Upscaling Wheat Grain and Forage Yield Response to Acidic Soils

Published online November 8, 2018 Crop Economics, Production, and Management

From Field Experiments to Regional Forecasts: Upscaling Wheat Grain and Forage Yield Response to Acidic Soils Romulo P. Lollato,* Tyson E. Ochsner, Daryl B. Arnall, Terry W. Griffin, and Jeffrey T. Edwards ABSTRACT We combined field studies to regional soil databases with the objective of presenting a protocol to forecast regional gains in winter wheat (Triticum aestivum L.) forage and grain yield and revenue, from liming or selecting a variety tolerant to acidic soils. First, we developed forage and grain yield response curves to soil pH using a variety by soil pH study conducted during 3 yr (2013– 2015) at two Oklahoma locations. Second, we used a database of soil pH samples representing 93% of the wheat growing region of the state (n = 11,905) coupled with 15-yr average county yield and harvested area to estimate potential gains for grain-only and dual-purpose scenarios. Relative grain yield maximized at soil pH of 5.8 for sensitive varieties and 4.8 for tolerant varieties. Forage yield maximized at soil pH of 6.0 for sensitive varieties and 5.5 for tolerant varieties. About 35% of the region had pH limiting to dual-purpose and 28% to grain-only production. Liming could improve grain-only statewide yield in 0.14 Mg ha–1 and revenue in US$19 ha–1, and adoption of tolerant varieties could increase yield in 0.11 Mg ha–1 and revenue in $10 ha–1. Liming could improve dual-purpose revenue in $37 ha–1 and variety selection in $28 ha–1 due to improved yield and forage. Potential additional statewide wheat production resulting from variety selection is 53,800 Mg and from liming 82,500 Mg. Our protocol can be used to aid development of agricultural policies and research prioritization at regional levels where acidic soils are prevalent.

Core Ideas • Wheat varieties differ in forage and grain yield in their tolerance to acidic soils. • Grain yield maximized at soil pH of 5.8 and 4.8 for sensitive and tolerant varieties. • Forage yield maximized at soil pH of 6.0 and 5.5 for sensitive and tolerant varieties. • We developed a protocol to estimate regional gains from variety selection or liming acidic soils.

A

cidic soils limit crop production in many regions across the world, as approximately 30 to 40% of current arable land has soil pH < 5.5 (von Uexküll and Mutert, 1995). Acidic soils usually develop naturally from parent materials low in base cations or in regions with large amounts of annual precipitation that leach base cations from the soil profile (Samac and Tesfaye, 2003). Acidic soils can also have an anthropogenic origin resulting from the removal of base cations as forage or grain harvest (Zhang and Raun, 2006) or due to the continuous use of ammonium-based fertilizers, which leads to topsoil acidification (Schroder et al., 2011). Major wheat growing environments characterized by acidic soils include China (Guo et al., 2010), Australia (Fisher and Scott, 1993; Tang et al., 2003), India (Behera and Shukla, 2015), France, United Kingdom, Germany, and other European countries (Panagos et al., 2012), countries in south and central America such as Brazil (de Sousa, 1998; Caires et al., 2008), Chile (Meriño-Gergichevich et al., 2010), and Mexico (Manske et al., 2001), and different regions within North America, including southeast United States (Gascho and Parker, 2001), northwest United States and southwest Canada (Mahler et al., 1985), and the US Great Plains (Johnson et al., 1997). Increased Al solubility in acidic soils can result in Al toxicity to the crop, which is one of the major causes of crop failure in extremely acidic soils (Schroder et al., 2011). Potassium chloride-extracted Al (AlKCl) and Al saturation (Alsat) have an inverse exponential relationship with soil pH, and small decreases in soil pH can lead to increases of great magnitude in AlKCl and Alsat at soil pH levels < 5 (Kariuki et al., 2007; Lollato et al., 2013). Aluminum toxicity symptoms include decreased root growth and reduced capacity of the root system to explore the soil for moisture and nutrients (Tang et al., 2003), and reduced aboveground biomass (Kaitibie et al., 2002; Kariuki et al., 2007) and grain yield (Kariuki et al., 2007; Valle et al., 2009). If the soil has acceptable levels of base cations, the presence of AlKCl per se may not induce Al toxicity (Johnson et al., 1997). Hence, a more consistent indicator of Al toxicity potential is the Alsat, a measure of Al concentration expressed as a percentage of total exchangeable base cations (i.e., Ca, Mg, K, and Na) of the soil (Sumner and Miller, 1996). Due to

Published in Agron. J. 111:1–16 (2019) doi:10.2134/agronj2018.03.0206 Supplemental material available online Available freely online through the author-supported open access option

R.P. Lollato, Dep. of Agronomy, Kansas State Univ., Manhattan, KS 66506; T.E. Ochsner, D.B. Arnall, J.T. Edwards, Dep. of Plant and Soil Sciences, Oklahoma State Univ., Stillwater, OK 74078; T.W. Griffin, Dep. of Agricultural Economics, Kansas State Univ., Manhattan, KS 66506. Received 26 Mar. 2018. Accepted 15 June 2018. *Corresponding author ([email protected]).

Copyright © 2019 by the American Society of Agronomy 5585 Guilford Road, Madison, WI 53711 USA This is an open access article distributed under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abbreviations: Al KCl, potassium chloride-extracted aluminum; Alsat, aluminum saturation; BI, Sikora buffer index; DAS, days after sowing; ECEC, effective cation exchange capacity; ETo, cumulative reference evapotranspiration.

A g ro n o my J o u r n a l • Vo l u m e 111, I s s u e 1 • 2 019

1

differences in soil chemical properties and texture, fields with similar soil pH can result in vastly different Alsat (Johnson et al., 1997) and crop grain yield responses (Kariuki et al., 2007). Broadcast and incorporated agricultural lime is the recommended method for managing acidic soils, and its effectiveness is well documented (Kaitibie et al., 2002; Liu et al., 2004; Tang et al., 2003). However, liming does not always result in increased wheat grain yield (Liu et al., 2004; Lollato et al., 2013) and its economics makes producers reluctant due to large amounts of product required and transportation costs (Sloan et al., 1995). Producers can adopt Al-tolerant wheat varieties (Johnson et al., 1997), as wheat sensitivity to soil pH is cultivar specific (Kariuki et al., 2007; Valle et al., 2009). Tolerance to soluble Al has been associated with exudation of organic acids by the roots, which increases nutrient solubility and can complex Al in the soil solution (Yang et al., 2004). Still, even cultivars with increased tolerance to Al may be impacted by high Al levels. Kariuki et al. (2007) showed that Alsat above 30% led to complete crop failure in Al sensitive cultivars and reduced yield in Al tolerant cultivars in 12 to 52%. Testing wheat cultivars in extremely low pH soils versus a limed control is one approach adopted to rank cultivars in their response to soil pH (e.g., Carver et al., 2003). However, this approach limits the possibility of optimizing pH for individual cultivars by restricting the range of pH needed to obtain a response threshold (Kariuki et al., 2007). Another methodology to quantify cultivar responses to soil pH is to measure root development of plants grown in nutrient solution with various concentration of Al (Bolt, 1996). Still, results obtained from nutrient-solution cultures may differ from field conditions and there are insufficient quantitative data to extrapolate results from one environment to the other (Kariuki et al., 2007). To overcome this barrier, testing modern wheat cultivars over a soil pH gradient (e.g., 4.0 to 7.0) is a valuable option to determine critical pH and Al concentration thresholds below which liming might be favored over variety selection. This approach has been adopted for crops such as grain sorghum (Butchee et al., 2012), sunflower (Sutradhar et al., 2014), and wheat (Valle et al., 2009; Kariuki et al., 2007). While these studies quantified individual varieties’ responses to a pH gradient, no attempts have been made to upscale results from such field experiments to explore potential yield gains at the regional scale due to liming or selecting a tolerant wheat variety. Our goal is to present a protocol including field experiments and up-scaling procedures to assess the regional effects of liming and wheat variety selection on wheat forage and grain yield, and revenue. We used Oklahoma, an important hard winter wheat growing region in the United States characterized by predominance of acidic soils, as a case study. Our specific objectives were to (i) quantify the effects of a pH and Al-concentration gradient on growth, development, and forage and grain yields of different wheat varieties grown under dual-purpose and grain only management and (ii) use a soil pH database to upscale field experiment results and quantify the probable positive regional-scale responses to liming versus selecting a tolerant wheat variety. MATERIAL AND METHODS Study Region Hard red winter wheat is sown in approximately 2.4 million hectares per year in Oklahoma, representing as much as 75% of total planted cropland in the state and around 20% of 2

the total winter wheat planted in the US Great Plains (USDANASS, 2017). Agricultural fields are characterized by mild slopes (1–6%) and predominant soil textures are silt loam and silty clay loam. While subsurface acidity is not a common concern (Zhang and Raun, 2006), more than 35% of the fields in the central wheat growing region of Oklahoma have topsoil pH < 5.5 (Zhang et al., 1998). Thus, acidic soils are a potential cause of wheat yield gaps (Patrignani et al., 2014). Wheat is managed as both a forage and grain crop (i.e., dual-purpose) in about one-half of the state’s wheat area (True et al., 2001), offering producers a potentially more stable source of income by producing both stocker cattle (Bos taurus L.) and grain (Redmon et al., 1996). Wheat managed as dual-purpose is generally sown mid-September at about 135 kg seed ha–1, which allows for greater biomass production and extended grazing period as compared to grain-only fields, which are typically sown mid-October at ~67 kg seed ha–1 (Edwards et al., 2011). As a tradeoff, dual-purpose production typically reduces winter wheat grain yield about 14% (Edwards et al., 2011). Field Experiment Sites, Experimental Design, and Treatments

