HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (21) Citing Articles via Google Scholar Google Scholar Articles by Sadler, E.J. Articles by Millen, J. A. Search for Related Content PubMed Articles by Sadler, E.J. Articles by Millen, J. A. Agricola Articles by Sadler, E.J. Articles by Millen, J. A. Related Collections Water Use Water Stress Maize Site-Specific Analysis Soil Analysis

SITE-SPECIFIC ANALYSIS

Site-Specific Analysis of a Droughted Corn Crop

II. Water Use and Stress

Coastal Plains Soil, Water, and Plant Research Center, USDA-ARS, 2611 West Lucas St., Florence, SC 29501-1242 USA

sadler{at}florence.ars.usda.gov

	ABSTRACT

TOP ABSTRACT INTRODUCTION Methods NOTES Results and discussion Summary and conclusions REFERENCES

 
In the southeastern USA Coastal Plain, spatial variation in soils causes extreme spatial variation in grain yield, as seen in yield maps. Corn (Zea mays L.) appears to be particularly susceptible to soil variation, especially during periods of drought. Our objectives were to compare variation in water use and stress of corn within and among soil map units. In one field, at two sites in each of four map units, we measured site-specific effects of soil variation on crop water use from 40 d after planting until after maturity using a time-domain reflectometer (TDR). On 4 d during vegetative growth, drought stress was evaluated on eight transects using infrared thermometer (IRT) measurements of canopy temperature (T_c). During the most severe drought, visibly stressed areas had canopy-air temperature differences (T_c - T_a) > 10°C, yet other areas remained <2°C. Two days after a 46-mm rain, T_c - T_a was near zero over the whole field, indicating little water stress. The time series of TDR measurements produced estimates of daily evapotranspiration, runoff, and infiltration; site-to-site differences in these dominated the water balance. Water stress, inferred from water use, matched that inferred earlier from yield components. In sum, corn at the eight sites arrived at final water use via fundamentally different paths. Further, variation between sites within soils was significant, indicating that soil map units are not homogenous with respect to water relations. These results underscore the need for within-season observations of crop water use and stress to augment interpretation of site-specific yield maps.

Abbreviations: AW, available water content • DAP, days after planting • DOY, day of year • Et, evapotranspiration • Et_a, actual evapotranspiration • Et_r, reference evapotranspiration • IRT, infrared thermometer • GPS, global positioning system • K_c, crop coefficient • K_s, soil coefficient • LAI, leaf area index • T_a, air temperature • T_c, canopy temperature • TDR, time-domain reflectometer • TSW, total soil water content • WUE, water use efficiency

	INTRODUCTION

TOP ABSTRACT INTRODUCTION Methods NOTES Results and discussion Summary and conclusions REFERENCES

 
THE NEW, RAPIDLY EXPANDING field of site-specific, or precision, farming presents a greatly increased demand for agronomic knowledge. However, classical statistical replicated experimental procedures are neither inexpensive nor well-suited to solving spatial problems. This new knowledge of plant–culture–soil relationships on a close spatial scale must be developed using a combination of spatial field measurements and computer simulation modeling (Robert, 1996).

Although the precision farming movement began with fertilizer management (Wollenhaupt and Buchholz, 1993), an increasing body of knowledge suggests that spatial variation in soil water relations may be an important factor in causing spatial variation in grain yield. First, there is the well-known dependence of water holding capacity on texture, which itself is known to vary spatially. Second, despite many attempts, there has been little success correlating spatial grain yield to spatial patterns in fertility (Pierce and Nowak, 1999). Third, the literature is replete with examples of spatial water stress inferred from spatial canopy temperature patterns observed using airborne and satellite remote sensing (Moran and Jackson, 1991; Kustas and Norman, 1996) and handheld infrared thermometers (Sadler et al., 1995). When these observations are considered in the context of the long-known dependence of yield on applied water (e.g., Howell et al., 1990) that gave rise to the concept of water use efficiency, one must conclude that water stress bears increased consideration as a candidate for causing spatial variation in yield.

What is not so easily concluded, however, is a suitable approach to measuring components of water relations with a useful spatial extent. It is not feasible to measure enough sites to fully map water use over the whole area using soil profile water balance measurements with, for instance, a time-domain reflectometer. However, local experiences with recurring patterns in yield often suggest locations from which spatial water use might be inferred, especially if supported by complementary data taken on a more dense spacing.

