A precise gravimetric method for simulating drought stress in pot experiments.
In the present work, a greenhouse null-balance lysimeter [an adaptation of the Hunter and Tonks (1979) technique] was evaluated as a means of simulating drought stress. Single plant water use was quantified gravimetrically by frequent, automated recording of pot weights, and transpired water was replaced by a computer controlled watering system such that actual soil water contents were maintained within a narrow, predetermined range. Potential advantages of this approach are (i) the rate at which the soil water deficit develops can be uncoupled from the transpiration rates of individual plants, so that soil dry-down of all experimental units can occur in a coordinated fashion, (ii) the dry-down rate can be preprogrammed and precisely controlled, so long as it is slower than what would occur with no addition of water, (iii) individual pots can be maintained at any chosen soil water content indefinitely, and (iv) automated recording of pot weights allows whole plant transpiration rates to be calculated over any desired time interval, thus providing a quantitative indicator of the plant response to the stress. There were two main objectives.
1. To characterize the distribution of soil water within pots as they were maintained at predetermined weights using the null-balance lysimeter. To our knowledge, replacement of small amounts of transpired water on a very frequent basis has not been previously tested with respect to its effect on spatial heterogeneity of soil water content in pots.
2. To determine if the system could be used to maintain single plants at several statistically distinct levels of water stress, as indicated by significantly different rates of whole plant transpiration.
Materials and Methods
Plant Culture Systems
Cotton (cv. ST474) and soybean (cv. Young) were grown at Athens, GA (34[degrees]N, 84[degrees]W), under greenhouse culture systems previously developed for each. All plants were grown in 2.5-L plastic food containers with plastic lids (Berry Plastics Corp., Evansville, IN) with interior dimensions of 15.5, 12.5, and 16 cm (top diameter, bottom diameter, and height, respectively). Each replication required four pots, but six were prepared so that the most uniform four plants could be selected. Soybean plants were grown essentially as described by Mian et al. (1998). The soil was a Pacolet sandy loam (clayey, kaolinitic, thermic, typic Hapludults) amended with sand to a texture of 800 g [kg.sup.-1] sand, 120 g [kg.sup.-1] silt, and 80 g [kg.sup.-1] clay, (final bulk density = 1.43 g [cm.sup.-3]). The containers lacked drainage holes. Cotton was grown in containers with drainage holes, in a Dothan loamy sand (fine-loamy, siliceous, thermic Kandiudults) whose texture was 910 g [kg.sup.-1] sand, 60 g [kg.sup.-1] silt, and 30 g [kg.sup.-1] clay (bulk density = 1.51 g [cm.sup.-3]). In both cases, 3500 g of soil was added to each pot. On each day that pots were prepared, an additional 3500 g of soil was placed in a forced air drier at 80[degrees]C for 1 wk, to estimate the oven-dry weight of soil that had been added to the pots (typically 3494-3496 g). This weight, plus the pot weight and lid weight was used as the "dry weight" value (WD). Seeds were sown four to a pot, and then thinned to one per pot after emergence. Replications were sown two at a time, 10 to 12 d apart, beginning 28 Feb. 2000 for soybean, and 24 Apr. 2000 for cotton. Pots containing soybean plants were weighed daily, and watered to 75% of soil saturation (see below for determination of soil water holding capacity). Pots containing cotton plants were given 100 mL of water per day when plants were young, and up to 250 mL daily as they grew. Plastic lids, each with a hole to accommodate the plant stem (with a slit from that hole to the outer edge of the lid) and another hole to permit addition of water, were applied to the soybean pots once cotyledons were fully expanded. Lids were not applied to the cotton pots until just before they were placed on the lysimeter. Soybean pots received 50 mL of a 0.8% (w/v) solution of fertilizer (20-20-20 plus micronutrients, Miller Greenhouse Special, Miller Chemical and Fertilizer Co. Corp., Hanover, PA) at planting, and again on the day that lids were applied. Cotton pots received 100 mL of a 0.25% (w/v) solution of the same fertilizer once per week. Greenhouse temperatures were maintained at 27 [+ or -] 4[degrees]C during the day and 20 [+ or -] 2[degrees]C during the night. Photoperiod was extended to 16 h with overhead 400-W metal halide lamps that produced a supplemental photosynthetic photon flux density (PPFD) of approximately 230 [micro]mol [m.sup.-2] [s.sup.-1] at the tops of the plants. Pots were placed on the lysimeter 35 to 40 d after sowing.
