Photosynthetic response to water stress in Themeda triandra and Eragrostis lehmanniana

The influence of water stress on the photosynthetic rate of the C 4 grasses Themeda triandra Forsk. and Eragrostis lehmanniana Nees was determined for the vegetative and reproductive phases. Gas exchange was determined with an infra-red gas analyser, while leaf water potential was used to quantify water stress. An open system of gas flow was used. The rate of photosynthesis was 0.3131 mg CO 2 m- 2 S-1 and 0.6287 mg CO 2 m- 2 S-1 respectively for T triandra and E. lehmanniana (P::;; 0.001) if water was not limiting. The rate of photosynthesis began to decline at a leaf water potential of -1 927 kPa and -1 625 kPa for T. triandra and E. lehmanniana respectively. This decline in rate of photosynthesis was significantly (p:s; 0.001) correlated with leaf water potential, and linear relationships with correla tion coefficients of 0.936 and 0.938 were obtained for T. triandra and E. lehmanniana respectively. whom correspondence addressed.


Introduction
South Africa is predominantly a dry country with two thirds of the land rl!ceiving less than 500 mm of rain per annum (Sc hulze 1979). In these arid and semi-arid regions, feed production is lim ited as a result of a shortage of avai lab le wate r and suitable soil while animal production is largely dependent on the condition of the veld (Snyman & Fouche 1993). In stock-farming areas, the veld is frequently subject to seasonal and extreme droug hts, leadin g to instability in farming (Sny man & Fouche:: 1991 ). Pre-and mid-summer droughts arc a normal phenomenon in semi-arid regions (Snyman 1993).
To measure the productivity of pasture plants under favourable conditio ns. it is necessary to determine the water use e fficiency under optimal conditions. It is well known that the rates of photosynthesis and plant production decline under water stress (Ashton 1956: Stalfclt 1956: Moss et aI. 1961: Hsiao 1973: Boyer 1968: Bussa & Richards 1995, but this has not bee n quantified for most pasture plants, and parameters to identi fy water stress quickly and accurately justify further investigation. According to Hsiao (1973). Turner (1981) and Venter (1988), leaf water potential is the plant characteristic most used to describe plant water status. If the leaf water potential can accurately be determined where the rate of photosynthesis and transpiration in grasses begins to decline, it can be used to identify the actual stress point. The accuracy of this parameter will depend on how well plant c haracteristics can be correlated with leaf water potential (Snyman et a Venter 1988). Sensitivity to drought stress also depends on the phenological or growth stage of the grass plant (Maul man et aI. 1996aaI. & b: Sieling et al. 1994.
The purpose of thi s study was to determine the rate of photosynthesis of Th emeda trinlldm and Eragrostis iehmalll1ialla in the vegetative and reproductive phases under optimal soil water and water-stress conditions. The pathway of photosynthetic CO 2 ass imilari on in both grao;;s species is via C-, metabolism. Leaf water potential was investigated as a parameter to identify and quantify water stress in grasses. The quantification of the reaction o f grasses to different water conditions can be used to explain t.:hanges in plant grow th and refine existing mathematical simulation model s. By using such a model. the probability of feed shortages at the end of the growing season can he deter-mined and stock numhers adapted in good ti me. T herefore for arid and semi-arid regions, quantitative determination of the influences of water stress is a necessity.

Study area
The study was conducted in ashl!stos pots (540 mm deep wi th a diameter of 210 mm) in a greenhouse under controlled climatologieal conditions. The pots were painted beforehand to t.:ounter a poss ible increase in pH of the soi l :.IS a resull of the asbestos. A 14-h photoperiod with day and night temperatures respectively of 30-32°C and 18°C was applied. The relative humidity ranged from 41 % to 58%.
Soil of the Shorrocks Series (Hutton Form) (Macvicar ct at. 1977) or. accord ing to the new so il classification system (Soil Classification Working Group 199 I), a fin e sandy loam soil of the Bloemdal Form (Roodepla,t family -3 200), sampled to a depth of 600 mm, was used as the growth medium. The first horizon (A: 0-200 mm) cO lllained 10.6% and 2.7%. and the seco nd horizon (82: 200-600 mm) 19.0% and 5. 1 % clay and silt respectively. The respective bulk densities for each horizon were 1 484 and 1 563 kg m·'.
Themeda triandra Forsk. and Emgro!itis Jelimmmiww Nces plants were obtained from Sydenham, the expe rimental farm of the University of the Orange Free State, 5 km west of Blol.!mfontein (29°06'S. 26°57'E; I 350 m.a.s.I). and is situaled in a semiarid summer-rainfall region (annual mean 560 mm, 55% of which falls during (he period January to April). In the central grassveld region, Tlzemeda triandra (decreaser species: Foran et a /. 1978) is the dominant grass species of veld which is in a good condition, while Eragrostis lelllnmltliwUl (increaser species: Fourie & Visagie 1985) dominates vdd in a moderate condition (Snyman & Fouche 1993). The experimenl was conducted on T.