The field experiment was designed to develop response curves of forage and grain yield of varieties differing in susceptibility to soil acidity. It was initiated in two locations in the 2012–2013 winter wheat growing season and was conducted over 3 yr for a total of six site-years. Soils were a Dale silt loam (fine-silty, mixed, superactive, thermic Pachic Haplustolls) near Chickasha (35.05° N, 97.91° W, and elevation 333 m), and a Easpur loam (fineloamy, mixed, superactive, thermic Fluventic Haplustolls) near Stillwater (36.13° N, 97.10° W, and elevation 270 m). A 6 × 4 complete factorial treatment structure was arranged in split-plot design with three replications. Whole plots were target soil pH (i.e., 4.0, 4.5, 5.0, 5.5, 6.0, and 7.0) and sub-plots were wheat cultivar: ‘Ruby Lee’, ‘Duster’, ‘TAM 203’, and ‘2174’. Whole plots were 7.6 × 7.6 m and arranged in a randomized complete block design, and subplots were 6.1 × 1.2 m completely randomized within main plots. Subplots were separated by a 1.5-m buffer alley to avoid potential influence from the neighboring target pH, resulting in larger whole plot than subplots. The varieties selected are commonly grown hard red winter wheat varieties in the US southern Great Plains varying in pH tolerance. Duster and Ruby Lee together accounted for 19.1% of the wheat area grown in Oklahoma during the 2014–2015 growing season (USDA-NASS, 2015). TAM 203 is highly susceptible to acidic soils (Edwards et al., 2012), Ruby Lee has moderate susceptibility (Edwards, 2013), 2174 has moderate tolerance (Kariuki et al., 2007), and Duster has high tolerance to acidic soils (Edwards et al., 2012). Hydrated lime [Ca(OH)2] and Al sulfate [Al2(SO4)3] were applied to each whole plot to obtain the target soil pH. The amount of material needed to reach a given target soil pH was determined with a laboratory experiment conducted prior to the field experiment to develop a response curve for the soil of each location. In this approach, composite soil samples were collected from both experimental sites to characterize initial soil pH at the 0–15 cm depth using a combination pH electrode in a 1:1 soil/deionized water suspension (Thomas, 1996). Subsamples weighing 500 g were taken from each composite sample and mixed with five incremental rates of Al2(SO4)3 or Ca(OH)2 .

Agronomy Journal • Volume 111, Issue 1 • 2019

Table 1. Initial soil pH, extractable Al (AlKCl), Ca, Mg, K, effective cation exchange capacity (ECEC), and Al saturation (Alsat) for the 0–15 and 15–45 cm soil layers at Chickasha and Stillwater, OK. Amount of hydrated lime [Ca(OH)2] and aluminum sulfate [Al2(SO4)3] required to change soil pH by a unit are also shown. Location Depth Initial soil pH AlKCl Ca Mg K ECEC Alsat Ca(OH)2 Al2(SO4)3 cm ————————— cmolc kg–1 ————————— % — Mg product per unit pH — Chickasha 0–15 6.2 0 7.6 4.1 0.5 12.2 0 2.4 1.6 15–45 7 0 9.1 5.3 0.4 14.8 0 Stillwater 0–15 5.2 γ

[1]

where Y is the response (percent wheat emergence or grain yield), X is soil pH, and γ is the level of soil pH in which an increase in soil pH did not result in increased response. Linear-plateau models were built using PROC NLIN in SAS Version 9.3 (SAS Institute, Cary, NC). Dynamics of canopy


3

development were modeled by variety, soil pH, location, and year, as a sigmoidal function of days after sowing (DAS) using the nonlinear regression model:

Y=

a 1+ e

 t −t  − 0   b 

[2]

where a is the asymptotic maximum percent canopy cover, t is time (DAS), t0 is the inflection point at which the rate in percent canopy cover increase is maximized (DAS), and b is a parameter determining the shape of the curve. Sigmoid functions are suggested when evaluating crop growth as function of time (Archontoulis and Miguez, 2015). Forage yield and relative grain yield across years were modeled as exponential rise to maximum function of soil pH using the nonlinear regression model: Y = β0 + α(1– e–β1x)

[3]

where β1 represents the responsiveness of Y to an increase in soil pH. Relative yield was determined by location, year, and variety, by expressing the yield of each plot relative to the average yield of the three plots with the highest soil pH for the same variety. The exponential model was then used to evaluate soil pH threshold in which forage yield or relative grain yield reached 95% of asymptotic maximum and to estimate the x-intercept, which indicates the pH below which yield is zero. This model has been widely used in experiments assessing crop responses characterized by decreasing marginal returns (Cerrato and Blackmer, 1990; Edwards and Purcell, 2005; Butchee and Edwards, 2013). Analyses based on Eq. [2] and [3] were performed using SigmaPlot 11 (Systat Software, San Jose, CA). Percent emergence, forage yield, grain volume weight, and grain protein concentration, were analyzed treating Alsat as a covariate. Variety was treated as a nominal variable whereas Alsat was a continuous variable in the same model, and we tested linear models to describe the relationship between the dependent and independent variables. Based on the shape of the response of each variety’s grain yield to Alsat at each site-year, grain yield was modeled as a plateau-linear or a linear function of soil Alsat. Threshold Alsat above which an increase in Alsat led to linear decrease in grain yields was established as the inflection point of the plateau-linear model. Relative grain yield was modeled as a linear function of Alsat. Analyses of covariance and plateau-linear models were performed in SAS Version 9.3 using PROC GLM and PROC NLIN. Upscaling Experimental Results to State Level Our second objective was to quantify the potential economic benefit from either liming or selecting a variety with greater tolerance to acidic soils at a state level. To do so, we used a large database of soil samples submitted to the Oklahoma State University Soil, Water and Forage Analytical Laboratory (SWFAL) for chemical analysis during the period 2013 through 2016, which encompassed the entire period when the field study was conducted and one additional year (i.e., 2016). This database was subjected to a five-step analysis as described below. First, we selected only surface soil samples that were collected from wheat fields across Oklahoma and that were evaluated for soil pH, for a total of 4347 samples in 2013, 4263 in 2014, 3055 4

in 2015, and 2662 in 2016. The short time period evaluated allowed us to assume minimal temporal effects in soil pH and combine the 4 yr for a total of 14,327 soil pH samples. County of origin was available for all samples, but we had no knowledge about specific sampling site. We filtered the initial dataset to select only counties corresponding to more than 0.5% wheat harvested area in the state and with a minimum of 15 reported samples over the 4-yr period to eliminate samples originating from regions where wheat is not representative and ensure a minimal number of samples was available from each studied county. The final dataset was comprised by 11,905 soil pH samples originated from 30 counties from the winter wheat growing region in the state, which represented 93% of total wheat growing region of Oklahoma (Fig. 1a). The second step was to evaluate the distribution of soil pH within the wheat growing region of Oklahoma to quantify the frequency of fields in which soil pH was in the responsive range to both lime and variety selection using thresholds developed from the field study. In this step, we used descriptive statistics and histograms of soil pH distribution (Fig. 1b). The third step was to estimate lime requirements for the samples resulting from steps one and two. Lime recommendations in Oklahoma are based on the Sikora buffer index (BI), which mimics and provides similar lime recommendations to those by the Shoemaker–McLean–Pratt buffer however without using hazardous chemicals (Sikora, 2006). Soil pH is independent of BI (Zhang and Raun, 2006); thus, the dataset had a large variability in BI for each soil pH range (Fig. 1c). Sikora BI was known for less than half of the samples in the database (n = 5000); thus, we performed the remaining analyses using three hypothetical soils representing a range of BI. The first hypothetical soil was a low BI soil, which represents a soil with high ECEC and therefore high buffer capacity [this soil corresponded to the quantile regression of BI (dependent variable) as affected by soil pH (independent variable) using quantile: 0.05]. The second hypothetical soil represented the mean BI of all soil samples submitted to SWFAL during the study period (quantile: 0.5), and the third hypothetical was soil characterized by high BI (quantile: 0.95, Fig. 1c), corresponding to soils with low ECEC and low buffer capacity (typically sandier soils). After developing the quantile regressions described above, three BI values were estimated for each soil pH value entered in the database (n = 11,905), and lime requirements to increase soil pH to 6.5 were calculated for each of the three hypothetical soils using the equation shown in Fig. 1d (Zhang and Raun, 2006). The fourth step was to calculate forage and grain yield as function of either liming or selecting a tolerant variety by county for both dual-purpose and grain-only systems. A diagram detailing each step adopted is shown in Fig. 2. Our cost analyses were performed for a scenario in which producers currently have seed of a pH sensitive variety (and have seed storage costs for that particular variety), and are considering whether to switch to a pH tolerant variety or to apply agricultural lime. Forage and relative grain yield were calculated for each soil pH value entered in the database for a sensitive (n = 11,905) and a tolerant (n = 11,905) variety using the equations developed in the field portion of the study (Fig. 2). We used the absolute forage values measured in our field experiment and their response curves to soil pH because the range in measured values is similar to long-term experiments measuring forage in central Oklahoma and there are no official