Thus, the combination of soil water contents and canopy response, coupled with collateral crop measurements reported in a companion paper (Sadler et al., 2000), would provide a comprehensive dataset with which to study spatial variation in crop water relations. However, simple analyses of cause-effect candidates were not expected to succeed because of the multiple, complex interactions among water and energy transfer processes and their dynamics during the season. For this reason, we expect final analysis for these data to require the use of process-level computer models of corn growth and yield. Such is beyond the scope of this paper, but results presented here should indicate processes that are important and should be considered in the models. As a result of the considerations presented above, water stress has been put forth as a candidate cause for site-specific variation in grain yield that should be considered in any explanation on sandy soils. Therefore, our objectives were to compare variation in water use and stress within and among soil map units and to evaluate variation in water relations as a cause for variation in grain yield.

	Methods

TOP ABSTRACT INTRODUCTION Methods NOTES Results and discussion Summary and conclusions REFERENCES

 
Overview
This research was conducted during the 12th cropping season of a study designed to document spatial variability of crop yield within a representative Coastal Plain field (Karlen et al., 1990; Sadler et al., 1993, 1995). The study followed a typical conventionally tilled corn–wheat (Triticum aestivum L.)–soybean [Glycine max (L.) Merr.] rotation. In 1993, the rotation phase was corn. During this season, the plan was to use soil and plant characteristics to evaluate causes of yield variation. Variation in plant characteristics is examined in Part I (Sadler et al., 2000); this work examines variation in water use and stress.

The 8-ha field, representative both of field size and soil variability in the Coastal Plains, includes 14 soil map units mapped on 1:1200 scale. Much of the variation is associated with two small, shallow depressions called Carolina Bays. The southwest corner of the field is located at 34°14'44'' N, 79°48'34'' W, as determined by averaging differential GPS readings for a period of 1 h.

Representative Sites
Four soil map units (USDA-SCS, 1986) were selected to represent the range of soils within the field. The descriptions include Goldsboro loamy fine sand (GoA; fine-loamy, siliceous, subactive, thermic Aquic Paleudult), Norfolk loamy fine sand (NkA; fine-loamy, siliceous, subactive, thermic Aquic Paleudult), Bonneau loamy fine sand (BnA; fine-loamy, siliceous, subactive, thermic Aquic Paleudult), and Coxville loam (Cx; fine-loamy, siliceous, subactive, thermic Aquic Paleudult). Two sites were chosen for each map unit (see Fig. 1 in Sadler et al., 2000, for a diagram). After TDR installation (see below), it was discovered that Site 1, which was to represent GoA, was placed by error on the boundary between GoA and Dunbar (Dn), but the difference between these two inclusions is less than between typical pedons of the two soils, and the profile was similar to the other GoA.

View larger version (34K): [in this window] [in a new window]   Fig. 1 Crop temperature measured with an infrared thermometer as a function of distance along Transect 1 for one date before and for three dates after a 46-mm rain on 12 June 1993. The area around Site 1 appeared to be the most stressed in the field

Soil Water Balance Calculations
Whole-profile water content was obtained by simple rectangular integration of soil water content over depth. The water balance of the profile was assumed to consist entirely of net rainfall and evapotranspiration (Et) during the drought, an assumption that was likely adequate at all sites except possibly Site 8 after the 12 June rain. Rain was measured at a weather station located 300 m from the field. Available water content was calculated using best estimates of drained upper limit and lower limit of plant-available water for each layer at each site. These estimates were obtained by inspection of TDR water contents, supplemented by measurements on similar soils and by literature values (Long et al., 1969; Peele et al., 1970). The rooting depth was estimated by inspection of the TDR traces to judge whether withdrawal occurred from a layer at a given time.

Time series of water content at each site were interpolated to a daily basis using the procedure described below (Schwab et al., 1993) to account for rain and expected evapotranspiration. The procedure used the same basic equation in two ways, depending on whether rain occurred between the TDR measurements.

When there was no rain between measurements, differences in soil profile water contents were used to estimate daily actual evapotranspiration (Et_a). Using this value of Et_a, a calculated reference Et (Et_r), and a soil (K_s) factor (see below), we solved for the crop (K_c) factor. When rains occurred between measurements, an interpolated value of the crop factor was used to estimate Et_a. The definition of the soil and crop factors follows from the equation

(1)

In the case without rain, Et_a and Et_r were known, and Eq. [1] was solved for K_c, using K_s taken from Haan et al. (1994):

(2)

where AW is available water content expressed as a percentage. In the case where rain occurred between measurements, so that K_c was not obtainable by direct solution, daily values were interpolated from the seasonal pattern of measured K_c values at each site using a simple cubic polynomial. Then, with Et_r, K_s, and K_c known, Eq. [1] was used to calculate Et_a.