Methods Treatments
The plan ts were allowl.!d lO establish well in the greenhouse bI::fore samples of T. Irialldm and E. lehmmlllicma were randomly divided into control plants and those to he subjected to water stress. Plants wc[e randomly allocated to cuvcttes in the same greenhouse fOf the determination uf rate o f gas exchange. The vegetative phase was first investigated, followed by the reproductive pha se . Six replicates per treatment wen: used. After allocaling a plant to a certain cuvette. it was alw,lYs placed under the same cuve tte for th e duration of thc experiment to excl ude the possihl e dfect of cuvette variation. The control plams were kept above wilting point as far as possible by regu lar watering. The amount of water hdd by thc soi l in the pot at field cap,tcilY was determi ned gravimetrically (Graven 1968). POlS we re weighed l!vcry second day and Ihc amount o f water needed to obtain a mass corresponding to R5% of field capacity was added.
The plants suhjected to water stress were not watered until the h!af wilter potential decreased to less than -6 000 kPa. The leaf water pntenlial of randomly selected leaves from every tuft was determined tiailv with a Scholander pressu re ho mb (Scholander ela/. 1965) after the ·plants began to show stress symptoms (Snyman el al. 1987). Care was taken that the length of leaf protruding ahove the ruhber sl.;ul was shorter than 20 mm (Wari ng & Cleary 1967) in order to minimize transpiration of that part of the leaf above the pressure homb and to cxd ude unnecess.lry variation. The pressure <It which water was ohserveti at (he top of the vascular bundles was taken as the potential of the water just hdore the leaf was removed (Waring & Cleary 1967). The leaf water potential was d!.!tcrmined immedi· atel), after measuring gas exchange.

Relationship between linear measurements and real area
The relationship between linear measuremen ts and real area of leaves and stems of other T triandra and E. leJmulllniana plants was estahlished on plants grown under the same conditions as those sub· jecled to water stress. Thl! ll!ngth and total breadth of leaves and · stems wl!re ch;: taminl!ti. as di scussed by K vet el 01. (197 1). and these values of every leaf and stem were multiplied to ca lculate surface area, after which the leaves and ste ms were removed and the area determin!.!d with a Licor LJ :3 000 planimeter. A linear regression between ca1cubted area and the determined area was applied . The mass of kaves and stems was also determined , after which the relationsh ip between real area anti mass was calculated in the same way. T hese relati ons hi ps were determined fo r hOlh the vegetat ive and reproductive growth stagl!s.

Plotting a growth curve by means of regreSS ion relationships
Two to three part s of every grass tuft were separated from the rest of the tu ft with a wife marker in order to ensure sampling within a rea· son able time. To monitor the growth of the plane, the length and breadth of all green leaves. vegetative and reprodu ctive stems in the marked plan! parts were measured and noted separately at regular intervals for the duration or (he ex periment. Afte r the experiment.
the whole phlllt was cut off and dried at 70 0 e fo r a period of 24 h.
The mass of the marked part s was determined separately.
The real areas of leaves. vegetat ive and reproductive stems respec· tively in the marked plant parts were muhiplied with the relati on hetween the total above·ground dry mass and the dry mass of lhe marked plant part s. This was done to extrapolate the rcal areas of the respective plant parts to tht: lotal plant. The ratio between mass of the tota l ahove-ground part of the plant and the marked plant parts was used after obtaining good relationships between dry mass and real determined areas. It was assumed that the measured plant part in relation to its cnd ma<;s increased the same as the whole plant. if the ~alllp le taken in the beginning was representative of the whole plant.
In the caSl! of pl:lnts subjected to water ~trcss. the leaf area was only dl!terminl!d until visible wilting: occurred. Wi th T. triandra, this was done as the leaves began to close, and with E. lehmalllJialJfl. when th!.!y hegan to curl up (the turgor pressure in the leaves dec reases dUl! to low leaf water potential). At this point, the leaf area cannot be accurately detcrminl!d as lhe regression equations arc no longer valid. S. Arr.J.llo!. 1997.6.1(1) Lay-out of the system for determination of gas exchange An open system of air fl ow was used (Jarvis & Catsky 1971). The six cuvclIes cons isted of a cyli ndrica l framework of aluminium (hei ght 1.5 m and diameter 500 mm ) cove red by polyethylene lilm. Inside the cuvelle. the air was evenly distributed over the plant with an electric fan. An air sample was takl!n berore entering th e cuvette and at the outlet of each of the six cuvettcs. The CO 2 concentration and vapou r pressu re of both sLlmp les were nnalysed by an ADC 22) MK II infra·red gas analyser accord ing 10 the absolute method for every cuvette (CLltsky et 01.1971). Temperature was controlled hy a separate system of air now (Vente r 1988) and maintained in the cuvettes at 32-3fi°C, which i~ thl! Dptimal range for growth or CJ plants (Lawlor 1976).
The soil around the pl ants in the pOlS. in which the f::ue of gas exchange was measured. was ~ea lc.d with plastic film. This was pasted to the bottom of the cuvette once the pot was placed in the cuvette. in order to ohtain an airtight seal. An adaptation period of 45 min was all owed after placing the plants in the cuvettes. and measurements of gas exchange were done on the total above-ground part of the plants.