Fig. 1. Description of the methodology used to upscale wheat grain and forage yield as affected by variety selection and liming from field experiments to state level. (a) Oklahoma counties selected for the analysis based on a minimum 15 soil pH measurements and wheat acreage of 0.5% or more of total state wheat acreage. (b) Distribution of 11,905 soil pH measurements retrieved during 2013– 2016 from the selected counties shown in Panel a, with soil pH thresholds for tolerant and sensitive varieties to soil acidity for grain-only (G.O.) and dual-purpose (D.P.) systems and respective number of samples with pH below the threshold. Percent values represent proportion over total number of samples, and database descriptive statistics are also shown. (c) Relationship between soil pH and Sikora buffer index (BI) in a subset (n = 5000) of samples shown in Panel b, with hypothetical soils with low BI (quantile = 0.05), average BI (quantile = 0.5), and high BI (quantile = 0.95). (d) Effective calcium carbonate equivalent (ECCE) requirement as function of Sikora buffer index, adapted from Zhang and Raun (2006). Solid line represents the fitted cubic function of lime requirement as function of Sikora buffer index, and dashed lines represent 95% confidence band of the fitted model.

statistics reporting wheat forage yield by county. For grain yield, we used the shape of the nonlinear response of relative grain yields developed in our field experiment coupled with wheat grain yield reported by county as the average of the 2008–2017 seasons (USDA-NASS, 2017). We assumed that dual-purpose wheat yields 93% and grain-only yields 107% of the average county yield (Edwards et al., 2011). Forage and grain yield for the hypothetical lime treatment were calculated per county (n = 30) using the values for a sensitive variety at a soil pH of 6.5 (i.e., non-limiting). To calculate net income, forage yield was first converted into beef yield considering that 1 Mg ha–1 wheat forage at 300 g kg–1 crude protein converts to 100 kg ha–1 beef (Doye et al., 2017). Then, gross income was calculated considering the grain commercialized at US$0.14 kg–1 and beef commercialized at $2 kg–1. Input costs were $0.032 kg–1 for lime, $29.6 ha–1 to buy seed of a resistance variety, and $8.8 ha–1 to save seed of a susceptible variety. Lime needs were calculated per sample for the three hypothetical soils described above (n = 35,745). Income differences were calculated between (i) switching to a tolerant

variety versus current sensitive variety for both grain-only and dual-purpose systems (n = 23,810); (ii) liming versus non-liming current sensitive variety for each of the three hypothetical soils (n = 71,430), and (iii) switching to a resistant variety versus liming for each of the three hypothetical soils (n = 71,430). The final step was to upscale grain and forage yield and income differences from sample to county- to state-level. In this step, state-level differences for variety selection or liming at both dual-purpose and grain-only systems individually or combined [the latter assuming that 50% of the wheat area in Oklahoma is grazed (True et al., 2001)] were estimated as the weighted average of the harvested wheat area in each studied county. The approach consisted of (i) collecting wheat harvested area per county between 2008 and 2017 (USDA-NASS, 2017); (ii) using the distribution of BI values from each county to assign weights to each hypothetical soil (e.g., low, average, and high BI) and estimate lime needs for each county; and (iii) estimating the weighted average grain and forage yield and net income for the state weighted by harvested wheat area by county.


5

Fig. 2. Flow diagram describing the methodology used to calculate differences in income, grain yield, and forage yield for a producer currently sowing a pH sensitive cultivar and considering purchasing seed of a tolerant variety or applying agricultural lime, and upscaling procedure to state-level. Equations used to calculate grain and forage yield for grain-only (GOgrain) and dual-purpose (DPgrain and DP forag.) were derived from the field experiments and are shown on Fig. 5 (forage) and 7 (yield).

RESULTS AND DISCUSSION Field Experiment Amendment Effects on Soil pH, Extractable Aluminum, and Aluminum Saturation

Hydrated lime or Al2(SO4)3 applied to the Easpur loam in Stillwater resulted in 49 out of 54 measured soil pH within ±0.4 from the target pH, creating a wide pH gradient (i.e., 4.1 to 7.4) for the dual-purpose analysis. Actual soil pH values were further apart from the targeted pH in the Dale silt loam at Chickasha during the 2012–2013 growing season, with final soil pH gradient ranging from 5.4 to 6.3. Additional amendments improved this relationship for a total of 29 out of 36 6

measured pH samples within ±0.4 from the target pH during 2013–2014 and 2014–2015, resulting in a soil pH gradient ranging from 4.4 to 7.6 for the grain-only analysis. Extractable AlKCl across the three studied growing seasons ranged from 0 to 1.53 cmolc kg–1 in Stillwater, and from 0 to 0.71 cmolc kg–1 in Chickasha (Fig. 3a), and Alsat ranged from 0.85, p < 0.001); thus, slight changes in soil pH when < 5 resulted in dramatic increases in both AlKCl and Alsat. For instance, a decrease in soil pH from 5 to 4.5 increased AlKCl from 0.24 to 0.66 cmolc kg–1 while a decrease in soil pH of same magnitude Agronomy Journal • Volume 111, Issue 1 • 2019

reference evapotranspiration (ETo, Allen et al., 1998) during the same period was 369 mm at Stillwater and 238 mm at Chickasha. Despite the dry fall, favorable spring weather allowed for average wheat yields of 4.16 Mg ha–1 at Stillwater and 4.26 Mg ha–1 at Chickasha. The opposite weather pattern occurred during 2013– 2014, when a moist fall was followed by a dry spring (Table 2). The severe spring water deficit (ETo minus precipitation of 302 mm in Stillwater and 277 mm in Chickasha) restricted wheat yields to 1.98 Mg ha–1 at Stillwater and 1.62 Mg ha–1 at Chickasha. The 2014–2015 growing season had available moisture during both fall and spring (Table 2), and yields averaged 3.33 Mg ha–1 at Stillwater and 3.44 Mg ha–1 at Chickasha. The differences in grain yield levels among studied growing seasons did not allow for combination of years when analyzing absolute wheat yields. The six site-years evaluated had very contrasting weather conditions, representing very well the 10-yr mean rainfall for central and eastern Oklahoma, which ranges from 369 to 853 mm (Patrignani et al., 2014). Nonetheless, yields were below maximum wheat yields attained under best management practices in the region (i.e., 3.5 to 7.7 Mg ha–1; Lollato and Edwards, 2015) and were below the longterm water-limited yield (~5.2 Mg ha–1; Lollato et al., 2017), likely due to low soil pH and dual-purpose management in Stillwater. Wheat Emergence and Canopy Dynamics as Affected by Soil Acidity

Fig. 3. (a) Potassium chloride extractable aluminum (AlKCl) and (b) aluminum saturation (Alsat) as affected by soil pH in an Easpur loam at Stillwater, OK, and in a Dale silt loam at Chickasha, OK, during the growing seasons 2012–2013, 2013–2014, and 2014–2015.

from 6.5 to 6 only increased AlKCl from 0.01 to 0.03 cmolc kg–1 in the Easpur loam in Stillwater. The Dale silt loam in Chickasha had less AlKCl at a given soil pH but the shape of the response to soil pH was similar (Fig. 3). Extractable Al levels are soil specific and different soils can result in vastly different AlKCl levels (Johnson et al., 1997). The Easpur loam in Stillwater had greater AlKCl and Alsat than the Dale silt loam in Chickasha at a given soil pH, and our results differ considerably than those previously reported for other soils. For instance, a soil pH of 4.7 in our study resulted in AlKCl of 0.44 cmolc kg–1 in Stillwater and 0.08 cmolc kg–1 in Chickasha, as compared to AlKCl of 1.41 cmolc kg–1 in a Konawa fine loamy soil (Kariuki et al., 2007), and AlKCl of 0.62, 0.39, and 0.32 cmolc kg–1 in a Teller sandy loam, a Taloka silt loam, and a Grant silt loam (Sutradhar et al., 2014). At a soil pH of 4.7, Alsat was 23.9% at Stillwater and only 3.3% at Chickasha due to a stronger base concentration. At the same soil pH level of 4.7, previous studies in other regions demonstrated soil Alsat ranging from 17.4 to 47.7% (Kariuki et al., 2007; Sutradhar et al., 2014). Weather Conditions

The growing season of 2012–2013 had a dry fall, with cumulative precipitation from sowing to December of 65 mm at Stillwater and 58 mm at Chickasha (Table 2). Cumulative

The linear-plateau model explained percent wheat emergence as affected by soil pH well (Supplemental Table S1). Low soil pH decreased percent wheat emergence during the three studied growing seasons at Stillwater and during the 2013–2014 and 2014–2015 growing seasons at Chickasha for the pH sensitive varieties (e.g., TAM 203 and Ruby Lee). The soil pH levels achieved at Chickasha in 2012–2013 were far from the pH the goal, resulting in Alsat below 1%, which may explain the lack of response in this site-year. Threshold soil pH for maximum wheat emergence (γ) ranged from 4.5 to 5.9 depending on growing season and wheat variety. On average, the resistant variety Duster achieved maximum percent emergence at soil pH ≥ 4.8, whereas the varieties 2174, Ruby Lee, and TAM 203, achieved maximum emergence at pH of 5.1, 4.9, and 5.2 at Stillwater. Wheat emergence for tolerant varieties (i.e., Duster and 2174) was not affected by low soil pH at Chickasha, while it decreased with soil pH < 5.2 for sensitive varieties (Supplemental Table S1). Analysis of covariance between wheat emergence and soil Alsat indicated that wheat variety was a significant factor affecting percent wheat emergence in the six studied site-years, and Alsat affected wheat emergence in all site-years except Chickasha 2012–2013. Increased Alsat decreased wheat emergence as much as 7.3% per unit increase in Alsat, and the effects of Alsat on wheat emergence were more apparent at Stillwater (Alsat < 22.5%) than at Chickasha (Alsat < 7.8%). For every unit increase in Alsat, percent wheat emergence decreased in 4.6% for the variety TAM 203, in 7.3% for Ruby Lee, in 6.1% for Duster, and in 5.7% for 2174. Although the effects of Al on wheat germination have been studied under controlled conditions (de Lima and Copeland, 1990; Jamal et al., 2006), most of the research conducted under field conditions do not report decreased wheat emergence due to low soil pH or high Al concentration. de Lima and Copeland (1990) reported that high concentrations of Al were necessary to inhibit the growth of the emerging roots and shoots of


