicating a slightly nonlinear response to water stress with a concave downward relationship (fig. 9). The use of larger values of pb in this study led to better predictions of crop ET and profile soil water contents at the 100% irrigation level, yet poorer predictions under deficit irrigation. Howell et al. (2004) noted a poor prediction of deficit irrigated cotton in their study using pb = 0.55, suggesting that the form of the FAO-56 crop coefficient model under non-standard conditions may be too restrictive to accommodate the greater water deficits in this study. By first square, lateral cotton roots have access to soil water in irrigated furrows (Hutmacher et al., 1999), and after irrigation, water contents would be expected to decline more at this location compared to below crop rows. Hence, the change in soil water determined using the neutron gauge positioned in the row would be expected to underestimate soil water contents in irrigated furrows (Bufon et al., 2012), the change in soil water after irrigations, and hence crop ET. This likely explains why pb converged to smaller values, simulating greater water stress and reduced crop ET, under the greater modeled soil water contents predicted with the assumed spatially uniform irrigation. This interpretation seems to be supported by the fact that as irrigation declined from 100% to 33%, predicted water contents were initially overestimated but progressively underestimated (fig. 9).

Figure 5. Measured water contents for three depth increments during the 2020 growing season for each cotton cultivar at the 100% and 33% irrigation level. Asterisks indicate water contents significantly differed (P < .05) for at least one pair of cultivars. Number signs (#) in the 100% irrigation level plots indicate there was a significant irrigation effect on soil water content. Water contents at 1.5 MPa is shown as a dashed horizontal line. Field capacity was approximately 0.33 m3 m-3 for each of the depth increments. The predicted water content threshold for crop water stress (?T) is shown as a thin, solid line.

Yield and Water Productivity

Across all hybrids, mean cotton lint yields at the 100% irrigation level were 1111, 1674, and 1806 kg ha-1 in 2016, 2018, and 2020, respectively. Potential lint yield based on total accumulated heat units (Gowda et al., 2007) was similar to the measured yield in 2016 (1180 kg ha-1) but considerably underestimated yield in 2018 (1218 kg ha-1) and 2020 (1358 kg ha-1). Irrigation effects on lint yield varied strongly with respect to the study year (table 4). In 2016, there was no significant response of lint yield to irrigation (P = 0.924) yet in 2018 and 2020, lint yield of all cultivars responded strongly to irrigation (P < 0.001). Lack of an irrigation response in 2016 is explained by the considerable amount of precipitation received from first white flower to first open boll in this year (186 mm) combined with reduced thermal energy during this period (GDD = 408°C) compared to 2018 and 2020 (GDD = 451 and 487°C, respectively). The combination of favorable precipitation yet reduced thermal energy during flowering likely increased the yield potential of the deficit irrigated cotton while limiting yield potential at the 100% irrigation level in 2016.

Figure 6. Mean 0.25-h canopy temperatures in 2016 for two cultivars from DOY 211–214 (first white flower) and from DOY 228–231 (three days after cutout) at the 100% and 33% irrigation levels. Shaded areas are the 90% confidence intervals for the PHY333 cultivar using the pooled variance. Portions of the line for the FM2011 cultivar that fall outside this shaded region indicate that the difference between mean canopy temperatures of the two cultivars is significantly different (P < 0.10). Ambient air temperature at 2 m is also shown. Number of observations: n = 6 for all cultivars and irrigation levels.

Cultivar influenced yield in all study years (P = 0.001), with the early maturity class cultivar FM2011 consistently in the highest yielding group and the medium maturity cultivars FM2484 and FM2334 in the lowest yielding group in 2016 and 2018 across all irrigation levels. The medium maturity cultivar FM2484 and Phy333 yields were significantly lower compared with the other cultivars in 2016, a year characterized by a cool, wet September and October. Analysis of the simple cultivar effect at each irrigation level in 2018 and 2020 shows that cultivars affected yield only at the 66% and 100% irrigation levels in 2018 (P = 0.01) and in 2020 (P = 0.05). In 2016, simple effect analysis showed that cultivars influenced yield across all irrigation levels (P = 0.02).