Determining Soil Water Holding Capacity
For the soybean experiment, one additional pot (with drainage holes) per replication was also filled with 3500 g of soil, and then watered to excess. The lid was applied to prevent evaporation from the soil surface, and the pot was allowed to drain until it reached a constant weight. The weight of water-saturated soil, plus pot and lid was used as the "wet weight" value ([W.sub.W]). For the cotton experiment, [W.sub.W] was not determined until the day before plants were transferred to the lysimeter. One extra pot per replication was watered to excess after removing the plant shoot, then the lid was applied and the pot was allowed to drain until it reached constant weight. Roots were then washed and their fresh weight determined, so that this weight could be subtracted from the total for calculation of [W.sub.W].
Lysimeter Design and Operation
The gravimetric lysimeter consisted of eight electronic balances with 6-kg maximum capacity and 0.1-g readability (Model XL-6100, Denver Instruments, Arvada, CO), connected to a personal computer via RS-232 cables and an 8-port RS-232 interface card (Model OMG-COMM8, Omega Engineering, Stamford, CT). The computer also contained an isolated digital input board that actuated eight Form C mechanical relays (Model CIO-RELAY16, Omega). Each relay in turn operated a normally closed two-way solenoid valve (Model SV-1201, Omega), each of which controlled water flow from a 100-L reservoir to one of the eight pots on the lysimeter balances. When a solenoid was actuated by the computer, water was conducted by gravity flow from the reservoir to the watering hole in the lid of the appropriate pot, via a length of 3-mm internal diameter vinyl tubing. The computer ran custom control software that was designed to read the weights from each balance approximately every 2 s, and then actuate the appropriate solenoid valve to replace transpired water if any pot weight had fallen by some predetermined amount (the threshold weight) below the target weight for that balance. The control software included logic to reject anomalous data caused by occasional faulty RS-232 communication, balance malfunction, tampering, etc., and posted error alerts whenever such events occurred.
When a run of the experiment was begun, the values of [W.sub.D] and [W.sub.W] were entered into the computer, along with the desired threshold weight. For these experiments, a threshold of 30 g was used, as this was near the minimum amount of water that would spread across the entire soil surface when added at the top of a pot. For each individual balance, values were then entered for the desired maximum relative soil water content (RSWC, a percentage between 0 an 100), and an estimate of the total plant fresh weight ([W.sub.P]), determined by measuring shoot and root weights of the two extra pots for that replication. The target weight ([W.sub.T]) for each balance was calculated by the software as:
 [W.sub.T] = [W.sub.D] + [W.sub.P] + RSWC ([W.sub.W] - [W.sub.D]).
New target RSWC values for the balances were entered manually into the computer during the course of an experiment as required. Pot weights were recorded automatically in a data file every 10 min. Whenever water was added to a pot, the amount added was also recorded in the file.
Experimental Design, Data Collection, and Analyses
Both the soybean and cotton experiments were randomized complete block designs with six replications and four treatments. Since the lysimeter had eight units it was possible to run two replications simultaneously. The treatments were four different final target RSWC levels of 75, 45, 30, and 20%. At the beginning of each run of the experiment, eight pots with plants of similar size were randomly assigned to the lysimeter balances, and the four treatments were randomly assigned to the balances within each of the two groups of four. The pots were placed on the balances at approximately 0900 h, calculation factors were entered into the computer, and the target RSWC was set to 75% for all balances. The balance assigned to the 75% RSWC treatment within each replication was maintained at this setting throughout the experiment. Settings for the other three balances in each replication were reduced to 45% RSWC 48 h after the beginning of the experiment. The settings for two of these balances were reduced to 30% 24 h later. Of these two balances, the setting for one was reduced to 20% after an additional 24 h. These final settings were maintained for another 48 h, during which time data were collected for calculation of whole plant water use (Fig. 1).