7

Table 2. Cumulative precipitation (Precip.), reference evapotranspiration (ETo), and water balance (WB; ETo –Precip.) for the fall, winter, and spring portions of growing seasons 2012–2013, 2013–2014, and 2014–2015 at Stillwater and Chickasha, OK. 2012–2013 2013–2014 2014–2015 Location Season Precip. ETo† WB Precip. ETo WB Precip. ETo WB ————————————————————— mm ————————————————————— Stillwater Fall‡ 65 369 –304 148 325 –177 228 315 –87 Winter§ 132 201 –69 43 227 –184 74 187 –113 Spring¶ 393 419 –26 197 499 –302 414 344 70 Growing season total 590 989 –399 388 1051 –663 716 846 –130 Chickasha Fall 58 238 –180 111 202 –91 202 225 –23 Winter 138 206 –68 46 226 –180 92 200 –108 Spring 456 444 12 254 531 –277 628 303 325 Growing season total 652 888 –236 411 959 –548 922 728 194 † Reference evapotranspiration, calculated according to the FAO 56 procedure (Allen et al., 1998). ‡ Fall encompasses the months of September, October, November, and December at Stillwater (dual-purpose site), and October, November, and December at Chickasha (grain-only site). § Winter encompasses the months of January, February, and March. ¶ Spring encompasses the months of April, May, and the first 15 d of June.

germinating seedlings. Similarly, Jamal et al. (2006) reported that seed germination was not affected by Al concentration up to 1.78 cmolc kg–1 applied to the seed, but root, shoot, and seedling length were. These reports support the strong acidity necessary to reduce wheat emergence measured in our study and suggest that the decreased emergence was likely due to an effect of soil acidity on the emerging roots and shoot of the seedling rather than decreased seed germination. The sigmoidal model explained dynamics of canopy cover development as affected by soil pH within site-year, variety, and pH range well (Fig. 4). Soil pH in the 4 to 4.5 range reduced the asymptotic maximum percent canopy cover (a) as compared to soil pH levels > 7 from 85.3 to 42.7% for Duster and from 89.2 to 28.9% for Ruby Lee. Additionally, low soil pH delayed the achievement of maximum rate of canopy cover development (t0). For example, in Stillwater 2012–2013, t0 in the variety Duster increased from 25.6 DAS at soil pH > 7.0 to 53.2 DAS in the pH range from 4–4.5 (Fig. 4a), whereas for Ruby Lee t0 increased from 25.9 to 55.4 DAS (Fig. 4b). Similar patterns of canopy cover as affected by soil pH occurred for all varieties in all site-years (Supplemental Table S2). A minimum percent canopy cover of 53% prior to winter dormancy and 62% at grazing termination is needed to maximize yields in dual-purpose wheat systems (Butchee and Edwards, 2013), and in our study, a minimum soil pH of 5 was needed for percent canopy coverage to reach the aforementioned thresholds (Supplemental Table S2). Critical Soil pH and Aluminum Concentration for Wheat Forage Yield

The exponential rise to maximum model explained the relationship between forage yield and soil pH across years very well (Fig. 5). According to the fitted equations, 95% of the asymptotic forage biomass would be achieved at pH of 5.5 for the variety 2174, at pH of 5.9 for both Duster and Ruby Lee, and at pH of 6 for TAM 203. The lower pH threshold for maximum forage yield by 2174 indicates tolerance to acid soil conditions. The β1 parameter was greater for the variety 2174 (2.15 ± 0.49), indicating a steep increase in forage production at pH values ranging from 4 to the 5.5 threshold for asymptotic yield. Interestingly, the fitted equations for Duster and Ruby Lee resulted in similar β1 values (1.64 ± 0.42 and 1.69 ± 0.37, respectively), indicating similar increase in forage production from pH of about 4 to the 8

Fig. 4. Dynamics of canopy cover development during the growing season as function of days after sowing for (a) the wheat variety Duster and (b) Ruby Lee grown in soil pH > 7 versus soil pH in the 4.0 to 4.5 range at Stillwater, OK, during the 2012–2013 growing season.

5.9 pH threshold. The most sensitive wheat variety to low soil pH was TAM 203, with β1 of 1.57 ± 0.45. Finally, the pH below which there was no forage yield, assessed as the x-intercept in Fig. 5, was 4.1 for the four studied wheat varieties, indicating that in extremely acidic soils, cultivar selection is not an option to overcome low soil pH. Similarly, Johnson et al. (1997) reported no measurable early-season forage production by wheat varieties sensitive to acidic soils when cultivated in sites with soil


Fig. 5. Forage yield prior to winter dormancy of wheat varieties (a) 2174, (b) Ruby Lee, (c) Duster, and (d) TAM 203, as affected by soil pH at Stillwater, OK, during the 2012–2013, 2013–2014, and 2014–2015 growing seasons.

pH of 4.5 and Alsat > 30%. Likewise, for each 1% increase in Alsat, wheat forage yield decreased linearly at 92 ± 12.5 kg ha–1 for 2174 (r 2 = 0.51, p < 0.001), 93.0 ± 12.8 kg ha–1 for Duster (r 2 = 0.51, p < 0.001), 96.4 ± 11.2 kg ha–1 for Ruby Lee (r 2 = 0.59, p < 0.001), and 89.2 ± 12.3 kg ha–1 for TAM 203 (r 2 = 0.50, p < 0.001). A negative linear association between wheat forage yields and Alsat has been previously reported (Kariuki et al., 2007), and occurs because Al3+ restrains root development (Zhou et al., 2007), consequently decreasing forage yield. Critical Soil pH And Aluminum Concentration for Wheat Grain Yield and Quality

The linear-plateau model explained each variety’s grain yield sensitivity to soil pH well in 2012–2013 and 2014–2015 at Stillwater, and the yield of the most sensitive varieties to soil pH in Chickasha during 2013–2014 and 2014–2015 (Table 3). Soil pH did not affect grain yield in Stillwater during the 2013–2014 growing season when a severe spring drought drastically reduced wheat yields; or at Chickasha (2012–2013) when soil amendments were not sufficient to create a large pH gradient. Threshold pH (γ) for grain yield plateau of sensitive varieties ranged from 4.3 to 5 in Stillwater, and from 5.1 to 5.2 in Chickasha (Table 3). Minimum pH threshold below which grain yields decrease had not been studied for TAM 203,

Duster, or Ruby Lee. In most cases when wheat yield showed a linear-plateau relationship to soil pH, wheat yield reduction was accentuated in the lowest pH experimental units (soil pH < 4.5) and was not as apparent in the remaining plots in which soil pH ranged from 4.5 to 7.5 (Fig. 6a). As a result, the linearplateau models had steep slopes in the linear portion, and γ at low pH values. The lack of wheat yield response to soil pH > 4.5 in our study agrees with literature evaluating acidity amelioration strategies in which liming acid soils did not always result in increased wheat yields at pH levels ~4.7 (Liu et al., 2004; Lollato et al., 2013). The extremely low soil pH needed to induce grain yield reductions in our study partially explains the lack of response of the more tolerant varieties Duster and 2174 at Chickasha (Table 3), which had soil pH > 4.5 and Alsat < 8%. Figure 6a illustrates the difference in sensitivity to soil pH of the varieties Duster and TAM 203 at Stillwater during the 2014–2015 growing season, in which the variety Duster resulted in γ of 4.5 as compared to 4.9 for TAM 203, indicating a greater tolerance to low soil pH of the former. Duster had a yield advantage at soil pH < 4.5, which ranged from 17 to 65% when compared to Ruby Lee, 14 to 117% when compared to TAM 203, and 24 to 58% when compared to 2174. Duster has been shown to be superior to other wheat varieties in low soil pH conditions (Edwards et al., 2012) due to the root-tip staining