Water productivity across all hybrids at 100% irrigation averaged 0.19, 0.27, and 0.26 kg m-3 in 2016, 2018, and 2020, respectively. These values are 0.02–0.09 kg m-3 greater than the water productivities reported by Howell et al. (2004) at the same location. Water productivity values reported by Bordovsky (2020) for medium maturity cultivars in an environment with greater heat accumulation usually exceeded 0.30 kg m-3.

Figure 7. Mean 0.25-h canopy temperatures in 2018 for two cultivars from DOY 200–204 (a few days after first square at the 100% and 33% irrigation level). Shaded areas are the 90% confidence intervals for the PHY333 cultivar using the pooled variance. Portions of the line for the FM2011 cultivar that fall outside this shaded region indicate that the difference between mean canopy temperatures of the two cultivars is significantly different (P < 0.1). Ambient air temperature at 2 m is also shown. Number of observations: n = 4 for FM2011 at 100%; n = 6 for PHY333 at 100%; n = 3 for FM2011 at 33%; n = 2 for PHY333 at 33%. Irrigation was not applied to the 33% level during this period.

Water productivity of cotton was moderately correlated to crop water use in 2018 (R2 = 0.57), and 2020 (R2 = 0.66) and increased by an average of 0.031 kg m-3 for each additional 100 mm of ETa (fig. 10). In 2016, crop water productivity was uncorrelated to water use (R2 = 0.16), that reflects the lack of an irrigation effect on yield in this year.

Discussion

In semiarid zones, a root system capable of extracting water deep in the profile is a valuable trait. In this study, cotton extracted water to a soil depth of at least 1.8 m that was independent of the cultivar. As inferred by water extraction patterns, a greater maximum effective rooting depth of 1.8 m in 2016 and 2020 compared with 1.4 m in 2018 was associated with plant available water deeper in the profile. This interpretation is supported by observations of initial volumetric water content at the 1.2–1.8-m depth increment of 0.215 m3 m-3 in 2018 compared with 0.260 and 0.296 m3 m-3 in 2016 and 2020, respectively. The lower limit of water extraction by cotton for this depth increment in the Pullman soil is approximately 0.18 m3 m-3 (Tolk and Evett, 2012). Cotton root proliferation into wetted zones and associated with gradual root zone drying is well documented (Taylor and Klepper, 1974), with such adaptation only occurring in immature cotton plants prior to boll formation (Carmi et al., 1993). For the few observed differences in soil water depletion during the growing season among cultivars within each irrigation level, marginally greater soil water use did not influence lint yields. For example, the cultivar DP1612 exhibited greater mid-season profile water use compared with FM2011 in 2018 and 2020 (figs. 4 and 5), yet DP1612 exhibited a significantly lower yield in 2018 and a similar yield in 2020 (table 4).

Figure 8. Mean 0.25-h canopy temperatures in 2020 for two cultivars from DOY 216 – 219 (cutout) and from DOY 231 – 234 (two weeks after cutout) at the 100% and 33% irrigation level. Shaded areas are the 90% confidence intervals for the FM2334 cultivar using the pooled variance. Portions of the line for the DP1612 cultivar that fall outside this shaded region indicate that the difference between mean canopy temperatures of the two cultivars is significantly different (p < 0.10). Ambient air temperature at 2 m is also shown. Number of observations: n = 3 for DP1612 at 100%; n = 4 for FM2334 at 100%; n = 4 for both cultivars at 33%.

The small and non-significant differences in observed canopy temperatures between cultivars during critical growth stages at the same irrigation level suggest that plant height and leaf pubescence characteristics did not alter the energy balance at the canopy level. Canopy temperature measurements also imply similar transpiration rates among cultivars, corroborating the crop water use estimates in this study. Previous field (Baker and Myhre, 1969; Mahan et al., 2016; Pettigrew, 2016) and chamber (Bednarz and van Iersel, 2001) studies also found that differences in photosynthetic rates, canopy temperatures, and CO2 exchange rates among cultivars with diverse leaf characteristics were small and likely not physiologically important with respect to yield.