Determination of the rate of photosynthesis
The equation, described by Ca tsky et af . (1971) for the determina· lion of CO 2 exchange, and adapted by Venter (1988). was u~cd in this study. Measurements to determine rate of gas exchange were only taken in full sun light so that light intensi ty would not be a limiting factor. A Licor quantum light sensor was used . Inside the cllvettes, values of I 100-1 500 ) . .unol m· 2 S-I were found. The rate of air flow was controlled in such a way that the change in CO 2 com· pensation point with water ~tress would not affect the measurement (Meidner 1967 ;Shearman el al. 1972;Lawlor 1976).
In the case of plants subjected to wal~r stress, the rate o f gas exchange was expressed a~ mg CO 2 plant·! S-I and not per m l . which is justified as the leaf area will not increase after permanent wilting (Cleland 1959: Probine & Preston 1962: Boyer 1968: Kramer 1969: Green el al. 1971Sharp el al. 1979). Measu rements were always taken between II :00 and 12:00. so that the angle of incidence of the sun on the plants would not di ffer for the duration of the experiment.

Processing of data
After calculating the rates of photosynthesis, the data was processed to characterize the react ion of every species to water stress. In calcu· lating the mean rate of photosynthesis, the read ing o f the cont rol plants was used together with that of the rest of the plants before stress. The leaf water potential, at which a continual decrease in rate of photosyt:lthesis was observed per plant. was taken as the onset of water stress regarding rate of photosynthes is. To characterize the decrease in rate of photosyn thes is with increasing water stress. the rate of gas exchange per plant and the corresponding leaf water potential were used. After withdrawing water from the plants subject to water stress, the rate of photosynthesis of the total plant increased initially. These highest values of photosynthesis. just before a steady decrease in both the rate of gas exchange and leaf water potential occurred. were taken as the reference values for each plant. These rates of gas exchange with increasing wate r stress were then expressed as a percentage of the reference values. The relative values of photosynthesis obtained in this way were related to the nbsolutc values of the corresponding leaf water potentials and expressed ns a relationship per species. The plants were co mpared with each other according to the analys is of variance technique for a complete ran· domized design. The F·tcst was used for the comparisons.

Relationships between plant measurements and area
The calc ulated areas of the stems and leaves of both species were related CO the real determined areas by rnc!ans of a linear regres· s. Afr.J. Bot. 1997,63(1) sian (r = a + hx) (Table I) where), is the real area and hOlh (/ and h the constants, w ith x thl! calculated area in mm 2 •

Rate of photosynthesis
Erngmsfis fehmallll imw developed very quickly from vegetative to reproductive phase and therefore rates of photosynthes is or only the reproductive phase were determined. Themeda tr;alldra plants dill'crl.!d very little in the vegetative and flowering stages (1' > 0.05) regarding rales of photosynthesis. fn Themeda Irian· dra the rate for both vegetative and reproductive phases was nearly half (p $ 0.01) that of E. lehmann;olla in the reproductive phase if water was not limiting ( Table 2). The lear wutl!r potell-tiaL where the tirst signs of a decrease in rate of photosynthesis was observed. was -I 927 kPa in Themeda triandra and -I 625 kPa in E. lehmannilllUl (p> 0.05). T. triafldm plants in the vegc~ talive and flowering stages did not differ signifi cantly (p > 0.05) regarding leaf water potential when the first signs of a decrease in rate of photosynt hes is were observed.
The linear relati onship (P 5 0.01)' between the absolute leaf wate r potential values and the relative rate of phoLOsyn thesis for the range 01'-1350 kPa LO -5850 kPa for T. triandm and E. [eli -mallllialla are presented in Figure 1. As mu ch as 88% of the va ri~ ation in photosynthetic rates for both species may be attributed to differences in leaf water potential. Emgrostis lehm{/l/lliana dif~ fered signifi cantly (p 5 0.01) from T. triandra regarding rate of photosynthesis with increasing water stress ( Figure I).