[FIGURE 1 OMITTED]
The normalized transpiration ratio (NTR) for each plant during Period A in Fig. 1 was calculated according to Ray and Sinclair (1997, 1998). That is, whole-plant transpiration during Period A ([T.sub.A]) was calculated for each plant (target RSWC = 75% for all pots), as the difference between beginning and ending pot weights, plus the total water added during the period. This was also calculated for Period B ([T.sub.B]), and then the transpiration ratio for each pot was calculated as TR = [T.sub.B]/[T.sub.A]. Finally, NTR for the 45, 30, and 20% RSWC treatments within a replication was calculated by dividing TR for each pot by the value of TR for the 75% RSWC pot in that replication.
Leaf gas exchange measurements and determinations of soil water distribution were conducted between 1100 and 1500 h on the day following Period B. Pots were removed one at a time for this purpose; pots remaining on the lysimeter continued to have their RSWC levels monitored and adjusted until they were removed. First, net C[O.sub.2] assimilation rate ([A.sub.N]), and stomatal conductance to water vapor ([g.sub.s]) were measured with an LI-6400 Photosynthesis Measurement System (LI-COR, Inc., Lincoln, NE). Measurements were made on youngest fully expanded mainstem leaves. The 6 [cm.sup.2] of leaf surface in the chamber received a PPFD of 1200 [micro]mol [m.sup.-2] [s.sup.-1] provided by a combination of red and blue light emitting diodes. Leaf temperature was maintained at 27 [+ or -] 0.3[degrees]C, by the chamber's thermoelectric coolers. Ambient air from within the greenhouse was passed through a 4-L buffer volume, and scrubbed of C[O.sub.2] by passing it through a column of soda lime. The C[O.sub.2] concentration was then adjusted to 400 [micro]L [L.sup.-1] (ppm) using the system's C[O.sub.2] injector. Airflow through the leaf chamber was maintained at 500 [micro]mol [s.sup.-1]. Once placed in the chamber, a leaf was allowed sufficient acclimation time (typically about 15 min) for [g.sub.s] and [A.sub.N] to reach near-steady state, as ascertained by monitoring gas exchange parameters, before logging the data point.
Immediately following gas exchange measurements, shoots were removed and their fresh weights determined. Then, four soil cores were extracted from the pot using a 1.2-cm internal diameter metal soil probe. The first core was taken about 1 cm from the pot wall, directly under the point where water had been supplied to the pot while on the lysimeter. The second core was extracted from the analogous position on the opposite side of the pot, very near the location of the plant stem. A third core was removed from the center of the pot, and a final one from against the pot wall, as far as possible from the other three cores. Each core was removed in four sections of equal length (that is, from four different depths in the pot), and each section was placed in a coin envelope and weighed. Finally, the roots were washed and weighed. Core sections were dried in an 80[degrees]C oven to constant weight, and then reweighed to calculate the original soil water content. The RSWC was then calculated for each core section, from the water saturation values determined previously for that soil.
Replication and treatment effects for all measured parameters were estimated by means of the ANOVA procedure in SAS (SAS Institute, 2001), and means comparisons between treatments were via protected LSD tests. Appropriate error terms were used for hypothesis testing and means separation where subsampling occurred (i.e., soil core RSWC measurements). A malfunction that affected two of the balances in the same replication occurred during the second run of the cotton experiment; all data from this replication were subsequently discarded, and an additional run (two replications) was performed, for a final total of n = 7.
Typical raw data from the lysimeter for a single replication of the soybean experiment are shown in Fig. 1. Except for the single malfunction mentioned in the materials and methods, the system was effective at maintaining the pots within a weight range of 30 g, equivalent to a range of approximately 5% RSWC. In these experiments, new target pot weights were always achieved within 24 h of adjusting the target RSWC. The actual mean RSWC maintained during Period B was always lower than the target value (Table 1), since weights were allowed to drop by 30 g below the target before rewatering. Also, actual values of [W.sub.P] determined destructively at the end of a replication were often higher than the estimates entered into the lysimeter computer at the beginning of the run. Final actual Wp for each pot was used for the calculation of RSWC and NTR once the experiment was completed (Table 1). The values of NTR for the 75% RSWC pots were not included in the ANOVA, since these values are all exactly 1.0 by mathematical definition; thus, including them would have artificially decreased the error variance. Nevertheless, the mean NTR value for the 45% RSWC treatment differed from NTR for the 75% treatment (1.0) by more than the LSD value in both experiments. The 45, 30, and 20% RSWC treatments also differed significantly from one another. Therefore, the NTR values for all four RSWC treatments were statistically different from one another, even though this is not indicated directly in Table 1. For leaf gas exchange measurements, both [g.sub.s] and [A.sub.N] changed in the expected direction with decreasing RSWC (Table 1). However, only two statistically distinct levels could be detected for [A.sub.N] in both experiments, and for [g.sub.s] in the cotton experiment. The LSD test resulted in three statistical classes for [g.sub.s] in the soybean experiment.