9

Table 3. Intercept (β0), slope (β1), threshold pH or aluminum saturation ( ± standard error) beyond which increases in independent variable did not result in increased response (γ), plateau (Plat.) following γ, and regression significance for the linear-plateau model in Eq. [1] describing wheat grain yield as function of soil pH or Alsat for the varieties 2174, Duster, Ruby Lee, and TAM 203 at Stillwater and Chickasha, OK, during the 2012–2013, 2013–2014, and 2014–2015 growing seasons. Soil pH Alsat Growing Location season Variety β0 β1 γ Plat. r2 β0 β1 γ Plat. r2 –1 –1 –1 –1 –1 –1 –1 –1 –1 Mg ha Mg ha pH pH Mg ha Mg ha Mg ha cmolc kg cmolc kg Mg ha Stillwater 2012–2013 2174 –31.6 8.37 4.3 ± 0.2 4.63 0.64*** 13.48 –45.24 0.20 ± 0.02 4.63 0.68*** Duster –7.29 2.59 4.5 ± 0.1 4.29 0.67*** 6.56 –13.59 0.17 ± 0.02 4.29 0.70*** Ruby Lee –20.65 5.75 4.4 ± 0.1 4.89 0.64*** 9.77 –28.77 0.17 ± 0.02 4.88 0.69*** TAM 203 –8.15 2.71 4.5 ± 0.1 4.15 0.72*** 5.01 –8.25 0.10 ± 0.03 4.21 0.76*** 2013–2014 2174 –3.1 1.08 4.5 ± 0.2 1.74 0.04ns† 2.41 –3.00 0.23 ± 1.00 1.72 0 ns Duster –20.26 0.95 4.5 ± 1.0 2.25 0.02ns 2.28 –0.36 0.12 ± 0.78 2.23 0.01 ns Ruby Lee 0.49 0.32 5.4 ± 0.9 2.21 0.14ns 2.79 –5.35 0.12 ± 0.06 2.15 0.21 ns TAM 203 0.45 0.28 5.6 ± 1.0 2.03 0.14ns 2.46 –4.35 0.12 ± 0.08 1.94 0.12 ns 2014–2015 2174 –11.27 2.99 4.9 ± 0.2 3.30 0.54** 4.69 –19.67 0.06 ± 0.08 3.47 0.66*** Duster –44.45 10.62 4.5 ± 0.1 3.40 0.72*** 4.36 –27.90 0.04 ± 0.01 3.37 0.57*** Ruby Lee –59.54 14.15 4.5 ± 0.1 4.43 0.86*** 5.77 –39.60 0.04 ± 0.01 4.40 0.66*** TAM 203 –21.75 5.22 4.9 ± 0.1 3.74 0.87*** 3.92 –29.88 –‡ – 0.81*** Chickasha 2012–2013 2174 –0.626 0.84 6.1 ± 0.6 4.50 0.18ns na§ na na na 0 ns Duster –2.561 1.25 5.9 ± 0.4 4.86 0.15ns 4.73 –50.41 – – 0 ns Ruby Lee 0.390 0.73 6.1 ± 0.7 4.83 0.16ns na na na na 0 ns TAM 203 0.018 0.75 5.3 ± 1.0 3.77 0 ns na na na na 0 ns 2013–2014 2174 –1.05 0.59 5.2 ± 0.6 2.00 0.23ns 2.61 –14.01 0.05 ± 0.05 1.94 0.12 ns Duster –0.83 0.52 5.2 ± 0.9 1.87 0.14ns 3.21 –20.00 0.07 ± 0.02 1.78 0.03 ns Ruby Lee –1.22 0.50 7.6 ± 2.0 1.56 0.27ns 2.91 –20.00 0.08 ± 0.02 1.36 0.15 ns TAM 203 –2.5 0.80 5.4 ± 0.4 1.63 0.28ns 3.35 –29.86 0.06 ± 0.02 1.49 0.14 ns 2014–2015 2174 –6.93 1.99 5.1 ± 0.1 3.24 0.27ns 3.25 –10.19 – – 0.13 ns Duster –3.64 1.45 5.1 ± 0.2 3.74 0.14ns 3.74 –5.04 – – 0.03 ns Ruby Lee –17.87 4.16 5.2 ± 0.1 3.66 0.59** 3.67 –28.38 – – 0.4** TAM 203 –21.94 5.05 5.1 ± 0.1 3.99 0.81*** 4.08 –4.08 – – 0.61*** **,*** Significant at p < 0.01, and 0.001, respectively. † ns, nonsignificant. ‡ Plateau not reached within the range in Al values studied. Linear model explained the relationship between wheat yield and aluminum saturation. Coefficients refer to the linear portion of the model. § Model failed to converge due to small range in aluminum saturation values in 2012–2013 at Chickasha, OK.

Table 4. Factor significances and mean separation for the analysis of covariance for grain volume weight and grain protein concentration as function of wheat variety, soil aluminum saturation (Alsat) and their interaction during the growing seasons 2012–2013, 2013–2014, and 2014–2015 at Stillwater and Chickasha, OK. Variety means are averaged across Alsat values because the latter was not significant with the exception of Chickasha 2013–2014. Stillwater Chickasha Variety 2012–2013 2013–2014 2014–2015 2012–2013 2013–2014 2014–2015 —————————————————— Grain volume weight (kg m–3) —————————————————— 2174 679 b† 708 a 700 750 a 695 a 671 Duster 697 a 666 b 687 725 b 698 a 697 Ruby Lee 702 a 689 a 688 749 a 697 a 695 TAM 203 659 c 647 b 698 699 c 669 b 701 Source of variation Variety *** *** ns‡ *** *** ns ns ns ns ns ns ns Alsat ns ns ns ns ns ns Variety × Alsat ———————————————— Grain protein concentration (g kg–1) ————————————————— 2174 148 b 191a 138 a 160 b 170 ab 139 a Duster 137 c 171 c 133 ab 150 c 164 b 133 b Ruby Lee 144 b 183 b 125 b 157 bc 175 a 133 b TAM 203 154 a 186 ab 137 a 169 a 174 a 139 a Source of variation Variety *** *** * *** *** *** ns ns ns ns *** ns Alsat ns ns ns ns ns * Variety × Alsat *, *** Significant at p < 0.05 and 0.001, respectively. † Values followed by the same letter within site-year are not statistically different at α = 0.05 based on Tukey’s Honest Significant Difference test. ‡ ns, nonsignificant.

10


Fig. 6. Grain yield of the wheat varieties (a) Duster and TAM 203 as affected by soil pH during the 2014–2015 growing season in Stillwater, OK; and (B) 2174 and Ruby Lee as affected by soil aluminum saturation (Alsat) during the 2012–2013 growing season in Stillwater, OK.

pattern (Heyne and Niblett, 1978) and a functional allele of the Al-induced malate transporter gene (Zhou et al., 2007). At soil pH of 7 or above, Ruby Lee had a yield advantage ranging from 4 to 46% over the other varieties. Analysis of relative yields using the exponential rise to maximum model resulted in more consistent parameters as data were analyzed across years (Fig. 7). Maximum attainable relative yield (e.g., 95% of the asymptotic) was 0.92 for 2174, 0.94 for Duster and Ruby Lee, and 0.96 for TAM 203. These relative yields were achieved at soil pH of 4.8, 4.8, 5.8, and 5.5, respectively, confirming results from individual years and indicating that 2174 and Duster have greater tolerance to acid soil conditions relative to Ruby Lee and TAM 203. The threshold pH values obtained in our study are lower than those reported by Kariuki et al. (2007), which ranged from 5.3 to 6.6 for different wheat varieties. The differences between studies may be attributed to soil chemical characteristics, as Kariuki et al. (2007) tested wheat tolerance to acidic soils in a Konawa fine loamy soil with Alsat values as great as 70%, which indicates a greater toxicity potential (Johnson et al., 1997). Site-specific symptoms of soil acidity on wheat yield of the same variety tested at locations where soil differed in chemical properties were also reported by Johnson et al. (1997). Additionally, all pH thresholds in our study were between 4.8

and 5.8, whereas the threshold pH range previously reported for other wheat varieties was 5.3 to 6.6 (Kariuki et al., 2007). These differences may be attributed not only to differences in soil chemistry but to the modern wheat varieties, such as Duster, released from a breeding program characterized by efforts to increase Al tolerance in wheat varieties (Johnson et al., 1997; Tang et al., 2002; Zhou et al., 2007). We provided empirical evidence for variety-specific wheat yield unaffected by soil pH > 4.8 to 5.8 at soils with ECEC ranging between 6.3 and 14.8 cmolc kg–1. Wheat grain yield was modeled as a plateau-linear function of soil Alsat at Stillwater during the 2012–2013 and 2014–2015 growing seasons, and at Chickasha for the more sensitive varieties during 2013–2014 and 2014–2015 growing seasons (Table 3). The linear decrease in grain yield as function of Alsat previously reported (Kariuki et al., 2007; Schroder et al., 2011; Valle et al., 2009) occurred for the varieties Ruby Lee and TAM 203 at Stillwater 2014–2015 and at Chickasha 2013–2014 and 2014–2015 (Table 3). In all other cases, grain yields plateaued at low Alsat levels until a minimum threshold Alsat was reached beyond which an increase in Alsat reduced grain yields. In Stillwater, threshold Alsat averaged 11.7% for the variety 2174, 9.6% for Duster, 8.7% for Ruby Lee, and 7.1% for TAM 203 (Table 3). Higher threshold Alsat indicates that the variety can endure greater levels of Al toxicity without the associated grain yield penalty. In Stillwater 2012–2013, when growing season precipitation during spring was plentiful, the yield of the variety 2174 plateaued up to Alsat values of 17%, whereas Ruby Lee plateaued up to Alsat of 15% (Fig. 6b), exemplifying the strong environmental effect on wheat sensitivity to acidic soils previously described, in which greater moisture availability might partially mitigate the effects of acidic soils (Johnson et al., 1997; Lollato et al., 2013). Analysis of covariance for wheat grain volume weight and grain protein concentration as function of variety and Alsat indicated that, in most site-years, grain quality parameters were affected by wheat variety but not by Alsat (Table 4). Grain volume weight was unaffected by Alsat in all studied site-years, and grain protein concentration increased with increased Alsat in Chickasha 2013–2014 but was otherwise unaffected. While Ruby Lee consistently resulted in greater or similar grain volume weight when compared to the other varieties, TAM 203 resulted in greater or similar protein concentration (Table 4). Our results demonstrate that wheat forage yield is more sensitive to acidic soil conditions than grain yield, as evidenced by greater threshold pH for forage yields to plateau (i.e., 5.5 to 6.0) as compared to grain yields (i.e., 4.8 to 5.8). The greater sensitivity of wheat forage yields as compared to grain yields agrees well with existing literature (Johnson et al., 1997; Kaitibie et al., 2002; Kariuki et al., 2007; Lollato et al., 2013) and may be due to forage yield formation occurring prior to full root development, whereas grain yield is formed after a more extensive rooting system is achieved (Kariuki et al., 2007). Many Oklahoma soils are only acidic on the surface (Schroder et al., 2011), which would allow for a more developed rooting system later in the season to explore deeper soil players and buffer grain yield formation if there is favorable weather during spring, when grain yield determination occurs (Lollato and Edwards, 2015).