Discussion
The rate of photosynthesis of T. trialldra and E. lehmalltJialla (1987), Polley et al. (1992) and Bamch and Fernandez (1993) in explaining diffen!I1ces in rate of photosynthesis between species where hoth species follow the same photosyn~ thetic pathway.
The leaf water potential. whe re the rate of pholOsynthesis of T.
fri(lndra and E. lehllIllIlIlifllw decreases, did not differe s i gn iri~ canlly. Beadle el (II. ( 1973) and Mel7 lehnulIIllilllw. where the rale of photosynthesis decreases. was respectively -I 927 kPa and -I 625 kPa in our sludy, but Snyman et nl. (l9R7) found Ihat these Iwo species first show visible signs of wilting at -2 450 kPa and -2 050 kPa respectively. The process of photosynthesis therefore undergoes water stress before the plants show visible signs of wilting.
Below the threshold value of lotal leaf water potential in T triandra and E. lehmalilliana the rate of photosynthesis decreased  linearly with dec reasing leaf wate r potential, whi ch is suppo rted by Policy et al. (1992) Comstock and Ehlcrin ger (1984), McCree and Richardson ( 1987). Vos and Oyarzun ( 1987) and Nic et al. (1992) also clearly showed that [here is a linear relationship between the rafe of photosynthesis and leaf water potential below a critical leaf water potential in Zea mays, Sorghum vulgare. Eflcelia Jmtes-('('liS, Bern vulgaris. Viglla rm&llicll/ata, Solaman ruheroswll and Alldmp o~oll ~erarclii.
In our study, the regression line of the relative rate of photosynthesis on water potential was purposely not lengthened to determine the leaf wate r potential where photosynthesis ceased. The work of Melzack et al. (1985) and Joh nson et al. (1987) shows that the decrease in rate of photosynthesis can suddenly level out at a very low value of photosynthesis and then maintain a constan t value. Th is tendency could not be d~termined in T. tri~ llIulm and £. lehlll(ulIliww as the Scholande r pressure bomb which we used could not measure such a low leaf water potential.

Conclusion
The beg inn ing of water stress mu st preferably be defined as the leaf water potential where the rate of photosynthesis begins to show a decrease. Soil water conlent is a misleading parameter to identify plant water stress as it does not accurately integrate both soi l waler conte nt and atmosp heric evaporation demand , whereas leaf water potential does (Snyman 1993). Plant water potential is a more sensitive indication of water stress than other parameters and should be used more in irri galion scheduling of cultivated pastures. A further advantage of leaf water potential as a p a rame~ tet in iden ti fying plant water st ress is the ease and speed of deter~ mination in the tield. We found that by measuring leaf water potential. it is possible to predict rdative rates of photosynthesis without determ inin g gas exc hange or plams.
Too few research results arc available 10 d irectly re late leaf water pOIential during water stress to production under veld con~ ditions. Quantitative n!sults must first be obtained on the innucnce of water stress on processes such as respiration (under~ ground and above g rou nd). ce ll division, cdl enlargement. pro~ tein synthesis. carbohydrate metabolism, and lear die-back, before accurate predictions of production under water stress can be done.
Tht: physiological imp lementation of vdd manageme nt in arid and sem i~arid regions is diflicult as knowll.!dge is scant of the influence of the intensity and duration of water stress on the carbohydrate status of the plant (Venter 1988. Busso et al. J 990), the distrihution of carbohydrates in the plant during wate r st ress (Snyman 1993), the influence or intensity of uti lization by animals on the recovery of the pastun! plant after a drough t (Danck~ werts & Aucamp 1985; Danckwerts & Stuart-Hill 1988;Snyman & van Rensburg 1990: Sny man 1993. and the response of photosynthesis to drought (Senock et al. 1991;Baruch 1994). Thl.! results obtained in this study can be used to refine sOllnd mathematical simulation models and to increase the predictive val ue of these in times of drought.