The effect of drought stress on the dry matter production, growth rate and biomass allocation of Anthephora pubescens Nees

The effect of drought stress on the dry matter production, growth rate and biomass allocation of Anthephora pubescens Nees was examined at three phenological stages, Le. vegetative stage (P1), initiation of reproduction (P2) and late reproductive stage (P3). Relative to control plants, total dry mass of P1 , P2 and P3 plants, harvested after an 8-day post-stress recovery period, was reduced by 78%, 60% and 35% respectively. However, at the end of the growing season, there were no differences (P> 0.05) in total dry matter accumulation between stressed and non stressed plants. This recovery of A. pubescens can possibly be explained by the phenotypic plasticity in leaf allocation in stages P1 and P2, and by the increase in the leaf area ratio 01 P1 and P2 plants, following drought stress.

In geneml. water delicits reduce plant production ( Grnwth analysis is the first step in tht'! analysis of primary producti on and is a useful tool to eX,illline the dfect of different cIl Vi r0111l1Cntai conditions un primary production (K"ct el af. 197 1; Causlo n & Venus 19S 1; lIunt 1<JH2, 1990). Biomass allocation refers to the proportion of total biomass stored in cacb organ (Harper & Ogden 1970). The biomass allocation pattcrn adopted by a species is determined by its genotype (Fenner 1985) and is reli ned through the process of natural selection to improve its c hances of survival (Barbour el al. 1980;Gross el al. 19X3;Bazzaz & Reekie 1985 ;Reekie & Banaz 1987).
The ,lim of this study was to examine th e effect of drought stress, imposed at different pht.mological stages, on the dry matter produc tion, growth rate an d biomass allocation pattern of A. pubescens.

Met hods
The expenment was cllnducted in a greenhouse at the Range and rorage InstItute from December 1990 to June 199 1. The mean minimum and maximum temperatures in the greenhouse for tbat period were 18°C and 30°C res pect ively and relative humid ity ranged from 41 % to 58%. A de tailed description of the seed so urce, so Ll type and general exp cnmcnL."111ayout is provided in Moolman et aI. (19%).
Plants of Ilnthephora pubescens, eco type VH20, were grown from seed in pOLS with a volume of 5 500 em] , ftlled wtth a IS-mm layer of g r~\lcl and topped with a sand y loam soil. Plan ts were grown at a density of one per pot and four replicates were used per treatment , The amount of water held by the soi l in the pot at field capacity (termed 'pal wate r capacity') was determined gravimetflcally (Graven 1(68). Pots were weighed every second day and the amount of water needed to obt,lin a mass cnrrespomling to 85 % of pot water capacIty was added.
Drought stress was imposed at three pheno logic al stages, i.e. PI : vegetative stage (6 weeks after germ inati on) ; P2: init iatio n of reproduction (8 weeks after germination); and P3 : late reproductive stage (11 weeks after germination). Two additional droughtstress treatments werc applied 2 weeks after germination onwards. i.e. 2 weeks drought stress, alternated with 1.5 weeks of water ing (AI ) and two weeks drou ght stress, alternated with 2.5 weeks of waterin g (A2).
Control plants received water every second day for thl;! duration of the experiment. Drought stress was mduced by withholding water from plants for 15 days. The leng th of this pe riod was prede termmed as the time needed to reduce the soil water potential 10 -I 500 kPa (permanent wilt ing point), at which stage the soi l water content was 3.3% (mlm ).
Fou r plants of treatments PI, P2 and P3, and their respec tive controls were harvested j ust prior to thc water-stress period b ein~ implemented, after an 8-day post-stress recovery period, as weI. liS al 18 weeks and 21 weeks after germi nat ion. Plants of the A i and A2 treatments were harvested o nly at 18 and 21 weeks afte[ germination. At each harvest, the plant ma terial was divided into laminae, tillers (including the leaf sheat h), inflorescences (including the peduncle), and roots. Leaf area was measured with a I.i Cor 3100 Jeaf area meter (Lienr, Lincoln, Nebraska 68504, USA).
The dry mass of each plant component was used to de termine dry matter production of the whole plant and the biomass allocation patterns of tiJJers, laminac, reproduct ive structures a nd roots. The bi omass allocation of a specific plant component was obtai ned by expressing its dry mass as a percen tage of the plant 's lotal dry mass. Relative growth rates (R) and leaf area ratios (LAR) wcre also dctcnnined. Relative growtb rates of the indi¥ vidual plant components were a lso calculated, i. c. that of the tillers (R t ) , leaf laminae (R]) and the roots (Rr)' Thc fomtulae used for the calcu1atj olls were based on those of Kvet el al. (197 1), Cousto n & Venus ( 198 I), Coombs el al. (1985) and Hunt ( 1990).
An analysis of variance (ANOYA) was don c by means of the

