Quantifying links between topsoil depth, plant water use, and yield in a rainfed maize field in the U. S. Midwest

Agricultural production in highly variable soils is a challenge, especially when those soils are shallow. Precision agriculture techniques were developed to improve yields and minimize spatial and interannual variability in profit. In the Central Claypan region of the Midwest United States, many of the precision agriculture techniques were based on the assumption that topsoil depth controlled plant available water, and therefore yield. But this assumption has not been empirically tested. In this study, we use measurements of sap flow installed on maize plants with a gradient in topsoil depth, caused by a claypan layer. We hypothesize that plants with higher water use have higher yield, plants in areas with thicker topsoil have higher water use, and soils in the areas with thicker topsoil have higher soil water content. Sap flow sensors were installed on 5 plants each at three locations with topsoil depths of 19.6 cm (shallow), 21.6 cm (medium), and 30.5 cm (deep) from June – September, 2022. An ANOVA analysis demonstrates that the average total season transpiration at the deep site (279 mm) was significantly larger than at the shallow site (151 mm), while the medium site was in the middle with transpiration not significantly different from either shallow or deep sites (218 mm). At the end of the season, the plants were harvested and total biomass and grain yield were measured. Increase in plant transpiration was significantly related to both increases in biomass and yield. Finally, we measured volumetric soil water content at each location and found higher soil water content at the site with thicker topsoil. Our results demonstrate the link between topsoil depth, soil water content, plant transpiration, and yield. These findings will help improve precision agriculture techniques in areas with highly variable topsoil thickness.