11

Fig. 7. Relative grain yield of the wheat varieties (a) 2174, (b) Ruby Lee, (c) Duster, and (d) TAM 203, as affected by soil pH during the 2012–2013, 2013–2014, and 2014–2015 growing seasons at Stillwater and Chickasha, OK.

Upscaling Results from Field Experiments to State Level Similar to our field study, analysis of the database of soil pH values suggested that dual-purpose wheat systems are more often limited by acidic soils than grain-only systems in the study region. A total of 4135 samples (or 35% of total samples) had pH below 6.0 and 28% (or 3307 samples) had pH below 5.8, which were the thresholds developed for pH sensitive varieties under dual-purpose and grain-only systems, respectively (Fig. 1b). These estimates appear to be an improvement over a similar analysis performed during 1995 and 1996, in which 39% of the soil samples analyzed in Oklahoma had soil pH less than 5.5 (Zhang et al., 1998). Average estimated grain yield for grain-only systems in the 3307 soil pH samples below 5.8, weighted by the harvested area by county, was statistically greater for a pH tolerant variety as compared to a pH sensitive (2.09 ± 0.04 vs. 1.93 ± 0.08 Mg ha–1, p < 0.001), with yield differences between tolerant and sensitive varieties consistently positive (Fig. 8a). As expected, this advantage decreased across the entire dataset (n = 11,905), in which tolerant and sensitive varieties yielded 2.13 ± 0.03 and 2.05 ± 0.08 Mg ha–1, but was still statistically significant (Fig. 8c). For the grain-only system across all pH samples evaluated, estimated county yield ranged from 1.72 to 2.48 Mg ha–1 for tolerant 12

varieties, and from 1.69 to 2.46 Mg ha–1 for sensitive varieties (Table 5). Liming to a pH of 6.5 instead of selecting a tolerant variety would have resulted in a weighted grain yield of 2.16 Mg ha–1 (Table 5). For the 4135 samples with soil pH less than 6.0 used in our dual-purpose analysis, weighted average grain yield was 1.81 ± 0.03 and 1.69 ± 0.07 Mg ha–1 for tolerant and sensitive varieties (p < 0.001), and weighted average forage yield was 2.49 ± 0.18 and 2.33 ± 0.22 Mg ha–1 for pH tolerant and sensitive varieties (p < 0.001, Fig. 8b). County grain yield ranged from 1.5 to 2.16 Mg ha–1 for tolerant varieties and from 1.47 to 2.14 Mg ha–1 for sensitive varieties, and forage yield ranged from 2.46 to 2.68 Mg ha–1 for tolerant varieties and from 2.32 to 2.64 Mg ha–1 for sensitive varieties (Table 5). Weighted yield based on liming instead of selecting a tolerant variety was 1.88 Mg ha–1 for grain and 2.6 Mg ha–1 for forage (Table 5). Revenue analysis of liming versus selecting a tolerant variety showed that the best strategy to maximize net return was field specific and dependent on lime requirements by the soil. Across all counties, revenue increase from switching from a pH sensitive to a tolerant variety in the grain-only system ranged from $2 to $20 ha–1, with a weighted average of $10 ha–1 for the study region (Table 5). Meanwhile, liming led to revenue increases of $4 to $27 ha–1, averaging $19 ha–1 (Table 5). For the dual-purpose system, increases in revenue due to either


Fig. 8. Histogram of grain and forage yields for tolerant and sensitive varieties at (a) the soil samples with pH below 5.8 and limiting to grainonly production, (b) the soil samples with pH below 6.0 and limiting to dual-purpose production, (c) for grain-only production across the entire dataset, and (d) for dual-purpose production across the entire dataset. Inset graphs show cumulative probability of grain yield in the tolerant variety being greater than the grain yield in the sensitive variety. Inset tables show mean and standard deviation grain yield and forage yield.

liming or selecting a variety tolerant to low soil pH were greater, mostly led by the increased forage production. Variety selection led to revenue increase of $10 to $47 ha–1, averaging $28 ha–1; and liming increased revenue anywhere between $17 and $64 ha–1, averaging $37 ha–1 (Table 5). The decision to lime or select a wheat variety with improved tolerance to low soil pH depended on county studied, as in some cases liming was more profitable and in other cases, variety selection was more profitable (Table 5). Usually, soils with high buffer index which required less lime to reach a pH of 6.5 led to greater profitability from liming, whereas soils with low buffer index most often led to greater profitability from variety selection (data not shown). To our knowledge, this is the first attempt to upscale field experiments examining variety response to soil pH to a regional perspective. Ideally, upscaling methods should cover a minimum of 50% of the area planted to a particular crop in a growing region, and weight location-specific results based on

harvested area (van Bussel et al., 2015); requirements met in our method. Upscaling methods like the one we presented have the potential to affect prioritization of funds for research and development, as well as to aid in the development of agricultural policies. For instance, the potential additional wheat production based on either liming or selecting a pH tolerant variety for a given region could be calculated based on the specific yield response of the different varieties coupled with information regarding soil pH distribution, variety adoption, and harvested wheat area by county. To expand on this example, 18.8% of the reported wheat varieties adopted in Oklahoma during the 2016 season were pH sensitive varieties, and an additional 24.9% of the surveyed area had unknown variety (USDA-NASS, 2016), also potentially sensitive to low soil pH. In a conservative estimate (e.g., considering 18.8% of the area was sown to sensitive varieties), switching to a pH tolerant variety would lead to a yield gain of 0.07 Mg ha–1 and an increase in 23,200 Mg or


13

Table 5. Winter wheat harvested area (Harea), current yield (Yc), yield simulated based on lime application for dual-purpose (Ydp), and grain-only (Ygo), grain yield for tolerant (Ytol) and sensitive (Ysen) varieties under grain-only and dual-purpose systems, and forage yield for tolerant (Ftol) and sensitive (Fsen) varieties. Increased revenue per hectare is also shown for variety selection or liming at both grainonly and dual-purpose systems. Means (bottom row) are weighted based on harvested area. Variety selection Revenue Lime application Grain-only Dual-purpose Forage Grain-only Dual-purpose County Harea Yc Ygo Ytol Ysen Ytol Ysen Ftol Fsen Variety Liming Variety Liming Ydp 1000 ha —————————————— Mg ha–1 —————————————— ————— $ ha–1 ————— Alfalfa 91 2.24 2.09 2.40 2.36 2.27 2.05 1.98 2.61 2.51 12 22 31 39 Beaver 50 1.70 1.58 1.82 1.80 1.78 1.56 1.55 2.67 2.63 2 25 10 18 Beckham 23 1.81 1.69 1.94 1.91 1.89 1.66 1.64 2.66 2.61 4 27 14 24 Blaine 67 1.97 1.83 2.11 2.08 2.01 1.80 1.75 2.62 2.53 9 20 25 32 Caddo 65 2.17 2.01 2.32 2.27 2.17 1.97 1.88 2.56 2.44 15 22 36 51 Canadian 65 2.09 1.95 2.24 2.20 2.12 1.91 1.84 2.60 2.50 12 19 30 37 Comanche 23 1.78 1.66 1.91 1.87 1.81 1.63 1.58 2.60 2.51 8 19 25 35 Cotton 38 1.84 1.71 1.97 1.93 1.85 1.68 1.61 2.56 2.45 11 16 32 43 Custer 71 2.13 1.98 2.28 2.24 2.19 1.95 1.90 2.62 2.54 8 25 22 34 Dewey 41 1.93 1.79 2.06 2.03 1.99 1.77 1.73 2.64 2.57 6 25 20 29 Ellis 23 1.64 1.53 1.76 1.73 1.69 1.51 1.47 2.64 2.56 6 22 21 28 Garfield 126 2.17 2.02 2.33 2.26 2.12 1.97 1.84 2.46 2.32 20 12 47 64 Grady 25 2.04 1.90 2.18 2.14 2.06 1.86 1.79 2.57 2.47 11 22 30 45 Grant 123 2.07 1.92 2.21 2.15 2.02 1.87 1.76 2.48 2.34 18 7 44 56 Greer 28 1.72 1.60 1.84 1.81 1.75 1.57 1.52 2.60 2.51 8 22 25 39 Harmon 18 1.63 1.52 1.74 1.72 1.71 1.50 1.48 2.67 2.63 2 26 11 19 Harper 32 1.73 1.61 1.85 1.83 1.81 1.59 1.58 2.68 2.63 2 25 10 17 Jackson 63 1.88 1.75 2.01 1.98 1.94 1.72 1.68 2.64 2.57 6 23 19 28 Kay 81 2.10 1.95 2.25 2.20 2.08 1.91 1.81 2.55 2.42 16 4 41 38 Kingfisher 77 2.04 1.90 2.18 2.14 2.06 1.86 1.79 2.58 2.48 11 22 30 42 Kiowa 85 1.92 1.79 2.05 2.02 1.95 1.75 1.70 2.60 2.50 10 17 27 34 Logan 24 1.95 1.81 2.08 2.05 1.95 1.78 1.70 2.58 2.47 13 14 35 38 Major 48 1.96 1.82 2.09 2.06 1.99 1.79 1.73 2.61 2.53 9 24 25 37 Noble 51 2.03 1.88 2.17 2.12 2.00 1.84 1.74 2.52 2.38 16 7 40 47 Roger Mills 14 1.90 1.77 2.03 2.00 1.96 1.74 1.71 2.64 2.57 6 27 18 31 Texas 83 2.35 2.18 2.51 2.48 2.46 2.16 2.14 2.68 2.64 2 27 10 18 Tillman 55 1.94 1.80 2.07 2.04 1.98 1.77 1.72 2.61 2.53 8 24 23 36 Washita 80 2.03 1.88 2.17 2.14 2.09 1.86 1.82 2.65 2.58 6 24 18 27 Woods 73 2.13 1.98 2.28 2.25 2.20 1.95 1.91 2.65 2.57 7 22 20 25 Woodward 29 1.86 1.73 1.99 1.96 1.94 1.71 1.68 2.67 2.61 4 25 14 21 Mean – 2.02 1.88 2.16 2.13 2.05 1.85 1.78 2.59 2.50 10 19 28 37