Dry matter production
The total dry mass of Pl. P2 and P3 plants wa~ less (I' < 0, (5) than that of their respective contro ls ( Figure I). A t stage P I. the dry mass of the tillers and laminae of stressed plants was less (P < 0.01) than that o f control plants. whereas there was no significant diffe rence (P> 0.05) in rool d ry mass. At stages P2 and P3, these relationships were reversed (Figu re 1, caption). The dry mass or the reprod uctive components of the plants stressed at all three pheno log ical stages ( Figure 1) was lower (P < 0.05) than that of control plants.
When drou gh t stress was applied during the vegetative stage (PI) of A. pubescens, the stressed plant's total dry mass was decreased by as much as 78%, whereas those stressed at the onset of reproduct io n (P2) s howed a redu ction of 60%, and those stressed in their late reproductive stage (P3) showed a reduction of only 3570.
At 18 weeks after germination. the total dry mass o f on ly the P3 and Al stressed plants was still significantly lower than that of control plants ( Figure 2). However. the dry mass of the repro· § , E -~,   Planlk. 1996, 62( 1) ductive structures of PI , P3, Al and A2 drought-stressed plants was lower than that of control plants ( Figure 2). In non e o f the treatments cou ld a difference in the dry mass of the roots be found . At 21 weeks after gemlination there were no s ignifica nt differ· ences in the total dry mass productio n between any o f the drought-stress treatments and that of the con traI ( Figure 3). How· ever, the reproductive dry mass of the plants which were repea t· edly drou ght stressed (AI and A2) was lower (P < 0.05) than thaI of control plants (Pigure 3).

Relative growth rate
The relati ve growth rate (R) of plants stressc!d at stage PI (-0.0022 g g.1 day· l) and P2 (0.0445 g g.1 day·l) was significantly lower (P < 0.05) than that of their respective controls (PI C = 0.0643 g g.1 day-I; P2C;;;; 0.0815 g g-1 day-I), whe reas there was no Significant difference in R between strl.!ssed and unstressed plants at stage P3 ( Table 1) . The redUction in R of stressed plants at stage PI was due to reductions in the rda tive growth rates of the lillers, laminae and roots. At stage Pl. only the relative growth ratc o f the roots ( Table 1) was reduced (P < 0'()5) com· pared with the control.

Leaf area ra tio (LAR)
The leaf area ratio o f a plant characterises the relativc size of its assimil atory organs, and therefore it expresses the efficiency of the plant as. a producer of lea f area (Coombs el al. 1985). It can be used to indicate differences between plants on the basis of genetic fa ctors. the environment , or d ifferent treatments ( Kvet et ai. 1971). The leaf area ratio of PI (56 .3 crn 2 g.J ) and P2 plants (47.6 cm 2 g.l) was higher than that of their corresponding can· tro ts (P IC = 29.6 em' g.l; P2C = 30.8 em' g"), whereas Ihe LAR of P3 plants was not affected by drought stress (Figu re 4). Therc-

Biomass allocation
Biomass allocation of plan ts from wh ich wa lcr was wi thheld at the thre e phenological stages is given in Figure 5. Drought stress, induced at stages PI and P2, resu lted in a significant increase in leaf allocat ion. Although leaf allocation of these droughtstressed plants was higher th an that of control plants, the amount of avail able matcrial was less on the stressed individuals. For  However, at 18 weeks after gennination, rool allocation o f P I, Al and 1\2 droughH;tressed plants was significan ll y higher than that of control plants ( Figure 6). This increase in roo t allocation in stressed plants was accompanied by a decrease in reproductive allocation ( Figure 6). However, 2 1 weeks after germination, there were no significant differences in biomass allocation between any of the stressed and comrol plants (Figure 7).