Introduction
Global food demand is increasing and to meet demand the clearing of additional land in more marginal areas is underway.The sustainable intensification of agriculture on marginal landscapes will be useful to feed a growing population and prevent further land clearing which has the potential to create large emissions of greenhouse gases (Tilman et al., 2011).In the U.S. Midwest, millions of acres of marginal land are currently being cultivated.For example, the Central Claypan Area (Major Land Resource Area 113) encompasses 3 million ha in Missouri and Illinois (USDA-NRCS, 2006).This region is characterized by claypan soils with high runoff potential due to the presence of a restrictive layer that makes the topsoil shallow.The low hydraulic conductivity in the claypan layer prevents infiltration from the topsoil, and leads to saturation excess runoff.The shallow soil and high runoff potential results in soil degradation that has been well characterized over the last 20 years (Baffaut et al., 2020a;Veum et al., 2015).Soil erosion is leading to shallow soils across much of the agricultural U.S. Midwest (Thaler et al., 2021).
Shallow soils may pose additional challenges to agricultural production as the climate changes.In the U.S. Midwest, precipitation events are expected to become more intense, but potentially less frequent (Byun and Hamlet, 2018;Feng et al., 2016).Shallow soils with lower water holding capacity will be incapable of absorbing large precipitation events (leading to increased runoff), which leads to not having sufficient soil water available for plants during the dry periods (Gautam et al., 2021).The claypan soils, combined with historical management practices, have resulted in agricultural land with high in-field variability of yield and profit margins (Conway et al., 2020;Kitchen et al., 2005a).Considerable effort has been underway to understand the yield variability and develop precision agriculture strategies to manage these fields (Yost et al., 2017;Kitchen et al., 2005b).Previous researchers have suggested that the water holding capacity of the soil controls this variability, but that link has not been demonstrated empirically (Kitchen et al., 2005b).
It has been known for decades that plant water use affects yield (Doorenbos and Kassam, 1979).Several studies in water limited agroecosystems have determined that maize plants with higher transpiration rates have higher yield (Feng et al., 2019;Zhang et al., 2021).In the water limited Nebraska farms of the U.S. Midwest, maize yield is linearly related to both evapotranspiration (ET) and transpiration (T) (Payero et al., 2006).In more humid regions, however, the relationship between T and crop yield is non-linear in well wetted areas and decreases in T do not necessarily result in decreases in yield (Purcell et al., 2007).Water stress during specific periods of plant development, however, does deplete yield (Çakir, 2004).Maize yield is most dependent on precipitation amounts during the months of June/July or July/August, demonstrating the connection between water availability and yield (Joshi et al., 2020;Jung et al., 2005).
Plant T is a function of both the atmospheric demand and the availability of soil moisture to meet that demand (Laio et al., 2001).Across the U.S. Midwest, the best hydrologic predictors of crop yield have been soil moisture and vapor pressure deficit (VPD), a measure of atmospheric demand (Zhou et al., 2020).The VPD depends on atmospheric conditions and is difficult for a single farmer to control; consequently, increasing the soil water availability may the best local strategy to create increased, stable yields.Soil water limitations on T have been well studied in dryland regions, but agriculture in these regions tends to be irrigated to offset the severe water limitations (Acevedo et al., 2022).In humid rainfed agriculture, the T limits imposed by soil moisture have not been studied as extensively (Falkenmark and Rockström, 2006).Plant available soil water depends on precipitation and on the ability of the soil to absorb and store that precipitation for plant use.This ability is constrained by interactions between soil type, plant communities, precipitation regime, as well as the topsoil depth (Pántano et al., 2017;Mei et al., 2018;Fellows and Goulden, 2017;Tromp-van Meerveld and McDonnell, 2006).
Various studies have shown that soil properties affect plant water use by allowing higher infiltration or preventing soil evaporation (Feng et al., 2019;Li et al., 2002;Rankoth et al., 2019).Soil layers that constrain the effective topsoil depth can have large impacts on plant productivity by limiting the plant available water holding capacity (Tilse et al., 2022).This is reflected in land surface models where accurate representation of topsoil depth plays a key role in improving the simulation of evapotranspiration (Gochis et al., 2010).Shallow soil limits the water holding capacity and therefore the ability of plants to extract stored soil water (Dang et al., 2006).In the Central Claypan region, shallow soils are common due to the presence of the claypan restrictive layer in places where soil erosion has reduced topsoil depth from pre-cultivation levels.Taken together, these previous findings highlight the importance of plant water use in predicting yield, as well as the role of soil moisture in controlling the plant water use.The link between topsoil depth, plant water use, and crop yield, however, has not been empirically demonstrated.
In this study we use measurements of sap flow, plant biomass, and grain production for maize plants located in a claypan soil field with high variability in topsoil depth.This field is in a humid region where the yield response to transpiration is expected to be non-linear.We aim to test three hypotheses.First, that plants with higher water use will have higher yield.Second, that areas with higher depth to claypan (i.e., deeper soil) will have higher soil water content.And finally, that plants in the locations with higher soil water content will have higher water use.We aim to quantify the impact of soil water in the previously reported relationships between topsoil depth and yield.

Study site
This research was carried out at the Central Mississippi River Basin (CMRB) Long-Term Agroecosystem Research (LTAR) site, which has observations dating to 1971 (Sadler et al., 2015b).Sensors were deployed in three locations in a field with a no-till maize-soybean-wheat-hay rotation with cover crops and precision fertilizer management.The study period is the cash crop growing season of 2022, which was the maize year of the rotation.The maize plants were planted directly into a standing cover crop that was terminated and crimped.Substantial cover crop residue was left on the field, which may have affected water dynamics, but impacts would have been similar across the entire field.Plants were seeded on May 12, 2022, but sensors could not be installed until the plants were large enough (vegetative growth stage V6).The soil type is consistent between the three locations and is Adco silt loam, part of the Central Claypan Area, MLRA 113 (USDA-NRCS, 2006).The field management uses conservation practices (no-till and cover crops) and as a result, soils at the site have higher nutrient content than conventional agricultural systems (Veum et al., 2015).A previous study found the carbon and nitrogen contents are highest in the surface layer (0-5 cm) and decay by the 5-15 cm layer.The organic carbon was 22.4 g/kg in the surface layer and 10.1 g/kg in the 5-15 cm layer while total nitrogen was 2.14 g/kg in the surface layer and 1.22 g/kg in the 5-15 cm layer (Veum et al., 2015).There is substantial variability in the depth to the claypan layer (DTC) at the CMRB site.Previous efforts at the site to determine soil properties have used the soil apparent electrical conductivity (EC a ) to map the DTC with high accuracy (the r 2 compared to validation samples was 0.88), data which is presented in Fig. 1 (Sudduth et al., 2010).Micrometeorological measurements were obtained from a micrometeorological tower placed in the middle of the primary CMRB LTAR field.Sap flow sensors were installed in locations within the measurement footprint of the micrometeorological tower with the largest differences in ECa, and therefore DTC.