Table 3. Fitted and fixed parameters and root mean square errors (RMSE) of the daily crop ET for the optimization of the crop water use for all study years. Maximum rooting depth was set equivalent to 1.8, 1.4, and 1.8 m in 2016, 2018, and 2020, respectively. The timing and duration of the mid-season peak water use were set from 12 days after first flower and extended to 18 days after cut-out.
90% CI
ParameterValueLowerUpper
Kc-ini0.380.330.43
Kc-mid1.19 1.001.30[a]
Kc-end0.08 [b}00.64
pb0.40 [b}
d (d mm-1)0.040
ETb(mm d-1)5
e0.910.5251.63
2016 ETc RMSE (mm d-1)0.886
2018 ETc RMSE (mm d-1)0.806
2020 ETc RMSE (mm d-1)0.888

    [a]    Upper confidence value for Kc-mid set by the constraints of the average crop water stress coefficient Ks exceeding 0.95 for the 100% irrigation level in 2020 from first square to 7 days after cut-out.

    [b]    pb was set to a fixed value of 0.4 and Kc-end was set to a fixed value of 0.08 for the optimization.

The fitted FAO-56 crop coefficient for the initial stage under SDI was relatively high (Kc-ini = 0.38) and similar to locally developed values reported by Farahani et al. (2008) for surface drip irrigation. Hunsaker and Bronson (2021) reported values of 0.2 for SDI applied directly under crop rows. Hence, much of the benefit of SDI is lost for the alternate furrow SDI because of the need to wet the seed zone and establish the crop with a considerable amount of irrigation in most growing seasons. Given an average ETo of 344 mm during the initial crop stage for the three study years, there is an additional unproductive water loss of 344*(0.38–0.2) = 62 mm required for wetting the seed zone and crop establishment under alternate furrow SDI at this location.

Maintaining soil water contents that impose a small degree of water stress was likely key to preventing regrowth of cotton after cutout and facilitating boll maturation under conditions of marginal thermal energy in September and October. Managed levels of soil water content are best exemplified in 2020, which did not have excessive rainfall after cutout, and soil water content observations could be continued until harvest. In this year, the average plant available water fraction (0–1.4 m) from cutout to first open boll was 0.59 at the 100% irrigation level. From first open boll (7 Sep) to harvest (22 Oct), the crop utilized 72 mm of water, of which 26 mm was extracted from the soil and the remainder supplied by two small irrigations (12.4 mm each) and precipitation. After first open boll, plant-available water ranged from a maximum of 0.60 to a minimum of 0.35 at harvest. In an environment that typically receives more thermal energy than our study location, Lascano et al. (2017) demonstrated that cotton irrigation could be terminated at 1000°C heat units accumulated after emergence without a yield loss compared to irrigations terminated later in the growing season.

Figure 9. Predicted (ETc) and measured (ETa) crop ET for all study years, predicted and measured plant available water fraction for the 100%, 66%, and 33% irrigation levels in 2020, and predicted (Kc) and measured crop coefficients during the 2020 growing season. Also shown is the fitted crop stress coefficient (Ks) relationship as a function of the depletion fraction of plant available water for a range in crop ET (Kc · ETo) in units of mm d-1. The predictions are based on the optimization in table 3 across all cultivars. The timing of first white flower (WF) and first open boll (OB) are indicated for the crop coefficient relationship.