Discussion
Drought stress generally reduces plant growth (Pande & Singh 1985;Rozijn & van der.Werf 1986;Alcoeer-Ruthling el al. 1989;Baruch 1994). The effect of drought stress on the production and biomass alloca tion patterns of Anthephora pubescens depends on stress recovery period, the total dry mass of PI, P2 and P3 plants was reduced by 78%, 60% and 35% respectively, relative to their controls. !-Iowever. at the end of the growing season. to tal dry matter production of stressed plants was not significan tl y lower th,m thal of con trol plants. Bus~o & Ric hards (1995) al so found that herbage accumu lation. by two tussock grasses in U tah , at the end of the season was no t reduced after a sing le season's drought stress . Drought stress durin g the vegetative stage or early reproductive stage o f A. pubescens is the most detrim e ntal in terms o f the plant's short-term production and growth rate . Drought stress did not affe ct the growth rate of plan ts if applied during the late reproduc ti ve stage.
The recovery o f A. pllbescens late in the growing season fo llowing earl ier drought stress is possibly because A. pubescens displays phenotypic plasticity in its biomass allocation. Leaf allocation was stimulated by drough t stress, induced at the vegetative stage o r during the onset of reproducti on. Inc reased investment in leaves led to an increased LAR, allowing the plant to com pensate for losses caused by drought stress. For examp le, the increase in dry m atter of con trol pl ant s between 18 ( Figure 2) and 21 weeks (Figure 3) after germination was low (3 g), when com pared to the increase in dry matter o f s tressed pla nts. especially that of P3 plants (21 g).
An increase in root allocation with an increase in moisture stress has been reported by vario us au thors (Gales 1979). In A.
pubescens, root allocation showed no immediate reac tion to drought stress . Onl y at the harves t at 18 weeks was a si gn ifi can t increase in root all oca ti o n noticeable in some of the treatments.
According to Tu r ner & Begg ( 1978), the cell numbers of plants exposed to water defici ts are of lhe same general o rder of magni tu de as control plants, although the cells are of a smaller The biomass allocation, 18 weeks after gcrminat ton. of plan ts subjected to different water-stress treatments (C: c{)nt rol, 1'1: stressed at the vegetative stage, P2: stressed at the onset of reproductio n, P3: stressed at the late reproductive slage; AI: 2 weeks water stress alternated with 1 j weeks of watering; and /\2: two weeks wa ler stress alternated with 2.5 weeks of watering.

Biomass allocation
Reproductive allocat ion

Root al location
Above-ground allocation LSD T . ,,,, ( P < 0.05) 20.4 11.5 11.5 size. Fu rthermore. plan ts that arc frequ en tl y water stressed exhibit more rapid growth whe n recovering from drought s tress. The sensitivity of cell e nlargement to water deficits also results in a red uction in the leaf area of a plant. In A. pilhescens, the e longation of leaves is rdarded duri ng drough t stress (Moolman 1993; Moolm an el al. 1996), but when the plan ts recover they produce thin ner le aves, which is demonstrated by the increase in the leaf area rat io and spec ifi c leaf area (Manlman et al. 1996) of es pc~ da ll y PI plan ts. T he reproduc tion of A. pubescens is negatively affected by drought stress, The dry mass of the reproduc tive st ructures decreased hy 93%, 90% and 63% follow in g stress induce d at stages P I, P2 and P3 respectively. This reduction could be the res ult of {he delay in the reprodu ction phase, as well as the abor~ tion of newly famled reproductive ti llers . However. on ly the dry mass of the reproductive structures of plants that were subjected to altern ati ng drought sIres!> throughout the growing season (AI and A2 ), was still signifi can tl y less th an that of control plants at the end of the growing season. The reaction of A. pub escens to drought stress d iffers fro m that of BOUle/oua scorpiodes as Aicoccr-Rulhiing et ai. (19R9) concl uded that the reproductive dry matt er production of H. scorpiodes was not in fluenced by drought stress in any of the phenological stages.

Conclusions
Althou g h drought stress ca used a signifi cant reduction in dry mattcr production in the short term, stressed plants apparenlly have the ability to compensate for this loss o ver thc entire growth season. In general, it can be con cluded that A. pubescens is m o re vu lnerable to drou ght-s tress con d itions in its vegetative stage, and to a lesser degree at the o nset of reproduction. Although drought stress negatively affects the produ ction of A. pubescens in the short tenn, only the reproductive dry ma tter produc ti on of plants that were stressed thro ughout the growing season was permanently hanned. This suggests that A. pubescens should not be used as su mmer-grazing pastu res in areas wh ic h are subjected to drought stress in Ihe growing seaso n. However. A. pubescens ca n be utili sed as fo rage in winter, as the effect of drought stress o n produ c tion is negligable.