The effect of drought stress on the morphology of Anthephora pubescens Nees

The effect of drought stress on the morphology of Anthephora pubescens Nees was examined at three phenological stages, i.e. during the vegetative stage (P1). at the onset of reproduction (P2) and during the late reproductive stage (P3). This was induced by withholding waler from the plants for 15 days. The morphology of A. pubescens was adversely affected by drought stress during its actively growing stage (P1) in that both leaf area and leaf length were reduced. However, it caused an increase in vegetative tiIIering of both P1 and P2 plants. Reproduction of A. pubescens was negatively influenced by drought stress, as reproductive tillering was reduced at both P1 and P2 stages, and the reproductive tillers formed at stage P3 were shorter.


Introduction
Anlhcphora pubescens Nees is a valuable species for dryland pastures in the semi-arid to arid regions of southern Africa bec ause of its dro ught-to lerance characteristics . Previous stud ies un dertaken on A. f'uhes ccns concentrated on establi shment, forage produ ct ion. fer ti lization requ ireme nts and nUlritiona l value (Donaldson el al. 1972;Rurger el at. 1975;Fou rie el al. 1984Fou rie el al. . 1987Du Pisani el al 1986;Dannhauser 199 1). Very little is known of the growth and development of this species under condit ions of drough t stress, and to increase the poten tial of A. p"bescclls as a pas turc species in arid regions, it is im pera tive to understand the response of this spcc ies to drought.
A1l1hepllora pubescells is classified as a short grass species because the internodes of the vegetative tillers arc not elongated (Baines 19H9). Short grasses, with a large number of till crs. are well adapted to regions with low, irregular rainfall, because it is advan tageous to di vide growth between mcristcms to ensure rapid regrowth after a period of drought stress. Ta ll grass species occur more frequently in areas with a more predictable rainfall where it is advant ageous to reduce the number of ti llers and to inv!.!st more in each individual liller.
The impact of drought stress on the morphology of a grass plant is greater than the effcct o n the physiology. and cell division appears to be kss sensitive to drought stress than cell enlargement (Turner & Begg 1978). The most im portan t effecls of drought s tres;.; arc therefore redu ced leaf area and decreased growth rate (Pande & Singh 1985;Rozijn & van der Werf 1986;Baruch 1994;Bussu & Richards 1995). Drought stress is also known to result in a decrease in total non-stru ctural carbohydrate and protein content (Pande & Singh 1985) as well as an in crease in prolin e and abscissic acid concentration (Frank 1994). Sensiti vity to drought stress also depends on the phenological or growth stage of the grass plant (A1cocer-Ruthling et al. 19H9;Siding et al. 1994). II has been reported that drought stress, Imposed al an early growth stage, red uces the nu mbe r of Vt.!geta-tJve til lers (Alcocer-Ruthling et al. 1989). If grass. pl an ts arc subjected to drough t stress bcCore the in itiation of the reproductive ph ase, the production of reproductive tillers m ay be either delayed or acceleratcd and the number of reprod uctive ti llers produ ced m ay also be less th an for non-stressed plants (Rozijn & van der Werr 1986;Alcoccr-Ruthling er al. 1989). Althnugh the lo tal growth of a grass plant m ay decrease duri ng drought stress, the negative effect on root growth is usually less than that on till er growth (Regg & Turner 1976;Turner & Begg 1978).
The aim of this study was to determin e the effects of drought stress, applied at different phenological s tages, on the morphOlogy of A. pllbescens. with special reference to the number of tillers (vegetative and reprodu ctive). leaf length and leaf area. Understanding plant growth and morphological deve lopment of A pubescens under drought stress may assist in the development of management practices aimed at the in creased su rvi val and forage production of this species when grown in dry land pastures.