Transect Measurements of Canopy Temperature
Corn growth during 1993 was extremely variable because of drought. On 10 June, the visibly stressed areas in the field were markedly distinct from others that were not apparently stressed. The plant heights ranged from 0.48 m in stressed areas to 1.34 m in less stressed areas. Other signs included a blue-gray cast and severe leaf rolling. In contrast, the areas with the tallest plants showed no visible stress.

Documenting the spatial patterns of stress was done by measuring canopy temperature on the eight transects with an infrared thermometer (IRT; Model 4000, 4° field of view, Everest Interscience, Tustin, CA). To do this, an IRT was connected to a datalogger (CR21X, Campbell Scientific, Logan, UT) mounted on a platform strapped to an operator's waist. The operator walked along the row at a steady pace, pointing the IRT forward and down at a 45° angle above the row. At 1-s intervals, the datalogger recorded the temperature. A manual switch allowed the operator to start and stop at known locations. Assuming the pace was steady (average was 1.4 m s^-1), the time of the individual measurements allowed computation of location. The first such measurements were made on 10 June during the most severe stress period. After the 46-mm rain on 12 June, measurements were made on 14, 17, and 19 June. Sky conditions were clear for all IRT measurements. On the four dates, the series commenced at 1430h, 1445h, 1240h, and 1145h (EST), and the total duration for each day was 30 min.

	Results and discussion

TOP ABSTRACT INTRODUCTION Methods NOTES Results and discussion Summary and conclusions REFERENCES

 
Spatial Canopy Temperature
Canopy minus air temperature data are shown in Fig. 1 and 2 for Transects 1 and 5. Transect 1 includes Site 1, which appeared to be the area with the most severe stress. Transect 5 includes the additional Site 11 (NoA) described in Sadler et al. (2000), which never showed visible signs of stress. The four sets of measurements show the severe drought stress on 10 June before the 12 June rain, near total relief of stress on 14 June after the rain, and the development of drought conditions nearly as severe as before on 17 June and 19 June. The variation in T_c - T_a increased with drought stress. This supports Aston and van Bavel's (1972) assertion that variation in canopy temperature could be used as an early indicator of the need for irrigation. In the humid Southeast, cloudiness may prevent irrigation scheduling using field-scale variation in IRT readings. However, one may choose a limited transect where measurements could be made during cloud-free periods. The first 40 m of Transect 5, which includes the GoA, NcA, and NoA soils, might be such an area because it spans the range of responses observed in the field.

View larger version (29K): [in this window] [in a new window]   Fig. 2 Crop temperature measured with an infrared thermometer as a function of distance along Transect 5 for one date before and for three dates after a 46-mm rain on 12 June 1993. The NoA site (Site 11) noted was never observed to be stressed, though final yields were quite low relative to norms

View larger version (36K): [in this window] [in a new window]   Fig. 3 Soil water balance at the eight original representative sites as a function of date during the season. Symbols represent measurements; dashed lines indicate the interpolated values using procedures listed in the text. Rainfall is indicated on the right y-axis

View larger version (34K): [in this window] [in a new window]   Fig. 4 Ratio of actual evapotranspiration (Eta) to reference evapotranspiration (Etr) as a function of date during the season for six of the eight sites (Sites 4 and 7 were close to 2 and 6, respectively). The lines were fit using the cubic spline smoothing interpolation feature (SM25) of PROC GPLOT (SAS Inst., 1990)

View larger version (24K): [in this window] [in a new window]   Fig. 5 Plot of soil coefficient (Ks) as a function of day of year (DOY). Points are connected by spline interpolation

View larger version (17K): [in this window] [in a new window]   Fig. 6 Plot of crop coefficient (Kc) as a function of day of year (DOY). Curves are cubic polynomials

View larger version (15K): [in this window] [in a new window]   Fig. 7 Plot of crop coefficient (Kc) as a function of leaf area index (LAI). Curve is a least-square fit rectangular hyperbola with the intercept forced to 0.2 by convention and the asymptote left unconstrained

View larger version (34K): [in this window] [in a new window]   Fig. 8 Plot of cumulative actual evapotranspiration (Eta) as a function of day of year (DOY). End-of-season values for cumulative Eta and grain yield (dry weight) were used to derive water use efficiency (WUE)