0.9% in regional production. Likewise, liming would lead to a yield gain of 0.11 Mg ha–1 and an additional production of 35,500 Mg or 1.3% at state level. Assuming that the remainder 24.9% unknown area was sown to a sensitive variety, additional production in the state would total 53,800 Mg (2%) and 82,500 Mg (3%) for variety selection and liming. Similar upscaling of potential additional production based on adoption of particular management practices was explored by Rattalino Edreira et al. (2017) for soybeans in the US Midwest; however, to our knowledge, this is the first quantification of wheat forage and grain yield based on liming or variety selection to overcome acidic soils. While our analysis focused on estimating exploitable yield gaps caused by acidic soils, many other suboptimal management factors contribute to the large wheat yield gap in this region, which ranges from 1.3 to 4.3 Mg ha–1 (Lollato et al., 2017). One factor potentially contributing to the yield gap that could also be explored using a similar procedure includes dual-purpose systems, in which the yield penalty averages about 14% in controlled experiments (Edwards et al., 2011) but no regional upscaling has identified its real impact on statewide production. Similarly, lack of crop rotation reduced 14

wheat yields in 10 to 22% in controlled experiments (Bushong et al., 2012), but no attempt to upscale these results to regional impacts have been performed. Other factors leading to the large yield gap include risk-aversion from producers leading to low investments in the crop due to a high year-to-year weather variability (Lollato et al., 2017), poor soil quality due to predominance of conventional tillage practices leading to soil erosion (Patrignani et al., 2014; Lollato et al., 2012), etc. While our goal was not to untangle the contribution of each factor to the wheat yield gap in the region, similar protocol can be developed to estimate their individual contribution. Uncertainties in our protocol include the assumptions that (i) soil databases derived from commercial laboratories are representative of the majority of the studied region (e.g., they might overestimate soil pH problems if most sampled fields were problematic, showing symptoms of nutrient deficiency or low soil pH; on the other hand, they may underestimate acidic soil problems if samples derived mostly from progressive growers who routinely analyze their fields to take informed decisions referring to nutrient management); (ii) the varieties studied and responses obtained in our field experiments represent those to


be obtained in commercial production fields (while the variability in environmental conditions experienced in our field study provides a strong argument for its representativeness, it is an assumption nonetheless); and (iii) government estimates for regional variety adoption (e.g., the USDA-NASS, 2016) are accurate. Additionally, our protocol was entirely based on soil pH and did not account for Al saturation or soil base cation saturation because the available database provided by SWFAL did not include these information for samples routinely analyzed. This information should be available for regions in which liming recommendations are based on base cation saturation (Raij et al., 1985), which could potentially improve our approach to account for this or for any other extra layer of information. Nevertheless, our analysis is innovative and demonstrates the potential uses of this protocol as a tool for impact assessments at regional levels, and can be adopted in other growing regions of the world characterized by acidic conditions to explore the potential response of different crops. It can aid in determining response curves and probabilities of yield gain at a regional level, as well as in establishing research priorities and agricultural policies, such as development of varieties tolerant to acidic soils or facilitation of soil amendment importation or transportation. ACKNOWLEDGMENTS We thank the contribution of the anonymous reviewers of Agronomy Journal in improving our manuscript. We thank the agronomists Robert Calhoun and Matt Knori for their support with field operations, and Giovana Cruppe, Luciano Cebogias Jr., and Victor Ricardo Bodnar for their help with field measurements. The field portion of this study was partially supported by the Oklahoma Cooperative Extension Service, and the upscaling portion of the study by the Kansas Agricultural Experiment Station and the Kansas Cooperative Extension Service. This research is contribution no. 18-520-J from the Kansas Agricultural Experiment Station. REFERENCES Allen, R.G., L.S. Pereira, D. Raes, and M. Smith. 1998. Crop evapotranspiration: Guidelines for computing crop water requirements. Irrigation and Drainage Paper 56. FAO, Rome, Italy. Archontoulis, S.V., and F.E. Miguez. 2015. Nonlinear regression models and applications in agricultural research. Agron. J. 107:786–798. doi:10.2134/agronj2012.0506 Behera, S.K., and A.K. Shukla. 2015. Spatial distribution of surface soil acidity, electrical conductivity, soil organic carbon content and exchangeable potassium, calcium and magnesium in some cropped acid soils of India. Land Degrad. Dev. 26:71–79. doi:10.1002/ ldr.2306 Bolt, V. 1996. Tolerance of wheat cultivars to soil acidity. Farmnote 2/96. Western Australia Dep. of Agric., South Perth, Australia. Bushong, J.A., A.P. Griffith, T.F. Peeper, and F.M. Epplin. 2012. Continuous winter wheat versus a winter canola–winter wheat rotation. Agron. J. 104:324–330. doi:10.2134/agronj2011.0244 Butchee, J.D., and J.T. Edwards. 2013. Dual-purpose wheat grain yield as affected by growth habit and simulated grazing intensity. Crop Sci. 53:1686–1692. doi:10.2135/cropsci2013.01.0033 Butchee, K., D.B. Arnall, A. Sutradhar, C. Godsey, H. Zhang, and C. Penn. 2012. Determining critical soil ph for grain sorghum production. Int. J. Agron. 2012:130254. doi:10.1155/2012/130254

Caires, E.F., F.J. Garbuio, S. Churka, G. Barth, and J.C.L. Corrêa. 2008. Effects of soil acidity amelioration by surface liming on no-till corn, soybeam and wheat root growth and yield. Eur. J. Agron. 28:57–64. doi:10.1016/j.eja.2007.05.002 Carver, B.F., E.L. Smith, E.G. Krenzer, R.M. Hunger, D.R. Porter, et al. 2003. Registration of ‘OK101’ wheat. Crop Sci. 43:2298–2299. doi:10.2135/cropsci2004.1474 Cerrato, M.E., and A.M. Blackmer. 1990. Comparison of models for describing corn yield response to nitrogen-fertilizer. Agron. J. 82:138–143. doi:10.2134/agronj1990.00021962008200010030x de Lima, M.L., and L. Copeland. 1990. The effect of aluminum on the germination of wheat seeds. J. Plant Nutr. 13:1489–1497. doi:10.1080/01904169009364170 de Sousa, C.N.A. 1998. Classification of Brazilian wheat cultivars for aluminum toxicity in acid soils. Plant Breed. 117:217–221. doi:10.1111/j.1439-0523.1998.tb01929.x Doye, D., D. Marburger, and R. Sahs. 2017. Should I buy (or retain) stockers to graze wheat pasture? Okla. St. Univ. Coop. and Ext. Serv. CR-212. Okla. St. Univ., Stillwater. Rattalino Edreira, J.I., S. Mourtzinis, S.P. Conley, A.C. Roth, I.A. Ciampitti, et al. 2017. Assessing causes of yield gaps in agricultural areas with diversity in climate and soils. Agric. For. Meteorol. 247:170–180. doi:10.1016/j.agrformet.2017.07.010 Edwards, J.T. 2013. Ruby lee: endurance parentage, top-end yield potential, and hessian fly resistance. Okla. St. Univ. Coop. and Ext. Serv. L-414. Okla. St. Univ., Stillwater. Edwards, J.T., B.F. Carver, G.W. Horn, and M.E. Payton. 2011. Impact of dual-purpose management on wheat grain yield. Crop Sci. 51:2181– 2185. doi:10.2135/cropsci2011.01.0043 Edwards, J.T., R.M. Hunger, E.L. Smith, G.W. Horn, M.S. Chen, et al. 2012. ‘Duster’ wheat: A durable, dual-purpose cultivar adapted to the southern great plains of the USA. J. Plant Reg. 6:37–48. doi:10.3198/jpr2011.04.0195crc Edwards, J.T., and L.C. Purcell. 2005. Soybean yield and biomass responses to increasing plant population among diverse maturity groups: I. Agronomic characteristics. Crop Sci. 45:1770–1777. doi:10.2135/ cropsci2004.0564 Fisher, J.A., and B.J. Scott. 1993. Are we justified in breeding wheat for tolerance to acid soils in southern New South Wales? In: P.J. Randall, E. Delhaize, R.A. Richards, and R. Munns, editors, Genetic aspects of plant mineral nutrition. Springer Science+Business Media, B.V., Canberra, Australia. doi:10.1007/978-94-011-1650-3_1 Gascho, G.J., and M.B. Parker. 2001. Long-term liming effects on Coastal Plain soils and crops. Agron. J. 93:1305–1315. doi:10.2134/ agronj2001.1305 Guo, J.H., X.J. Liu, Y. Zhang, J.L. Shen, W.X. Han, W.F. Zhang, P. Christie, K.W.T. Goulding, P.M. Vitousek, and F.S. Zhang. 2010. Significant acidification in major Chinese croplands. Science 327(5968):1008–1010. doi:10.1126/science.1182570 Heyne, E., and C. Niblett. 1978. Registration of ‘Newton’ wheat. Crop Sci. 18:696. doi:10.2135/cropsci1978.0011183X001800040055x Jamal, S.N., M.Z. Iqbal, and M. Athar. 2006. Phytotoxic effect of aluminum and chromium on the germination and early growth of wheat (Triticum aestivum) varieties anmol and kiran. Int. J. Environ. Sci. Technol. 3:411–416. doi:10.1007/BF03325950 Johnson, J.P., B.F. Carver, and V.C. Baligar. 1997. Productivity in great plains acid soils of wheat genotypes selected for aluminium tolerance. Plant Soil 188:101–106. doi:10.1023/A:1004268325067 Kaitibie, S., F.M. Epplin, E.G. Krenzer, and H.L. Zhang. 2002. Economics of lime and phosphorus application for dual-purpose winter wheat production in low-ph soils. Agron. J. 94:1139–1145. doi:10.2134/ agronj2002.1139