Distribution of soil water within the pots for each treatment is shown in Fig. 2 and Fig. 3. Each treatment was analyzed separately, because of nonhomogeneity of variance between the treatments. No significant interactions between horizontal position and depth were detected (P > 0.2), and so only main effects are presented. In general, soil water was distributed quite uniformly within the pots. Within a treatment, differences in mean RSWC with horizontal position of the core were generally nonsignificant. The exception was the 30% RSWC treatment in both experiments, for which the core located directly under the watering hole had a higher RSWC than one (soybean) or all (cotton) of the other cores (Fig. 2). Vertical distribution of soil water was somewhat less uniform than horizontal distribution, with a significant trend toward slightly wetter soil at the tops of the pots in all treatments, and also wetter soil at the bottoms of the pots in the 75% RSWC treatment (Fig. 3). In the cotton experiment, the mean RSWC at the top of the pots in the 20% treatment was numerically higher (no statistical test made) than the mean RSWC at the bottom of the pots in the 30% treatment. This was the only case where there was any overlap in mean RSWC values between treatments. In general, vertical distribution of soil water was more uniform in the soybean experiment than in the cotton experiment.
[FIGURES 2-3 OMITTED]
The multibalance lysimeter and automated watering system were successful in maintaining pot weights within a narrow range, such that fluctuations in bulk RSWC could be limited to 5% or less. Visual observations made during root washing indicated that roots were well distributed in the pots, exploring most of the soil volume. This situation, in combination with uniform soil water distribution within the pots, should make bulk RSWC calculated from pot weights a reliable indicator of water deficit experienced by roots. While there was some systematic heterogeneity of soil water content within the pots, especially differences with depth in the cotton experiment, in general the distribution of soil water was quite uniform. Thus, it appears that frequent replacement of transpired water at the tops of pots need not result in large gradients of RSWC, at least for the two specific culture systems examined. In both cases the growth medium had a very high sand content, which may have enhanced the uniformity of the vertical soil water distribution within the pots. Adding the water in 30-g amounts probably helped to achieve the horizontal uniformity of soil water, since this amount of water was sufficient to ensure that the entire soil surface was flooded at each watering event.
An advantage of using relatively small pots in combination with frequent replacement of transpired water is that there is great flexibility with respect to the rate at which stress treatments are imposed. In the present work, drought stress was allowed to develop quite quickly in the 20% RSWC treatment (RSWC declined from 75-20% in just 3 d). However, pots could be maintained within any 30-g weight range indefinitely, and so more gradual soil dry-down could easily be simulated with this system, with close synchronization between experimental units if so desired. In practice, the mean RSWC levels achieved were always substantially lower than the target values (Table 1). This was primarily caused by the structure of the watering subroutine, which attempted to maintain pots between the target weight and a weight 30 g lower than the target. The software has since been altered to trigger watering when a pot weight declines by 50% of the threshold weight below the target value. Pots are then watered up to the target weight plus 50% of the threshold, resulting in a mean weight over time that is very near the target value, given accurate estimates of [W.sub.P].
Even the 20% RSWC treatment used in these experiments could be considered a relatively mild stress. Even though whole plant water use was reduced by over 50% relative to the 75% RSWC treatment, AN was reduced only 15% (soybean) or 21% (cotton) (Table 1). This system has subsequently been used to apply more severe stresses, reducing [A.sub.N] of cotton by 90% relative to control plants (Earl and Hufstetler, unpublished).