Sap flow observations
Heat balance sap flow sensors (Baker and van Bavel, 1987) were installed on 15 maize plants throughout the study field.Three locations were chosen with maximum differences in DTC within the micrometeorolocial tower footprint and are called shallow, medium, and deep in reference to their relative topsoil depths.To minimize the impact of lateral water transfers, via either surface run-on or subsurface redistribution, we intentionally selected sites with similar slope values away from convergent zones.The slope is 0.46%, 0.55%, and 0.53% at the shallow, medium, and deep DTC sites, respectively.The number of locations was kept to 3 to minimize both costs for datalogger equipment and interference with field operations.At each location, SGEX-25 sensors (Dynamax Inc., Houston TX) were installed on 5 randomly selected plants once the plant diameters were large enough and the sensors were connected to a Flow32-1k sap flow system.The sensors have been shown to be effective at measuring plant transpiration (Lascano et al., 2016).Sensors were installed between the second and third nodes of the maize plant and following manufacturer recommendations.We removed the outer leaf, wrapped a layer of plastic wrap around the stem, applied G4 silicone compound to the sensor heater, and installed the sensor.Each sensor was kept in place with a flexible Velcro strip and a Goretex wrap was applied to keep water from precipitation out.Foam insulation as well as a reflective heat shield were wrapped around the sensor to minimize heat transfer with the air.Sap flux values were measured every minute and recorded for 30-minute averages during the growing season.The sheath conductance represents the heat conductance through the sensor material and is calibrated daily during no-flow conditions using a manufacturer supplied algorithm.We define no-flow conditions as the pre-dawn period from 3 am to 5 am.Gauges were installed on June 30, 2022 and removed September 26, 2022, before harvest.

Soil measurements
At the start of the measurement season the DTC at each site was measured by extracting a soil core from within the crop row and identifying the diagnostic pedogenic horizon.The DTC at the shallow site was 19.6 cm, 21.6 cm at the medium site, and 30.5 cm at the deep site.Installing soil moisture sensors for short term measurements, which would have been required in the actively managed agricultural field, is unreliable due to the time required for soil to settle after disturbance.Because of this, we took weekly gravimetric samples of soil moisture.For each sampling date, two cores were extracted at each of the DTC class locations (shallow, medium, deep) and mixed together.Multiple cores were used to reduce the impact of small scale spatial heterogeneity on the final soil moisture value.Each sample was sealed in a plastic bag, brought to the laboratory and weighed, then dried for 24 h at 105 • C and weighed again.To convert from gravimetric to volumetric soil moisture we used the bulk density.When soils are dry, as was the case throughout the summer growing season, bulk density is difficult to estimate from soil cores because the soil does not stick together and soil grains fall out the bottom of the auger.Because of this, we used a value of 1.24 g/cm 3 for bulk density, which was previously estimated at this field (Mudgal et al., 2010).Previous studies at the site characterized root distributions and found that the claypan layer inhibited root growth and that while some roots penetrated beneath the claypan, they were small (Myers et al., 2007).Thus, it is likely that the primary water source for plants is the topsoil above the claypan.To assess the total amount of water in the topsoil we multiplied the volumetric water content by the depth of the topsoil (i.e., DTC).

