The measurement of crop water stress under field conditions is fraught with technical and logistical problems. Although leaf water potential has become a standard measurement it has spatial and temporal sampling limitations. In the current study, a rice (Oryza sativa L.) crop was used to compare eight measurements indicating crop water status, namely leaf water potential Ww)> stomatal resistance (rj, transpiration rate (T), net photosynthesis rate (PN). canopy temperature (Tc), canopy minus air temperature (Tc — TJ, crop water stress index (CWSI), and visual leaf rolling score (LRS). The eight measurements were compared across seven water stress levels created by a line source sprinkler irrigation system. The methods were compared for accuracy, sensitivity, sampling time, and the destructive-disruptive nature of the sampling procedure. Accuracy was estimated by comparison with ^w and by the interaction between water stress level and time of day. All methods except PN were significantly correlated with laf at the 1% level. However, Tc, Tc — T. and CWSI showed less interaction between water stress level and time of day when total variance was partitioned into its relative components; water stress level, time, and the interaction between water stress level and time of day. All methods, with the exception of PN, were equally sensitive to the water stress gradient when "goodness of fit" response functions across the water stress gradient were compared. The visual LRS was the most rapid while the canopy temperature-based measurements, T,., Tc — T. and CWSI, were nearly three times faster than the gas exchange techniques and about two and a half times faster than lo>{. Leaf water potential sampling was both destructive and disruptive to the crop plant community. The gas exchange methods were nondestructive but repetitive sampling was disruptive. Only the remotely sensed Tn T€ T., CWSI and LRS were neither destructive nor disruptive to the crop. The interpretive value of various measurements is discussed. The CWSI was found to be highly correlated with mean daily PN and represents a significant advancement in crop level detection and measurement of water stress. Additional index words: Rice, Oryza sativa L., Leaf water potential, Stomatal resistance, Transpiration rate, Net photosynthesis rate, Canopy temperature, Crop water stress index, Leaf rolling. T abiJity to quantify plant or crop water deficits is fundamental to research on the response of plant communities to water stress. Great progress toward this end was made in the 1960s and 1970s (Turner, 1981) particularly because of the development of the pressure chamber for measurement of leaf water potential. Because of its portability and lack of variation with temperature, the pressure chamber (Scholander et al., 1965) is well suited to field observations (Ritchie and Hinckley, 1975). However, due to the dynamic diurnal nature of individual leaf, tiller and plant water status, the limited number of observations per hour, and the logistical problems associated with large scale field sampling, adequate sampling of crop level water status with the pressure chamber continues to be problematical. In addition to the difficulty associated with sampling, leaf water potential soon lost its hoped for wide applicability as interpretation of tissue water potential was found to be dependent upon the history of the crop's water relations (Thomas et al., 1976; Hsiao et al., 1984) and growth stage (Morgan, 1977; Sinha et al., 1982). Plant water potential of a single crop species has been found to vary across locations or growth conditions with reference to the apparent degree of stress required to bring about a particular physiological response, i.e., stomatal closure (Boyer, 1976; Thomas et al., 1976). Thus, indirect rapid methods of indicating plant water status have been sought to aid in screening germplasm and measurement of physiological activity per se in large scale field experiments. In recent years leaf rolling and canopy temperature have been suggested as indirect methods of quantifying crop water stress. Leaf rolling scores of cereals has been linearly related to leaf water potential over a range of potentials (O'Toole and Moya, 1978; Jones, 1979; Begg, 1980). Likewise canopy or foliage-to-air temperature difference has been related to leaf water potential (Idso et al., 1981b). 1 Contribution from the Agronomy Dep., The Int. Rice Res. Inst., P.O. Box 933, Manila, Philippines. Received 12 Mar. 1984. 1 Agronomist, visiting scientist, former research assistant, postmasteral fellow, and statistician, Int. Rice Res. Inst. Present address of N.C.T.: CSIRO Dryland Crops and Soils Research Program, Private Bag, P.O., Wembley, WA 6014, Australia. 1122 CROP SCIENCE, VOL. 24, NOVEMBER-DECEMBER 1984 In a recent series of papers (Idso et al., 1981a; Jackson et al., 1981), canopy temperature measurement has been used in the development of a plant or "crop water stress index" (CWSI). Briefly stated, the concept holds that at any given vapor pressure deficit there is a theoretical upper and lower limit of the canopy-to-air temperature differential (To --Ta). When a measure of foliage or canopy temperature (To) is made it can be related to the ratio of actual evapotranspiration to potential evapotranspiration (ETa/ETp) by the prevailing conditions of vapor pressure deficit, and air temperature (Idso, 1981 a). From these inputs the CWSI can be calculated. The index is theoretically analogous to 1--(ETa/ETp) (Jackson et al., 1981). The physiological measurements used to detect water deficits at the leaf level are leaf ex.pansion, stomatal resistance and rates of transpiration and photosynthesis. Although frequently not linearly related to leaf water potential (Turner, 1981) these measurements represent processes of production that can be markedly affectedby water stress. In the present study we sought to compare various methods used in the detection and interpretation of crop water deficits. We sought not simply to compare the ability to detect stress but also the accuracy and speed of measurement, as these factors strongly affect the choice of methods for use in field conditions. MATERIALS AND METHODS The upland rice (Oryza sativa L.) cultivar IRAT 13 was grown in a silty clay loam soil (Typic Hapludoll) at a dryland site on the experimental farm of the International Rice Research Institute (IRRI), Los Bat3os, Philippines (14°13’ N Lat, 121°15’ E Long). Details of the site are given in Cruz and O’Toole (1984). On 3 Feb. 1983 seed was sown in rows --~25 cm apart at a rate of 10 g m-~ to give a final plant population of 35 plants per m of row, i.e., 140 plants m-*. Twenty eight kg ha-~ of N, 6 kg ha-~ of P and 15 kg ha-t of K were applied at seeding and a further 9 kg ha -1 of N and 2 kg ha-t of P were applied 42 days after sowing. The main plot was 16 m long, 2 m wide and had a 2 m border of rice surrounding it. For the first 42 days after sowing the plot was evenly irrigated by overhead sprinklers to keep the soil near field capacity. From 43 to 55 days after sowing, the plot was irrigated with a line source sprinkler system (Hanks et al., 1976; Puckridge and O’Toole, 1981) that applied a continuously decreasing amount of water along the length of the plot. Fifty days after seeding eight water stress measurement techniques were used to sample subplots across the irrigation gradient. Within the main plot seven subplots were positioned perpendicular to the line source irrigation system. Each supblot was approximately 2.3 m long and 2 m wide and received decreasing amounts of irrigation water (Table 1). Plants within these subplots were used for measurement of water stress. The rice rows were grown parallel to the line source irrigation system. To enhance specific measurement comparisons, techniques which required single plant or leaf sampling were confined to one row near the subplot center (Table 1). The following water stress detection measurements were used: leaf water potential (~b~e,f), leaf diffusive resistance (r,), transpiration rate (T), net photosynthesis rate (PN), canopy temperature (To), canopy minus air temperature (To --Ta), crop water stress index (CWSI), and visual leaf rolling score Table 1. Distance from the line source sprinkler of single rows selected for repetitive sampling within each subplot. Changes in amount of water applied, soil water content, and canopy light interception from 42 days after seeding, the last day of uniform irrigation, to 50 days after seeding when the eight stress measurement methods were compared. Data represent changes at four locations on the differential irrigation gradient.