The NTR appeared to be a very sensitive indicator of water deficit stress (Table 1). This was expected, since NTR integrates water use over a long period of time (24 h in this case) for an entire plant. By contrast, leaf gas exchange measurements are subject to error variance caused by temporal variation (leaf activity at the time of the measurement may not have been representative of activity over the whole day), and spatial variation (the single leaf sampled may not have been representative of the entire shoot). Even so, mean values of [g.sub.s] were well correlated with NTR values. For instance, the 20% RSWC treatment in the soybean experiment produced a mean NTR value of 0.42; the ratio of [g.sub.s] values between the 20 and 75% treatments was 0.44. Linear regression of mean NTR on mean [g.sub.s] values gave an [r.sup.2] of 0.99 for the soybean experiment, and 0.91 for the cotton experiment (not shown). It should be noted that in addition to [g.sub.s], whole plant transpiration as quantified by NTR might also be affected by treatment effects on leaf area development. That is, decreases in NTR with RSWC would arise not only from reduced transpiration per unit leaf area, but also a reduction in the rate of leaf area expansion. It is not known if this effect contributed significantly to observed responses of NTR to RSWC in the present work, but the data presented in Table 1 suggest that differences in g constituted the major effect.
Response of whole plant water use to soil water availability has often been quantified with respect to fraction of transpirable soil water
(FTSW), rather than RSWC (e.g., Sinclair and Ludlow, 1986; Ray and Sinclair, 1997). The FTSW is the current soil water content divided by the difference between the saturated water content and the water content when the plant can no longer extract any additional water. Because of the very low rate of plant water use at low soil water contents, in practice it is difficult to determine experimentally the soil water content at which plant available water is completely exhausted. Instead, this point is sometimes arbitrarily defined as the soil water content where NTR drops below a particular value. For example, Sinclair and Ludlow (1986) calculated FTSW defining FTSW to be zero when NTR dropped below 0.1. Subsequent experimentation with the culture systems used in the present work has shown that NTR reaches 0.1 when RSWC is approximately 0.075 for the soybean system and 0.109 for the cotton system (unpublished data). Thus, RSWC in the present work can be converted to FTSW as defined by Sinclair and Ludlow (1986) by first subtracting 0.075, then dividing the difference by 0.925 for soybean; for cotton, 0.109 is subtracted, and the difference is divided by 0.891.
In conclusion, gravimetric lysimetry combined with frequent, automated replacement of transpired water is a useful approach to simulating drought stress under greenhouse conditions. Stable levels of soil water deficit, with uniform water distribution throughout the rooting volume, can be maintained over time with this technique given the appropriate plant culture system. Bulk RSWC determined gravimetrically is a valid indicator of soil water deficit experienced by roots in such experiments, and NTR can serve as an unusually sensitive indicator of plant response to water stress. This method should prove advantageous when it is experimentally necessary to uncouple the rate at which drought stress develops from the rate of whole plant water use or from environmental conditions, or when maintenance of a stable level of soil water deficit is desired.
Abbreviations: [A.sub.N], leaf net C[O.sub.2] assimilation rate; [g.sub.s], stomatal conductance to water vapor; NTR, normalized transpiration ratio; PPFD, photosynthetic photon flux density: RSWC, relative soil water content.
Table 1. Effect of relative soil water content (RSWC) on the normalized transpiration ratio (NTR), stomatal conductance to water vapor ([g.sub.s]) and leaf net C[O.sub.2] assimilation rate ([A.sub.N]). Target maximum Actual mean RSWC RSWC NTR [g.sub.s] [A.sub.N] Soybean experiment (n = 6) % mol [micro]mol [m.sup.-2] [m.sup.-2] [s.sup-1] [s.sup-1] 75 70 (a) ([dagger]) (1) 0.50 (a) 19.5 (a) 45 40 (b) 0.86 (a) 0.43 (a) 19.8 (a) 30 26 (c) 0.71 (b) 0.33 (b) 18.1 (ab) 20 19 (d) 0.42 (c) 0.22 (c) 16.6 (b) LSD0.05 3 0.06 0.09 2.1 Cotton experiment (n = 7) % mol [micro]mol [m.sup.-2] [m.sup.-2] [s.sup-1] [s.sup-1] 75 68 (a) (1) 0.43 (a) 19.2 (a) 45 39 (b) 0.91 (a) 0.35 (ab) 18.8 (a) 30 25 (c) 0.82 (b) 0.28 (ab) 17.0 (ab) 20 14 (d) 0.53 (c) 0.20 (b) 15.2 (b) LSD0.05 2 0.09 0.15 3.2 ([dagger]) Values followed by the same letter within a column and within an experiment are not significantly different at the P < 0.05 level, according to a protected LSD test.