Comparison of Paired Samples
The 38 dates for which TDR values were measured for all sites provide a dataset amenable to evaluation of the variation within sites. Total soil water content (TSW), compared using a t-test, was significantly different ( = 0.05) between the two GoA sites (1 and 2) and also between the two NkA sites (3 and 4), as seen in Table 1 . The Cx and BnA sites were not different according to this criterion. However, because of the large contrast between soil water retention curves for sand and clay, site-to-site differences in depth to clay would be expected to cause large differences in TSW. These can easily be accounted for using linear regression of seasonal mean TSW against depth to clay, which explained 70% of the variation in TSW. When this line was subtracted from TSW for each site, thus eliminating the known effect of depth to clay, differences between the Cx sites went from marginally insignificant to significant, NkA remained significant, BnA went from nonsignificant (0.21) to nearly significant (0.055), and GoA went from significant at 0.0003 to insignificant at 0.9947. Thus, essentially all differences between the two GoA sites were accounted for by the differences in depth to the clay horizon. In the final analysis, the NkA sites were different, the Cx sites were different if adjusted for depth to clay, the GoA sites were different until adjusted for depth to clay, and the BnA sites were not different in either case.

Evaluation of Transect Data
Attribution of the IRT data to soil map unit provided a dataset amenable to analysis of variance using soil map unit as a class variable. The simple model, T_c - T_a = soil map unit, though significant for all dates at P > F = 0.0001 (n = 800), explained only 30, 7, 35, and 21% of the field variation in T_c - T_a for the four dates. Even in the first, third, and fourth dates, where 20 to 35% of the variation was explained, 65 to 80% of the variation occurred within soil map units.

Examination of the dataset of T_c - T_a within 5 m of the eight sites allowed an examination of variance attributable to soil map unit differences, to site-to-site differences, and to within-site differences. The simple model T_c - T_a = soil map unit on the four soil map units at the sites was always significant at = 0.05 (n = 56) and explained 38, 23, 20, and 19% of the variation for the four dates. However, the model T_c - T_a = site for the eight sites explained 56, 40, 71, and 35% of the variation, which indicates that 18, 17, 51, and 16% of the variation occurred between sites within soils. Even so, there remained 44, 60, 29, and 65% of the variance unexplained, and thus attributable to either measurement error or variation in T_c - T_a within the 10-m distance at each site.

Linking T_c - T_a and Soil Water
Given the relative ease of obtaining canopy temperatures, either with handheld IRTs or by remote sensing in thermal wavebands, it would be quite useful to have a relationship between T_c or T_c - T_a and soil water content, measurement of which is both difficult and time-consuming. The subset of T_c - T_a values within 5 m of each of the eight sites allows such a relationship to be examined for the range of soil water contents that existed on the four IRT measurement dates. The TDR measurements were conducted on 10 June, 14 June, 16 June, 18 June, and 21 June, so direct use of the first two dates was possible. Values corresponding to 17 June and 19 June were obtained by linear interpolation. Linear regression showed that nearly 60% of the variation in T_c - T_a was explained by fraction of available water content (Fig. 9) . Although not conclusive, this result certainly suggests that this concept merits further study.

View larger version (19K): [in this window] [in a new window]   Fig. 9 Plot of Tc - Ta as a function of fraction of available water content for the four infrared thermometer measurement dates

	Summary and conclusions

TOP ABSTRACT INTRODUCTION Methods NOTES Results and discussion Summary and conclusions REFERENCES

By inspection, t-test, and analysis of variance, it was shown that significant differences in total soil water content and T_c - T_a exist between sites for some soil map units, and also in short distances (<10 m) for T_c - T_a. Thus, reliance on a single soil description for an entire soil map unit, even one obtained on 1:1200 scale, will likely mean that variance in the field cannot be explained.SAS Institute 1990; 1986

	NOTES

TOP ABSTRACT INTRODUCTION Methods NOTES Results and discussion Summary and conclusions REFERENCES

 
¹ Mention of trade names is for informational purposes only. No endorsement is implied by the USDA-ARS.

Received for publication June 15, 1998.