Lint yield under the 100% irrigation level was similar to average yields reported for irrigated production in counties with similar thermal energy accumulations in the Texas High Plains (Bell and Byrd, 2016; Bell and Heflin, 2018; Bell et al., 2020). Heat unit accumulations for 2016, 2018, and 2020 represent the 78th percentile or greater of the mean cumulative GDD from 1992–2020 at Bushland ( = 1127; s.d. = 132), and consequently, yields in these years would be expected to exceed what could be achieved in a normal year. Considering that a cotton stand was only successfully established in half of the study years and that in most years heat accumulation will be suboptimal, cotton production poses a relatively large production risk compared with other crops in the region. A high production risk level is also evident from the crop insurance indemnities for the portion of the THP overlying the High Plains Aquifer, 71% of which were attributed to cotton crop failure (Reyes et al., 2020). Cultivar selection for yield was important in every study year, especially at the high irrigation levels. The medium maturity class cultivars usually had the lowest yields across all irrigation levels, especially in 2016, a year with considerable precipitation and less heat accumulation during boll development and maturation. Within a given irrigation level, greater lint yields were not associated with greater profile water use, nor were they related to canopy temperature differences. Significantly greater yields for some cultivars may be a result of more effective partitioning of dry matter to lint yield. Growing short season cultivars in environments with limited heat units may benefit producers, especially those with low irrigation capacities.

Table 4. Mean lint yield and water productivity (WP) across cultivar (C) and irrigation (I) and the associated analysis of variance for main effects and interactions. Means followed by a different letter are significantly different within each effect at the 0.05 adjusted Tukey-Kramer probability level. Water productivity is determined using the crop water use evaluated for each I × C combination.
201620182020
MeansLint YieldWPLint YieldWPLint YieldWP
kg ha-1kg m-3kg ha-1kg m-3kg ha-1kg m-3
Cultivar
(C)		 DP1212/DP16121259a0.2381160b0.2211398ab0.235
FM20111423a0.2681356a0.2551532a0.258
FM2484/FM2334847b0.1591290ab0.2481252c0.210
Phy333972b0.1841378a0.2601354bc0.228
Irrigation
(I)		 100%1111a0.1921674a0.2711806a0.259
66%1117a0.2081237b0.2401406b0.236
33%1148a0.241978c0.219939c0.192
I × C I=100% C =DP1212/DP16121269abcd0.2171494cd0.2471896a0.271
I=100% C =FM20111546a0.2671804a0.2901988a0.287
I=100% C =FM2484/FM2334725g0.1261614bc0.2611630b0.233
I=100% C =Phy333903efg0.1581782ab0.2851713b0.245
I=66% C =DP1212/DP16121287abc0.2421034f0.1981427cd0.241
I=66% C =FM20111290abc0.2441289e0.2491525bc0.255
I=66% C =FM2484/FM2334932efg0.1701323de0.2611265de0.212
I=66% C =Phy333960defg0.1791300de0.2521407cd0.236
I=33% C =DP1212/DP16121220bcde0.259952f0.214872g0.177
I=33% C =FM20111433ab0.297975f0.2141082ef0.221
I=33% C =FM2484/FM2334884fg0.187933f0.214861g0.176
I=33% C =Phy3331054cdef0.2211052f0.235942fg0.194
Effect I0.924<0.001<0.001
C<0.0010.001<0.001
I × C0.3750.1510.405
Figure 10. Water productivity as a function of growing season measured crop evapotranspiration (ETa). Only the trendline is shown for 2016.

Conclusions

Within a given irrigation level, crop water use during the growing season did not differ among the cultivars evaluated in this study. Stored soil water exhibited few observed significant differences among cultivars throughout the growing season for the examined depth increments in each year of the study. Measured canopy temperatures from first white flower to two weeks after cutout did not significantly differ among hybrids within each irrigation level, which, to some extent, corroborates our interpretation that crop water use was similar among cultivars. The fitted mid-season crop coefficient (1.19) was similar to the tabulated FAO-56 value; however, the duration of this period was short (20 d) and began well after first flower, extending from a couple days before cutout to 18 days after cutout. The late season crop coefficient converged to values near zero and was considerably lower than reported by other studies or the FAO-56 tabulated value. This suggests that water savings could be achieved by reducing irrigation during boll maturation. Lint yield was significantly affected by cultivar in all study years and by irrigation in two of the three study years. Cultivar effects on yield were significant only at the 66% and 100% irrigation levels. Despite similar water use, across irrigation levels, the yield of one cultivar exceeded others in every study year, possibly due to the more effective partitioning of available photosynthate to productive bolls. Medium maturity class cultivars usually yielded less than the early maturity cultivars, especially in the year with less accumulation of thermal energy.