Plant biomass and yield
At the end of the growing season each plant was destructively harvested by clipping the aboveground biomass.The plants were dried and weighed to measure the total biomass of each plant.Then, the corn grain was removed from the cob, dried, and weighed.Two 5 m transect surveys were performed to estimate the density of plants in the field to be 7.66 plants per m 2 .This is used to scale transpiration from the plant scale to an areal average with units of mm/day, which is not required to test our hypotheses, but is useful for comparison with other sites and previous studies.

Meteorological measurements
A micrometeorological tower is deployed the middle of the CMRB LTAR field (Schreiner-McGraw et al., 2023) and several meteorological measurements for this study are derived from the data at the tower (Fig. 1).Net radiation is measured with a 4-component net radiometer (CNR4, Kipp and Zonen).Air temperature and relative humidity are measured using a heat shielded sensor (HC2A, Rotronic Measurement Solutions USA, Hauppauge, NY), which is located above the canopy and varied between 1.5 and 3.3 m height as the maize crop grows.A weighing bucket rain gauge has been continuously measuring precipitation at the site since 1991 and was used to measure precipitation (Sadler et al., 2015a).We used the 31-year timeseries of precipitation to calculate the standardized precipitation index (SPI), at a 3-month scale, to quantify the relative dryness of the study period growing season and antecedent conditions (Mckee et al., 1993).We calculated the SPI using a non-parametric approach that removes the assumption that monthly precipitation is normally distributed (Farahmand and AghaKouchak, 2015).

Hypothesis testing
We tested three hypotheses in this study.The first-plants with higher water use will have higher yield-was tested using the sap flow and yield data.We built relationships between total season sap flow and plant biomass as well as between total season sap flow and grain yield.We deemed the hypothesis substantiated if the slope of a linear relationship between each set of two variables was significantly different from 0, which indicates that there is a relationship.The total season sap flow consists of the sap flow that occurred between June 30, 2022 and September 26, 2022, the period of time when plants were large enough to support the sap flow sensors.Therefore, the early growing season water use is not accounted for.The second hypothesisthat locations with deeper topsoil will have higher water usewas tested with ANOVA and a post hoc Tukey HSD test, and we accepted the hypothesis if the locations with deeper topsoil had statistically significant higher water use at a confidence level of p < 0.05.Finally, the third hypothesis-plants with higher soil moisture had higher water usewas tested by comparing the total water use and soil moisture between the three locations.We use a repeated measures ANOVA and a post hoc Tukey HSD test, to test if the mean seasonal soil water content in the topsoil is significantly different between the three locations and if the relative differences in soil water content match the differences in total plant water use.We repeat the ANOVA and post hoc Tukey HSD tests on the volumetric soil water content.

Growing season weather conditions and sap flow
Between June 30 and September 26, 2022, which we term the growing season for this study, there was 287 mm of precipitation.The 3month SPI value during these months was 0.29, which indicates that the precipitation was approximately average during the study period.The early growing season, however, had moderately low precipitation, relative to the historical average.The 3-month SPI for the April, May, and June period was − 0.57.This indicates a "mild drought" during the period when the maize plants were establishing (Mckee et al., 1993).
A.P. Schreiner-McGraw and C. Baffaut This lack of precipitation carried through most of July until July 25 when 67 mm of precipitation fell (Fig. 2).As a result, for the 3-month period where sap flow measurements are available, the SPI was 0.29, which represents approximately average conditions.Therefore, the crop water use represents a growing season with average precipitation, but below average antecedent wetness.The VPD at the site reflected the dry conditions during the early growing season as values more negative than − 1 kPa were frequent during July, but rare in August and September.Daily air temperatures decreased slightly throughout the growing season but were relatively stable.The net radiation presented a seasonal decline after the summer solstice and large spikes were observed, caused by cloud cover.
The seasonal timeseries of sap flux, presented in Fig. 3a, suggests persistent differences in sap flux between locations.For most days throughout the growing season the daily average sap flow was highest at the deep DTC class and lowest at the shallow DTC class.Of the micrometeorological variables measured, daily sap flow was most tightly correlated to the available net radiation followed closely by the air temperature (Table 1).This phenomenon is most evident from July 26-28 when a large precipitation storm and cloudy skies persisted several days, decreasing both net radiation and air temperature.Daily values of sap flux for individual plants ranged from near 0 to greater than several thousand grams of water per day (Fig. 3b).The mean daily sap flux was higher for the plants with deeper topsoil and the maximum values were also highest in regions with deeper topsoil.The mean total season sap flux at the shallow DTC location was 19,652 g, 28,442 g at the medium DTC location, and 36,439 g at the deep DTC location, which is equivalent to 151 mm, 218 mm, and 279 mm, respectively.