	REFERENCES

TOP ABSTRACT INTRODUCTION Methods NOTES Results and discussion Summary and conclusions REFERENCES

 

• Aston A.R., van Bavel C.H.M. Soil surface water depletion and leaf temperature. Agron. J. 1972;64:368-373.[Abstract/Free Full Text]
• Baker J.M., Allmaras R.R. System for automating and multiplexing soil moisture measurement by time-domain reflectometry. Soil Sci. Soc. Am. J. 1990;54:1-6.
• Haan C.T., Barfield B.J., Hayes J.C. Design hydrology and sedimentology for small catchments. San Diego: Academic Press, 1994.
• Howell T.A., Cuenca R.H., Solomon K.H. Crop yield response. In: Hoffman G.J., et al. , ed. Management of farm irrigation systems. St. Joseph, MI: Am. Soc. of Agric. Eng, 1990:91-122.
• Karlen D.L., Sadler E.J., Busscher W.J. Crop yield variation associated with Coastal Plain soil map units. Soil Sci. Soc. Am. J. 1990;54:859-865.[Abstract/Free Full Text]
• Kustas W.P., Norman J.M. Use of remote sensing for evapotranspiration monitoring over land surfaces. Hydrol. Sci. J. 1996;41(4):495-516.
• Long F.L., Perkins H.F., Carreker J.R., Daniels J.M. Morphological, chemical, and physical characteristics of eighteen representative soils of the Atlantic Coast flatwoods. Res. Bull. 59. Athens, GA: USDA-ARS and UGA, College of Agric. Exp. Stn, 1969.
• Moran M.S., Jackson R.D. Assessing the spatial distribution of evapotranspiration using remotely sensed inputs. J. Environ. Qual. 1991;20:725-737.[Abstract/Free Full Text]
• Peele, T.C., O.W. Beale, and F.F. Lesesne. 1970. The physical properties of some South Carolina soils. Tech. Bull. 1037, South Carolina Agric. Exp. Stn., Clemson Univ., Clemson, SC.
• Pierce F.P., Nowak P. Aspects of precision farming. Adv. Agron. 1999;67:1-85.
• Robert P.C. Appendix I. In: Robert P.C., et al. , ed. Precision Agriculture: Proc. 3rd Int. Conf., Minneapolis, MN. 23–26 June 1996. Madison, WI: ASA, CSSA, and SSSA, 1996:1193-1194 Summary of workgroups: Precision agriculture research & development needs..
• Sadler E.J., Bauer P.J., Busscher W.J. Site-specific analysis of a droughted corn crop: I. Corn growth and yield. Agronomy J. 2000;92:395-402 (this issue).[Abstract/Free Full Text]
• Sadler, E.J., and W.J. Busscher. 1993. Soil water content and water use for conventional and conservation tillage in the SE Coastal Plain. p. 327. In 1993 Agronomy abstracts. ASA, Madison, WI.
• Sadler E.J., Busscher W.J., Karlen D.L. Site-specific yield histories on a SE Coastal Plain field. In: Robert P.C., et al. , ed. Soil specific crop management. Madison, WI: ASA, CSSA, and SSSA, 1995:153-166.
• Sadler E.J., Evans D.E., Busscher W.J., Karlen D.L. Yield variation across Coastal Plain soil mapping units. In: Robert P.C., et al. , ed. Soil specific crop management. Madison, WI: ASA, CSSA, and SSSA, 1993:373-374.
• SAS Institute. SAS/GRAPH Software: Reference. Cary, NC: Version 6. 1st ed. Vol. 1–2. SAS Inst, 1990.
• Schwab G.O., Fangmeier D.D., Elliot W.J., Frevert R.K. Soil and water conservation engineering, 4th ed New York: John Wiley & Sons, 1993.
• USDA-SCS. 1986. Classification and correlation of the soils of Coastal Plains Research Center, ARS, Florence, South Carolina. South National Technical Center, Ft. Worth, TX.
• Wollenhaupt N.C., Buchholz D.D. Profitability of farming by soils. In: Robert P.C., et al. , ed. Soil specific crop management. Madison, WI: ASA, CSSA, and SSSA, 1993:199-211.

This article has been cited by other articles:

				T. W. Katsvairo, W. J. Cox, H. M. Van Es, and M. Glos Spatial Yield Response of Two Corn Hybrids at Two Nitrogen Levels Agron. J., July 1, 2003; 95(4): 1012 - 1022. [Abstract] [Full Text] [PDF]

				J. K. F. Roygard, M. M. Alley, and R. Khosla No-Till Corn Yields and Water Balance in the Mid-Atlantic Coastal Plain Agron. J., May 1, 2002; 94(3): 612 - 623. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (21) Citing Articles via Google Scholar Google Scholar Articles by Sadler, E.J. Articles by Millen, J. A. Search for Related Content PubMed Articles by Sadler, E.J. Articles by Millen, J. A. Agricola Articles by Sadler, E.J. Articles by Millen, J. A. Related Collections Water Use Water Stress Maize Site-Specific Analysis Soil Analysis