Acknowledgments

The authors gratefully acknowledge the important contributions of Bridgette Hiltbrunner, Beau Hiltbrunner, Grant Johnson, and Don McRoberts, Gavin Goodin (2016), Cache McClure (2016), Jonathon Russell (2017–2018), Caroline Huseman (2018–2020), and Nathan Ruthhardt (2017–2018) for plot establishment, maintenance, and data collection. This research was supported in part by the Ogallala Aquifer Program, a consortium between the USDA Agricultural Research Service, Kansas State University, Texas A&M AgriLife Research, Texas A&M AgriLife Extension Service, Texas Tech University, and West Texas A&M University.

References

Allen, R. G., Pereira, L. S., Raes, D., & Smith, M. (1998). Crop Evapotranspiration: Guidelines for computing crop water requirements. Irrigation and Drainage paper 56. Rome, Italy: United Nations FAO.

Allen, R. G., Walter, I. A., Elliott, R. L., Howell, T. A., Itenfisu, D., Jensen, M. E., & Snyder, R. L. (2005). The ASCE standardized reference evapotranspiration equation. Reston, VA: ASCE Press. https://doi.org/10.1061/9780784408056

Baker, D. N., & Myhre, D. L. (1969). Effects of leaf shape and boundary layer thickness on photosynthesis in cotton (Gossypium hirsutum). Physiol. Plant., 22(5), 1043-1049. https://doi.org/10.1111/j.1399-3054.1969.tb07463.x

Bange, M. P., & Milroy, S. P. (2000). Timing of crop maturity in cotton: Impact of dry matter production and partitioning. Field Crops Res., 68(2), 143-155. https://doi.org/10.1016/S0378-4290(00)00116-7

Baumhardt, R. L., Schwartz, R., Howell, T., Evett, S. R., & Colaizzi, P. (2013). Residue management effects on water use and yield of deficit irrigated cotton. Agron. J., 105(4), 1026-1034. https://doi.org/10.2134/agronj2012.0361

Bednarz, C. W., & van Iersel, M. W. (2001). Temperature response of whole-plant CO2 exchange rates of four upland cotton cultivars differing in leaf shape and leaf pubescence. Commun. Soil Sci. Plant Anal., 32(15-16), 2485-2501. https://doi.org/10.1081/CSS-120000386

Bell, J. M., & Byrd, S. (2016). 2016 Texas Panhandle Cotton Variety Trials. Amarillo, TX: Texas A&M AgriLife Extension. Retrieved from http://amarillo.tamu.edu/files/2017/02/2016-Texas-Panhandle-Cotton-Variety-Trials-Final-2-10-17.pdf

Bell, J. M., & Heflin, K. (2018). 2018 Texas Panhandle Replicated Agronomic Cotton Evaluation (RACE). Amarillo, TX: Texas A&M AgriLife Extension. Retrieved from https://amarillo.tamu.edu/files/2019/01/2018-Texas-Panhandle-RACE-Trial-Report.pdf

Bell, J. M., Maeda, M., Heflin, K., Naylor, C., & Sirmon, P. (2020). 2020 Texas Panhandle Replicated Agronomic Cotton Evaluation (RACE). Amarillo, TX: Texas A&M AgriLife Extension. Retrieved from https://amarillo.tamu.edu/files/2021/02/2020-Texas-Panhandle-RACE-Report.pdf

Bell, J. M., Maeda, M., Heflin, K., Naylor, C., & Sirmon, P. (2021). 2021 Texas Panhandle Replicated Agronomic Cotton Evaluation (RACE). Amarillo, TX: Texas A&M AgriLife Extension. Retrieved from https://amarillo.tamu.edu/files/2022/01/2021-Texas-Panhandle-RACE-Report.pdf