Methods
The experiment was conducted under controUcd environmental conditions at the Range and Forage Insti tute from Decem ber 1990 to June 1991. The mean minimum and maximum temperatures for that period were 18°C and 30 0 e res pecti vely and re lative hum idi ty ranged from 41 to 58%.
Caryopsis of Anthephora pubescens, ccotype VH20. were obtained fr om the Biesievlaktc Research Station near Vryburg in the Northern Ca pe (24°28'E; 25°57'S) . Eight caryopses were sown per pot and germination took place afte r 4 days. Afte r 4 weeks the secd~ lings were thinned to one per pol. The pots. with a volumc of 5500 cm) (300 mm deep) were filled with a 15-mm layer of gravel and topped with a sandy-loam soil (82.8% sand, 8.7% loam. 8.5% cby, and a pH of 5.3) to a mass of 7.95 kg. The amount of water held by the soil in the pot at field water capacity was determined gravimetricall y (G raven 1968). This value was termed 'potwater capacity'. Every second day the pots were weighed and the amount of waler needed to obtain a mass, corresponding to 85% of pot water capacity. was added. Corrections werc made for the fresh ma ss of the plant mate rial in the pOL Control plants rece ived watcr every second day for the duration of the experiment. All plants received UAN 32 (ureum ammon iu m nitra te: 10 ml UAN 32/5000 ml H 2 0) once, early in their vegetative stage. The experimen tal layout was a random ized block design with four replicates of each of [lve treatments (bl ocks) and four harvest times per trcalmenl. Pots wcre rotated fortnightly to avoid the effect s of uneve n temperatures and light condi· tions within the gree nhouse.
The effect of drought stress on the ph enology of A. pubesceltS was examined al lhree phenological stages i.e. vegetative SL:'1ge (6 weeks after germination: PI), onse t of reproductio n (8 wceks after germination: P2) and late reproductive stage (11 weeks after germination: P3 ). Drought stress was imposed by withholding watcr from the plants fllr 15 days. The kng th of tlus pen ud was predetcrmllled and ta ken as the tim\! needed to reduce the soil wa te r content to 3.3% (m/ m) , Willdl represenh!d a soil water poteilltal of -I SOD kPa (permanen t wdting pom t) .
(2) num ber o f live leaves per marked tiller, and (3) hltaJ leaf length or the living part of the leaves pt!r marked Polyn omial [uncllons were fitled for the data pOints collected over time, i.c. number of tlllers per plan t and the to tal leaf length per marked tiller. The method of Groc neveJ dt (1970) was used to determine the confidence limits (P < 0.05; P < 0.0 1) fo r the linear combinations o f contrasts among polynomials. When only two treatments were compared. the Student's (·test (Sakal & Rohlf 1982) was: used La determine statis tically significant differences at a signiftcance level of a = 0.05 and a = 0.01.
Analysis o f variance (ANOVA) was done to compare morc than two treatments. Bart le tt's test was used to determine whether the variance was homogeneous. and Bonferroni's test was used to determine Sk'1tlstically s ign ific ant differences at a level of a == 0.05 (Sakal & Rohlf 1982).  Figure 3 The number of (a) vegetative, and (b) lo tal number of ti llers per pl ant, of con trol plan ts (P3C) and plants that were water stressed at the la te reproductive stage (P3 ).

Number of till ers
S.-AfrTydsk r.Planlk . 1996.62 ( 1) stressed plants was significantly less (P < 0.01) than that of control plants ( Figure Ib). After 40 days recovery, stressed plants overcame this back log. For plants stressed at stage PI , from 4 days afte r drought stress was induced un ti l 13 days into the re covery period, the total number of tillers was significantly less than that of control plants ( Figure Ic) . After a 27-day rec overy period the total number of tillers of s tressed plants (PI) was sign ifica ntly greater (P < 0.0 1) than that of control plants, and after 68 days this diffcrence between treatments was no longer apparent. The number of vegetative tillers of plants s. tressed at stage P2 did not differ significantly (P > 0.01) from that of control plants during the drought stress period (Figure 2a. c). However, after a 6-day post-s tress recovery period the number o f vegeta ti ve tillers on stressed plants (P2) was sign ifi cantly greater (P < 0.0 1) than that of con trol plants. This difference disappeared after a 41 -day recovery period. The total number of tillers followed a similar trend. Six days after drought stress was induced, the number of reprodu c tive tillers on stressed plants was signifi cantly less than th at of control plants (Figure 2b). Stressed plan ts were only able to overcome this backlog after a recovery period of 22 days.
At stage P3 drought s tress did not affect vegetative (Figure 3a

Total leaf length and leaf area
During the firs t six days afte r drou ght stress was induced. no signifi cant difference in the total leaf len gth betwee n plan ts stressed at stage PI and con tro l pla nts was noted (shaded areas , Fig ure   4a). However, after 6 days of drough t stress the total leaf length of stressed plants was sign ificantl y less than that of con trol plants. This difference was main tai ned during the post-stress rec ove ry period . In the case of plants stressed at stage P2, the lol a l leaf lengt h did not differ sign ifica ntly from that of the contro l plan ts (sh aded area, Figure 4b) during the first three days of the d rought s tress peri od. However. from the fou rth to the eight h day of the drought stress period, the total leaf length of the stressed plants was significan tl y greater than that of the control plants. After 9 days of stress the total leaf length of the stressed plants decreased and for the rest of the monitoring period there we re no sig nifican t differences in to tal h:af length between stre ssed and control plants.
After an 8-day post-stress recovery peri od, o nly plants stressed at stage PI had a significantly lower (P < 0.05) leaf area than control plants ( Figure 5) while the difference between leaf areas of stressed and non-stressed pl an ts was not signifi cant at stages P2 and P3 ( Figure 5).

Height of the flag leaf, length of the peduncle and inflorescence axis
Only at st age P3 were there en ough reproducti ve tillers to measure these param e ters. There was no s igni fican t difference (P > 0.01) in the height of the nag leaf between stressed and control plan ts (Tabl e 1). H owever, the length of the pedu ncle and the inf1 0rcscencc axis of the stressed plants (P3) were significan tl y lower (P < 0 .(1) than those of the control plan ts.