Plants with higher water use have higher yield
The first hypothesis we tested is that plants with higher water use will have increased biomass production and higher grain yields.Dry plant biomass at harvest was significantly related to the total seasonal transpiration (Fig. 4a).There is considerable scatter in the relationship with an r 2 value of 0.41 and a standard error from the regression line of 63 g.But the p-value for the slope of the linear regression is 0.009 providing high confidence that the slope is significantly different from zero.Upon examination of the data, a linear relationship may not be appropriate, but with the limited number of data points a more complex function is outside the scope of this experiment.In addition to a statistically significant relationship between growing season transpiration and biomass, there is a significant relationship between growing season transpiration and maize grain yield (Fig. 4b).Again, there is considerable scatter in the data points with a r 2 value of 0.35 and a standard error from the regression of 45 g.But the slope of the linear regression is significant with a p-value of 0.02.Based on this evidence we accept the first hypothesis that there is an increase in plant biomass and grain yield with increases in transpiration.Fig. 5.

Plants with higher soil water availability have higher transpiration
By accepting our first hypothesis, we have established that higher plant transpiration leads to higher plant biomass and grain yield.Our second hypothesis is that higher soil water availability leads to higher transpiration when the atmospheric demand is uniform in space, as is generally the case within a single field.The time series of topsoil available water demonstrates that the location with deep topsoil consistently had higher available soil water than the other locations (Fig. 6a).For sampling dates from mid-July to mid-August, the medium DTC site had higher available water in the topsoil than the shallow site, but that did not persist through the end of the growing season.This may be because the plants were larger and were able to transpire more water than those at the shallow DTC site (Fig. 3 and Fig. 4).The differences in topsoil available water were not only caused by the thicker topsoil at the medium and deep sites; the average volumetric soil moisture content was 20% at deep DTC and 17% at the shallow and medium DTC sites (Fig. 6b).A repeated measures ANOVA test on the mean growing season soil moisture (treating each sample date as a replicate) with a Tukey post-hoc HSD test found that the deep DTC site had significantly higher topsoil water content and volumetric soil water content than the other two sites.Unfortunately, the weekly sampling frequency limits our ability to fully interpret the data.But as the deep site consistently had higher soil water available, we accept the hypothesis that plants with higher soil water availability have higher transpiration.A deeper investigation into the role of soil moisture on plant transpiration is warranted to fully quantify the connection.

Plants with deeper topsoil have higher water use
Finally, we test the hypothesis that plants with higher DTC have higher water use.As we noticed in Fig. 3, there appears to be a relationship between the DTC and the seasonal transpiration.This is confirmed in Fig. 6 where a single factor ANOVA test revealed that there are significant differences in the mean seasonal transpiration between the three sites (p = 0.018).A post-hoc Tukey HSD test found that the difference between the shallow and deep DTC are statistically significant (q = 4.76, q critical = 3.77), but that differences between shallow-medium DTC and medium-deep DTC are not significant.The same patterns of statistical significance are found if we use the mean daily flux or the median daily flux, rather than the total seasonal transpiration.This indicates that the relationship is robust to different metrics to calculate plant water use.The mean seasonal T was 19,652 ± 4368 for shallow DTC, 28,442 ± 7909 for medium DTC, and 36,438 ± 8207 for deep DTC.Therefore, we accept the hypothesis that the transpiration is highest on the deep DTC class.There is a significant difference between deep and shallow DTC and while the transpiration from medium DTC is in the middle of the deep and shallow sites, our evidence is not strong enough to differentiate it.