Bordovsky, J. P. (2019). Low-energy precision application (LEPA) Irrigation: A forty-year review. Trans. ASABE, 62(5), 1343-1353. https://doi.org/10.13031/trans.13117

Bordovsky, J. P. (2020). Preplant and early-season cotton irrigation timing with deficit amounts using subsurface drip (SDI) systems in the Texas High Plains. Irrig. Sci., 38(5), 485-499. https://doi.org/10.1007/s00271-019-00661-3

Bordovsky, J. P., & Mustian, J. T. (2020). Evaluation of cotton establishment strategies using subsurface drip irrigation (SDI) in the Southern High Plains of Texas. Appl. Eng. Agric., 36(6), 891-904. https://doi.org/10.13031/aea.14098

Bordovsky, J. P., Mustian, J. T., Ritchie, G. L., & Lewis, K. L. (2015). Cotton irrigation timing with variable seasonal irrigation capacities in the Texas South Plains. Appl. Eng. Agric., 31(6), 883-897. https://doi.org/10.13031/aea.31.10953

Bourland, F. M., Oosterhuis, D. M., & Tugwell, N. P. (1992). Concept for monitoring the growth and development of cotton plants using main-stem node counts. J. Prod. Agric., 5(4), 532-538. https://doi.org/10.2134/jpa1992.0532

Bufon, V. B., Lascano, R. J., Bednarz, C., Booker, J. D., & Gitz, D. C. (2012). Soil water content on drip irrigated cotton: Comparison of measured and simulated values obtained with the Hydrus 2-D model. Irrig. Sci., 30(4), 259-273. https://doi.org/10.1007/s00271-011-0279-z

Burke, J. J., & Upchurch, D. R. (1989). Leaf temperature and transpirational control in cotton. Environ. Exp. Bot., 29(4), 487-492. https://doi.org/10.1016/0098-8472(89)90027-0

Camp, C. R. (1998). Subsurface drip irrigation: A review. Trans. ASAE, 41(5), 1353-1367. https://doi.org/10.13031/2013.17309

Carmi, A., Plaut, Z., & Sinai, M. (1993). Cotton root growth as affected by changes in soil water distribution and their impact on plant tolerance to drought. Irrig. Sci., 13(4), 177-182. https://doi.org/10.1007/BF00190033

Conaty, W. C., Burke, J. J., Mahan, J. R., Neilsen, J. E., & Sutton, B. G. (2012). Determining the optimum plant temperature of cotton physiology and yield to improve plant-based irrigation scheduling. Crop Sci., 52(4), 1828-1836. https://doi.org/10.2135/cropsci2011.11.0581

Evett, S. R., & Steiner, J. L. (1995). Precision of neutron scattering and capacitance type soil water content gauges from field calibration. Soil Sci. Soc. Am. J., 59(4), 961-968. https://doi.org/10.2136/sssaj1995.03615995005900040001x

Evett, S. R., Tolk, J. A., & Howell, T. A. (2003). A depth control stand for improved accuracy with the neutron probe. Vadose Zone J., 2(4), 642-649. https://doi.org/10.2136/vzj2003.6420

Farahani, H. J., Oweis, T. Y., & Izzi, G. (2008). Crop coefficient for drip-irrigated cotton in a Mediterranean environment. Irrig. Sci., 26(5), 375-383. https://doi.org/10.1007/s00271-007-0101-0

Gerik, T. J., Faver, K. L., Thaxton, P. M., & El-Zik, K. M. (1996). Late season water stress in cotton: I. Plant growth, water use, and yield. Crop Sci., 36(4). https://doi.org/10.2135/cropsci1996.0011183X003600040017x

Gowda, P. H., Baumhardt, R. L., Esparza, A. M., Marek, T. H., & Howell, T. A. (2007). Suitability of cotton as an alternative crop in the Ogallala Aquifer Region. Agron. J., 99(6), 1397-1403. https://doi.org/10.2134/agronj2006.0275