15

Kariuki, S.K., H. Zhang, J.L. Schroder, J. Edwards, M. Payton, et al. 2007. Hard red winter wheat cultivar responses to a ph and aluminum concentration gradient. Agron. J. 99:88–98. doi:10.2134/ agronj2006.0128 Large, E.C. 1954. Growth stages in cereals illustration of the feekes scale. Plant Pathol. 3:128–129. doi:10.1111/j.1365-3059.1954.tb00716.x Liu, D.L., K.R. Helyar, M.K. Conyers, R. Fisher, and G.J. Poile. 2004. Response of wheat, triticale and barley to lime application in semi-arid soils. Field Crops Res. 90:287–301. doi:10.1016/j.fcr.2004.03.008 Lollato, R.P., M.A. Lollato, and J.T. Edwards. 2012. Soil organic carbon replenishment through long-term no-till on a Brazilian family farm. J. Soil Water Conserv. 67(3):74A–76A. doi:10.2489/jswc.67.3.74A Lollato, R.P., and J.T. Edwards. 2015. Maximum attainable wheat yield and resource use efficiency in the southern great plains. Crop Sci. 55:2863–2876. doi:10.2135/cropsci2015.04.0215 Lollato, R.P., J.T. Edwards, and H.L. Zhang. 2013. Effect of alternative soil acidity amelioration strategies on soil ph distribution and wheat agronomic response. Soil Sci. Soc. Am. J. 77:1831–1841. doi:10.2136/sssaj2013.04.0129 Lollato, R.P., J.T. Edwards, and T.E. Ochsner. 2017. Meteorological limits to winter wheat productivity in the U.S. southern Great Plains. Field Crops Res. 203:212–226. doi:10.1016/j.fcr.2016.12.014 Mahler, R.L., A.R. Halvorson, and F.E. Koehler. 1985. Longterm acidification of farmland in northern Idaho and eastern Washington. Commun. Soil Sci. Plant Anal. 16:83–95. doi:10.1080/00103628509367589 Manske, G.G.B., J.I. Ortiz-Monasterio, M. van Ginkel, R.M. Gonzalez, R.A. Fisher, S. Rajaram, and P.L.G. Vlek. 2001. Importance of P uptake efficiency versus P utilization for wheat yield in acid and calcareous soils in Mexico. Eur. J. Agron. 14:261–274. doi:10.1016/ S1161-0301(00)00099-X Meriño-Gergichevich, C., M. Alberdi, A.G. Ivanov, and M. Reyes-Diaz. 2010. Al3+- Ca2+ interaction in plants growing in acid soils: Alphytotoxicity response to calcareous amendments. J. Soil Sci. Plant Nutr. 10:217–243. Panagos, P., M. van Liedekerke, A. Jones, and L. Montanarella. 2012. European soil data center: Response to European policy support and public data requirements. Land Use Policy 29:329–338. doi:10.1016/j. landusepol.2011.07.003 Patrignani, A., R.P. Lollato, T.E. Ochsner, C.B. Godsey, and J.T. Edwards. 2014. Yield gap and production gap of rainfed winter wheat in the southern Great Plains. Agron. J. 106:1329–1339. doi:10.2134/ agronj14.0011 Patrignani, A., and T.E. Ochsner. 2015. Canopeo: A powerful new tool for measuring fractional green canopy cover. Agron. J. 107:2312–2320. doi:10.2134/agronj15.0150 Raij, B.V., H. Cantarella, J.A. Quaggio, and A.M.C. Furlani. 1985. Recomendações de adubação e calagem para o Estado de São Paulo. Campinas: Instituto Agronômico de Campinas, 1985. 107 p. (Boletim técnico 100). www.iac.sp.gov.br/publicacoes/boletim100/versaoimprenssa.php (accessed 23 may 2018). Redmon, L.A., E.G. Krenzer, D.J. Bernardo, and G.W. Horn. 1996. Effect of wheat morphological stage at grazing termination on economic return. Agron. J. 88:94–97. doi:10.2134/agronj1996.0002196200 8800010020x Samac, D.A., and M. Tesfaye. 2003. Plant improvement for tolerance to aluminum in acid soils–a review. Plant Cell Tissue Organ Cult. 75:189–207. doi:10.1023/A:1025843829545 Schroder, J.L., H.L. Zhang, K. Girma, W.R. Raun, C.J. Penn, and M.E. Payton. 2011. Soil acidification from long-term use of nitrogen fertilizers on winter wheat. Soil Sci. Soc. Am. J. 75:957–964. doi:10.2136/ sssaj2010.0187

16

Sikora, F.J. 2006. A buffer that mimics the SMP buffer for determining lime requirements of soil. Soil Sci. Soc. Am. J. 70:474–486. doi:10.2136/sssaj2005.0164 Sloan, J.J., N.T. Basta, and R.L. Westerman. 1995. Aluminum transformations and solution equilibria induced by banded phosphorusfertilizer in acid soil. Soil Sci. Soc. Am. J. 59:357–364. doi:10.2136/ sssaj1995.03615995005900020013x Sumner, M.E., and W.P. Miller. 1996. Cation exchange capacity and exchange coefficients. In: D. Sparks, A. Page, P. Helmke and R. Loeppert, editors, Methods of soil analysis part 3—chemical methods. SSSA Book Ser., Madison, WI. p. 1201–1253. Sutradhar, A., R.P. Lollato, K. Butchee, and D.B. Arnall. 2014. Determining critical soil ph for sunflower production. Int. J. Agron. 2014: 894196. doi:10.1155/2014/894196 Tang, C., Z. Rengel, E. Diatloff, and C. Gazey. 2003. Responses of wheat and barley to liming on a sandy soil with subsoil acidity. Field Crops Res. 80:235–244. doi:10.1016/S0378-4290(02)00192-2 Tang, Y., D. Garvin, L. Kochian, M. Sorrells, and B. Carver. 2002. Physiological genetics of aluminum tolerance in the wheat cultivar atlas 66. Crop Sci. 42:1541–1546. doi:10.2135/cropsci2002.1541 Thomas, G. 1996. Soil ph and soil acidity. In: D. Sparks, A. Page, P. Helmke, R. Loeppert, P. Soltanpour, M. Tabatabai, C. Johnston and M. Sumner, editors, Methods of soil analysis. Part 3—Chemical methods. p. 475–490. True, R.R., F.M. Epplin, E.G. Krenzer, and G.W. Horn. 2001. A survey of wheat production and wheat forage use practices in oklahoma. B-815. Okla. Coop. Ext. Serv., Stillwater, OK. USDA-NASS. 2015. United states department of agriculture. National agricultural statistics service. Oklahoma wheat variety report. www. nass.usda.gov/Statistics_by_State/Oklahoma/Publications/Oklahoma_Crop_Reports/2015/ok_wheat_variety_2015.pdf (accessed 17 Jan. 2018). USDA-NASS. 2016. United states department of agriculture. National agricultural statistics service. Oklahoma wheat variety report. www. nass.usda.gov/Statistics_by_State/Oklahoma/Publications/Oklahoma_Crop_Reports/2017/ok_wheat_variety_2017.pdf (accessed 17 Jan. 2018). USDA-NASS. 2017. United states department of agriculture. National agricultural statistics service. www.nass.usda.gov/Quick_Stats/ (accessed 13 Dec. 2017). Valle, S.R., J. Carrasco, D. Pinochet, and D.F. Calderini. 2009. Grain yield, above-ground and root biomass of Al-tolerant and Al-sensitive wheat cultivars under different soil aluminum concentrations at field conditions. Plant Soil 318:299–310. doi:10.1007/s11104-008-9841-8 van Bussel, L.G.J., P. Grassini, J. Van Wart, J. Wolf, L. Claessens, et al. 2015. From field to atlas: Upscaling of location-specific yield gap estimates. Field Crops Res. 177:98–108. doi:10.1016/j.fcr.2015.03.005 von Uexküll, H.R., and E. Mutert. 1995. Global extent, development and economic impact of acid soils. Plant Soil 171(1):1–15. doi:10.1007/ BF00009558 Yang, Z.-M., H. Yang, J. Wang, and Y.-S. Wang. 2004. Aluminum regulation of citrate metabolism for Al-induced citrate efflux in the roots of Cassia tora L. Plant Sci. 166:1589–1594. doi:10.1016/j. plantsci.2004.02.012 Zhang, H., and W.R. Raun. 2006. Oklahoma soil fertility handbook. Oklahoma St. Univ., Stillwater, OK. Zhang, H.L., G. Johnson, G. Krenzer, and R. Gribble. 1998. Soil testing for an economically and environmentally sound wheat production. Commun. Soil Sci. Plant Anal. 29:1707–1717. doi:10.1080/00103629809370061 Zhou, L.L., G.H. Bai, H.X. Ma, and B.F. Carver. 2007. Quantitative trait loci for aluminum resistance in wheat. Mol. Breed. 19:153–161. doi:10.1007/s11032-006-9054-x