Discussion
Drough t stress imposed a t the vegetative stage (Pt ) of A. pube.fcens in itially caused a significant decrease in Ihe number of vegctative as well as the to tal number of tillers. A reductio n in ti lleri ng wi th a decrease in soil moisture availabilit y was also reported by Turn er & Degg (1978) and Alcocer-Rut hlin g et al.
(1989) for o ther g rass species. Upo n rewa te rin g P I-stressed Leng th of the in florescence axis (mm) 92.7 (7.9)*'* 107.5 (15.(,) Anthephora pubescem plants, stimulated tHlering res ulted in s ignificantly more vegetative tillers on stressed plllnts. Similar results were obtained in Agropyron smithii (Turn er & Begg 1978), w ith tillering bein g stimulated by a decrease in soi l moisture availability. However, tow ards the end of the observation period there were no sign ifican t differences in vegetative and total tiller num bers on stressed and con trol plants of A. pulJescens. The effec t of drought stress at stage P2 (onset of reproductio n) fo llowed a similar pattern, except thaI there was no initial decrease in tiller production. Reproductive ti ll er production was inh ibi ted by drou ght stress in bot h stages P I and P2. bu t no t in stage P3 (latc reproductive stage). Similarly. Alcoccr-Ruthl ing el aL (1989) found tha t drought stress imposed duri ng the flowering stage did not affect the reproductive capacity of Boute/oua scorpiodes. In a study o n the grai n yield of cere als. B aruch (1994) fou nd that the yield decre ased when droug ht stress was applied at a later phenological st age. Total le af len gt h per tiller in A. pubescens was less when the plants were stressed durin g the vegetative (PI ) stage. According to Turner & Begg (1978). the reduction in leaf e longa tion caused by drought stress is main ly due to the inhibition of cell enl a rge~ menL The elongation of leaves of P2~s tressed plants ceased 7 days afte r the onset of drough t st ress (leaf water potenti al < -1.2 mPa). Leaf elongation of Panicum maximum ceased at a leaf water po ten tia l of -1.0 mPa (Turner & IJ egg 1978). Reductions in green lea f number, rate of leaf initiation. height and total leaf area were reported for two tuft g rasse s in Utah, when leaf wa ter potentiat fe ll below -2.5 mPa (B usso & Richards 1995). During pot trials carried out on Themeda lriarldra and Sporoboius flmbriallls. Dan ckwerts ( 1988) found that the green leaf lengt h per tiller increased steadily while water was abundan t, reach in g a maximum at 40% water depletion, thereafter dege ner~ ating rapidly.
Drought stress in the vegetative stage (P I) of A. pubescens, reduced leaf area by 58%. However, drought stress induced either at the onset of reprod uction (P2) or later during the reproductive stage (P3) did not red uce leaf area significantly. In co ntrast, leaf area of Doule/aua scorpiodes plants (Alc ocer~Ruthling et af. 1989) as well as maize (Denmead & Shaw 1960) was reduced by drought stress at all phenological stages.
According to Denm ead & Shaw ( 1960), tiller elo ngation and the length of the inflorescence axis of maize was lowered by drought stress induced only at the vege tative stage, sugges ting that drought stress could o nly influence plant growth during the ac tively growing vegetative period of a plant. However, in the case of A. pubescens, ped uncle length as well as infloresce nce axis length were lower all P3-stressed plants than 011 control plants .
Arllhepora pubescens is a valuable cultivated pasture in the scmi~arid regions of South Africa. Although it is no t a high~ yiclding pasture in terms of kilograms dry matter per hec tare, it b vcry palatable and nutri tious (Fair 1989;Dannhauscr 1991). lIigh animal performance on A. pubescens pasture. (herdorc, large ly offsets the low dry-matter yield (Fair 19851;Dannhauser 1991) . In contrast to many other pasturc grasses. the palatability and nu tritional value of A. pubescem do not ded inc in aulumn and it can therefore provide good-quality grazing in autumn and winter (Fair 19R!J). Resu lls of this study indicatc that A. pllhescens is sensit ive to drought stress in its actively growing s.tages, i.e. the vegetative stage (Pl) and during the initiation of reproduction (P2), as reproductive ti llcring, tPla1 leaf lenglh and kaf area a[e reduced. Rcproductivt:: tillers produced o n stressed plants :I.[e shorter due to a [eduction in the length o f the peduncle and mflorescence axis. I Inwcver, provided water stress is not continuous.A. puhes-('ens pla nts have the ahil ity to compensate fnr the effect of drought strl.!ss by increased tii!cring. On the basis of thesl: resu lts it seems adv isable llnt to usc A. {Juh escens that has heen subj ec ted to drought stress early in the season, it should rather he used as autumn and winter fodde r.