Gwathmey, C. O., Bange, M. P., & Brodrick, R. (2016). Cotton crop maturity: A compendium of measures and predictors. Field Crops Res., 191, 41-53. https://doi.org/10.1016/j.fcr.2016.01.002

Gwathmey, C. O., Leib, B. G., & Main, C. L. (2011). Lint yield and crop maturity responses to irrigation in a short-season environment. J. Cotton Sci., 15(1), 1-10. Retrieved from http://www.cotton.org/journal/2011-15/1/upload/JCS15-1.pdf

Howell, T. A., Evett, S. R., Tolk, J. A., & Schneider, A. D. (2004). Evapotranspiration of full-, deficit-irrigated, and dryland cotton on the Northern Texas High Plains. J. Irrig. Drain. Eng., 130(4), 277-285. https://doi.org/10.1061/(ASCE)0733-9437(2004)130:4(277)

Hunsaker, D. J., & Bronson, K. F. (2021). FAO56 crop and water stress coefficients for cotton using subsurface drip irrigation in an arid US climate. Agric. Water Manag., 252, 106881. https://doi.org/10.1016/j.agwat.2021.106881

Hutmacher, R. B., Keeley, M. P., Davis, K. R., Ayars, J. E., & Peters, M. S. (1999). Cotton root systems in a clay loam soil: effects of growth stage, irrigation and nitrogen treatment. Proc. Beltwide Cotton Conf. National Cotton Council.

Kemmer, G., & Keller, S. (2010). Nonlinear least-squares data fitting in Excel spreadsheets. Nat. Protoc., 5(2), 267-281. https://doi.org/10.1038/nprot.2009.182

Ko, J., Piccinni, G., Marek, T., & Howell, T. (2009). Determination of growth-stage-specific crop coefficients (Kc) of cotton and wheat. Agric. Water Manag., 96(12), 1691-1697. https://doi.org/10.1016/j.agwat.2009.06.023

Lascano, R. J., Baumhardt, R. L., Goebel, T. S., Baker, J. T., & Gitz III, D. C. (2017). Irrigation termination thermal time and amount on cotton lint yield and fiber quality. Open J. Soil Sci., 7(9), 216-234. https://doi.org/10.4236/ojss.2017.79016

Mahan, J. R., & Upchurch, D. R. (1988). Maintenance of constant leaf temperature by plants — I. Hypothesis-limited homeothermy. Environ. Exp. Bot., 28(4), 351-357. https://doi.org/10.1016/0098-8472(88)90059-7

Mahan, J. R., Payton, P. R., & Laza, H. E. (2016). Seasonal canopy temperatures for normal and okra leaf cotton under variable irrigation in the field. Agriculture, 6(4), 58. https://doi.org/10.3390/agriculture6040058

Main, C. L., Barber, L. T., Boman, R. K., Chapman, K., Dodds, D. M., Duncan, S.,... Bronson, K. F. (2013). Effects of nitrogen and planting seed size on cotton growth, development, and yield. Agron. J., 105(6), 1853-1859. https://doi.org/10.2134/agronj2013.0154

McGuire, V. L. (2009). Water-level changes in the high plains aquifer, predevelopment to 2007, 2005-06, and 2006-07. Scientific Investigations Report 2009-5019. Reston, VA: USGS. https://doi.org/10.3133/sir20095019

McGuire, V. L. (2017). Water-level and recoverable water in storage changes, High Plains aquifer, predevelopment to 2015 and 2013–15. Scientific Investigations Report 2009-5019. 24. Reston, VA: USGS. https://doi.org/10.3133/sir20175040

McMaster, G. S., & Wilhelm, W. W. (1997). Growing degree-days: One equation, two interpretations. Agric. For. Meteorol., 87(4), 291-300. https://doi.org/10.1016/S0168-1923(97)00027-0