Relevance for precision agriculture
The precision agriculture movement has helped to stabilize yields both temporally and spatially, but it has yet to be demonstrated that precision agriculture approaches can substantially improve overall agricultural yields (Dhoubhadel, 2021;Yost et al., 2017).Precision agricultural practices have traditionally focused on managing farm inputs such as nitrogen and/or herbicides.But much of the premise was based on the idea that different zones of the field require different inputs because differences in water availability allow plants to grow more rapidly in certain areas (Kitchen et al., 1999).This hypothesis has been supported by model-based experiments (Mudgal et al., 2012).Previous studies have found that soil thickness is associated with increases in crop yields (Xiao et al., 2021) and that maize yields can be depressed by even short periods of water stress at inopportune timing (Çakir, 2004).
Research has also been performed to track sources of water in crops growing on shallow soils, for example by using isotopic tracers (Zhao et al., 2016).Our results are, to our knowledge, the first observational experiment suggesting that these hypotheses are correct and that the water availability is key to producing higher grain yields.We explicitly link topsoil depth to increased water availability, increased transpiration, and increased yield.This suggests that precision agriculture approaches should focus on crop management that is responsive to water needs of the plants.
An important caveat, however, is that higher volumetric water content (as opposed to higher soil column water from deeper soils) eventually results in there being too much water for the plants.For maize, this overabundance of water may promote denitrification and cause nitrogen stress, whereas for soybean oxygen limitation becomes problematic (Nótás et al., 2014;Kuzma et al., 1999).Research on soybean has shown that when soils are too wet in the early season, yield can be reduced by 25-30% (Bajgain et al., 2015).Tile drainage and irrigation can be an effective strategy to manage root zone soil water and boost grain yield, but it is expensive to install and often not possible in shallow soils (Singh and Nelson, 2021).Therefore, increases in soil available water may not always lead to increases in transpiration.If the volumetric soil water content is excessive, plant productivity may suffer.The size of this impact in maize plants, however, may be minimal.In a claypan soil in Illinois, tile drainage resulted in a 0.8 t/ha increase in maize yield, while tile drainage plus irrigation resulted in a larger increase of 4.8 t/ha (Walker et al., 1982).Regardless of the size of the impact, deeper soils will allow larger amounts of water storage before reaching saturation conditions.The study period in this experiment had

Table 1
Correlation coefficient between mean daily sap flux from each of the three DTC classes and meteorological variables including vapor pressure deficit (VPD), air temperature (TA), net radiation (R n ), and wind speed (US).approximately average precipitation, with an SPI value of 0.29.If precipitation was well above average, we may have seen a reversal of some of the relationships.In a wet year, relations between the soil water content and transpiration may reverse where waterlogged soils have lower transpiration rates and smaller plants.Further, long-term research to test the impact of DTC on crop water use and yields across a variety of precipitation conditions is warranted.