Allen, L.H., Jr., R.R. Valle, J.W. Mishoe, and J.W. Jones. 1994. Soybean leaf gas exchange responses to carbon dioxide and water stress. Agron. J. 86:625-636.
Andrew, C.S., and J.L. Cowper. 1973. Automatic plant pot watering machine. Lab. Pract. 22:364-367.
Babu, R.C., M.S. Pathan, A. Blum, and H.T. Nguyen. 1999. Comparison of measurement methods of osmotic adjustment in rice cultivars. Crop Sci. 39:150-158.
Cornic, G., I. Papageorgiou, and G. Louason. 1987. Effect of a rapid and a slow drought cycle followed by rehydration on stomatal and non-stomatal components of leaf photosynthesis in Phaseolus vulgaris L. J. Plant Physiol. 126:309-318.
Ekanayake, I.J., S.K. De Datta, and P.L. Steponkus. 1993. Effect of water deficit stress on diffusive resistance, transpiration, and spikelet desiccation of rice (Oryza sativa L.). Ann. Bot. 72:73-80.
Farrant, J.M., K. Cooper, L.A. Kruger, and H.W. Sherwin. 1999. The effect of drying rate on the survival of three desiccation-tolerant angiosperm species. Ann. Bot. 84:371-379.
Hunter, M.N., and J.W. Tonks. 1979. Tilting auto-watering pot system (TAPS). Queensland J. Agric. Animal Sci. 36:1-7.
Kramer, P.J. 1983. Water relations of plants. Academic Press, New York.
Ludlow, M.M. 1987. Contribution of osmotic adjustment to the maintenance of photosynthesis during water stress, p. 161-168. In J. Biggens (ed.) Progress in photosynthesis research, Vol. 4. Martinus Nijhoff Publ., Dordrecht, the Netherlands.
Mian, M.A.R., D.A. Ashley, and H.R. Boerma, 1998. An additional QTL for water use efficiency in soybean. Crop Sci. 38:390-393.
Nissanka, S.P., M.A. Dixon, and M. Tollenaar. 1997. Canopy gas exchange response to moisture stress in old and new maize hybrid. Crop Sci. 37:172-181.
Pennypacker, B.W., K.T. Leath, W.L. Stout, and R.R. Hill, Jr. 1990. Technique for simulating field drought stress in the greenhouse. Agron. J. 82:951-957.
Ray, J.D., and T.R. Sinclair. 1997. Stomatal closure of maize hybrids in response to drying soil. Crop Sci. 37:803-807.
Ray, J.D., and T.R. Sinclair. 1998. The effect of pot size on growth and transpiration of maize and soybean during water deficit stress. J. Exp, Bot. 49:1381-1386.
Saccardy, K., G. Cornic, J. Brulfert, and A. Reyss. 1996. Effect of drought stress on net C[O.sub.2] uptake by Zea leaves. Planta 199:589-595.
SAS Institute. 2001. The SAS System for Windows, Release 8.02. SAS Institute Inc., Cary, NC.
Schwaegerle, K.E. 1983. A method for maintaining constant soil moisture availability for plants. Soil Sci. Soc. Am. J. 47:608-610.
Sinclair, T.R., and M.M. Ludlow. 1986. Influence of soil water supply on the plant water balance of four tropical grain legumes. Aust. J. Plant Physiol. 13:329-341.
Turner, N.C. 1997. Further progress in crop water relations. Adv. Agron. 58:293-338.
HUGH J. EARL *
Dep. of Crop and Soil Sciences, Univ. of Georgia, Athens, GA 30602-7272. This research was funded by state and Hatch funds allocated to the Georgia Agric. Exp. Stn., and by grants from the Univ. of Georgia Office of the Vice President for Res. and the Georgia Agric. Commodity Commission for Cotton. Received 8 Aug. 2002. * Corresponding author (firstname.lastname@example.org).
|Printer friendly Cite/link Email Feedback|
|Author:||Earl, Hugh J.|
|Date:||Sep 1, 2003|
|Previous Article:||Genetic diversity of soybean cultivars from China, Japan, North America, and North American ancestral lines determined by amplified fragment length...|
|Next Article:||Comparison of thermal units derived from dally and hourly temperatures.|