Pabuayon, I. L., Mitchell-McCallister, D., Lewis, K. L., & Ritchie, G. L. (2021). Yield and economic response of modern cotton cultivars to nitrogen fertilizer. Agronomy, 11(11), 2149. https://doi.org/10.3390/agronomy11112149

Pettigrew, W. T. (2004). Moisture deficit effects on cotton lint yield, yield components, and boll distribution. Agron. J., 96(2), 377-383. https://doi.org/10.2134/agronj2004.3770

Pettigrew, W. T. (2016). Cultivar variation in cotton photosynthetic performance under different temperature regimes. Photosynthetica, 54(4), 502-507. https://doi.org/10.1007/s11099-016-0208-8

Reyes, J., Elias, E., Haacker, E., Kremen, A., Parker, L., & Rottler, C. (2020). Assessing agricultural risk management using historic crop insurance loss data over the Ogallala Aquifer. Agric. Water Manag., 232, 106000. https://doi.org/10.1016/j.agwat.2020.106000

SAS Institute. (2014). SAS/STAT online documentation. Version 9.4. Cary, NC: SAS Institute.

Schwartz, R. C., Domínguez, A., Pardo, J. J., Colaizzi, P. D., Baumhardt, R. L., & Bell, J. M. (2020). A crop coefficient-based water use model with non-uniform root distribution. Agric. Water Manag., 228, 105892. https://doi.org/10.1016/j.agwat.2019.105892

Schwartz, R. C., Schlegel, A. J., Bell, J. M., Baumhardt, R. L., & Evett, S. R. (2019). Contrasting tillage effects on stored soil water, infiltration and evapotranspiration fluxes in a dryland rotation at two locations. Soil Tillage Res., 190, 157-174. https://doi.org/10.1016/j.still.2019.02.013

Taylor, H. M., & Klepper, B. (1974). Water relations of cotton. I. Root growth and water use as related to top growth and soil water content. Agron. J., 66(4), 584-588. https://doi.org/10.2134/agronj1974.00021962006600040031x

Tolk, J. A., & Evett, S. R. (2012). Lower limits of crop water use in three soil textural classes. Soil Sci. Soc. Am. J., 76(2), 607-616. https://doi.org/10.2136/sssaj2011.0248

Unger, P. W., & Pringle, F. B. (1981). Pullman soils: Distribution importance, variability, and management. Texas A&M Bulletin B-1372. College Station, TX: Texas Agricultural Experiment Station. Retrieved from https://hdl.handle.net/1969.1/126205

USDA-NASS. (2022). Cotton - Planted Acreage. Washington, DC: NASS. Retrieved from https://www.nass.usda.gov/Data_and_Statistics/index.php

USDA-RMA. (2020). Research of Current and Potential Subsurface Irrigation Methods for United States Department of Agriculture (USDA) Federal Crop Insurance Corporation (FCIC) Program. Final Research Report, Solicitation: RFQ1396983. College Station, TX: AgriLogic Consulting, LLC. Retrieved from https://www.rma.usda.gov/-/media/RMA/Publications/Risk-Management-Publications/ALC-SubSurface-Drip-Irrigation-Report_noappendices.ashx?la=en

Wanjura, D. F., Upchurch, D. R., & Mahan, J. R. (1992). Automated irrigation based on threshold canopy temperature. Trans. ASAE, 35(5), 1411-1417. https://doi.org/10.13031/2013.28748

Wanjura, D. F., Upchurch, D. R., & Mahan, J. R. (2004). Establishing differential irrigation levels using temperature-time thresholds. Appl. Eng. Agric., 20(2), 201-206. https://doi.org/10.13031/2013.15892

Witt, T. W., Ulloa, M., Schwartz, R. C., & Ritchie, G. L. (2020). Response to deficit irrigation of morphological, yield and fiber quality traits of upland (Gossypium hirsutum L.) and Pima (G. barbadense L.) cotton in the Texas High Plains. Field Crops Res., 249, 107759. https://doi.org/10.1016/j.fcr.2020.107759