Relevance for sustainable agricultural practices
Precipitation patterns are changing in much of the world and a common theme is that precipitation is becoming less frequent, but more intense (Byun and Hamlet, 2018;Coelho et al., 2022;Li et al., 2022).Thus, even if the total annual precipitation does not change, the ability of the soil to store and retain that water for plants to use can be greatly affected.Soil water holding capacity is a key factor in controlling how farming systems will respond to altered precipitation patterns (Brempong et al., 2023).Soils with high water holding capacity allow greater water storage, which reduces erosion and provides a water source during subsequent dry periods.Our results show that deeper soils with higher water holding capacity were linked to greater plant water use, growth, and yield.Previous work on soybean demonstrated that plant root growth is inhibited near the claypan layer but that smaller roots penetrate through the claypan to access deeper water stores (Myers et al., 2007).The lack of roots near the claypan likely accentuates the impact of the claypan on water stress in shallow soils by further limiting the plant rooting depth.Some roots that penetrate the claypan, however, may access more temporally stable sources of water to limit the detrimental impacts of shallow topsoil.Further research to quantify how topsoil thickness affects where in the soil profile plants draw water from is warranted.
In a changing climate the importance of water holding capacity is only likely to increase.Thus, minimizing topsoil erosion, which is widespread in the U.S. Midwest, will be key to preserving crop yields, especially in claypan regions (Baffaut et al., 2020b;Thaler et al., 2021).Proposals for mitigating the impacts of climate change on crop water availability through the large-scale development of retention ponds coupled with irrigation are promising, but costly (Baker et al., 2012).Improving soil water holding capacity may provide some of the benefits at a small fraction of the cost.Practices such as cover crops and no-till are known to increase the water holding capacity of soil and should be considered in farmland with restrictive layers or shallow soils (Abdallah et al., 2021;Blanco-Canqui et al., 2015).Deeper soils are capable of holding more water without becoming saturated, which helps maintain plant available water while ensuring sufficient oxygen available for plant roots.For soils with claypan layers, perennial grasses can penetrate the restrictive layer with permanent roots, which adds soil structure and can improve the infiltration and water holding capacity (Zaibon et al., 2016(Zaibon et al., , 2017)).The available water holding capacity is an important indicator of soil health but is measured from disturbed samples (Moebius-Clune et al., 2016).While the laboratory measured available water capacity may not be physically meaningful due to soil disturbance, our results suggest that soil water holding capacity is important to consider and that the important role of topsoil depth should be considered in soil health analyses in areas with potential for restrictive soil layers.

Conclusions
Previous research highlighted the connections between DTC and crop yield, where deeper soils led to higher yield.Researchers hypothesized that the DTC affected the crop yield because it limits the water holding capacity of the soil, which exposes crops to drought stress even after very short periods with no precipitation.In this study, we have shown the link between DTC, soil water content, plant transpiration, and crop yield.We used sap flow sensors installed on maize plants in three DTC classes and showed that plants with deeper topsoil have higher water use.After the growing season, each of the plants was manually harvested and the total biomass and grain yield were measured.We found that plants with higher total season transpiration had higher total biomass and higher yield.Finally, we showed that the deep DTC class had higher soil water content, likely leading to the higher transpiration.This information will inform precision agriculture and crop water management practices in a changing climate.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Beshears' help to install and maintain the instrument networks.We also thank Megan Metz, Teri Oster, as well as the many other USDA-ARS employees who made this work possible.This research was supported by the U.S. Department of Agriculture, Agricultural Research Service, project 5070-12000-001-000D.This research was a contribution from the Long-Term Agroecosystem Research (LTAR) network.LTAR is supported by the United States Department of Agriculture.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.USDA is an equal opportunity provider and employer.

Fig. 1 .
Fig. 1.Location of the study field within the state of Missouri, USA.The apparent electric conductivity (ECa) of the field, which has been shown to be inversely related to depth to claypan (DTC), is shown.Locations of the micrometeorological (met) tower and sap flux sensors with their relative DTC descriptors are indicated.Each sap flux location had 5 plants instrumented.

Fig. 3 .
Fig. 3. (a) Timeseries of the daily average sap flux at each of the three depth to claypan (DTC) classes (shallow = 19.6 cm; medium = 21.6 cm; deep = 30.5 cm), daily precipitation is also displayed.(b) Jitterplot of the daily sap flux.Each point represents the daily sap flux from one of the instrumented plants.The mean value for each DTC class is displayed with a horizontal line.

Fig. 4 .
Fig. 4. Relationships between seasonal transpiration and (a) plant biomass and (b) grain yield for the 15 plants instrumented (black circles).Statistically significant linear relationships are presented with a solid line.

Fig. 5 .
Fig. 5. Average weekly (a) topsoil water content (mm) and (b) topsoil volumetric water content(-) for each of the three locations.

Fig. 6 .
Fig. 6.Mean transpiration from each of the depth to claypan classes.Error bars represent the standard deviation and letters indicate groups with statistically significant differences in mean.