Plant and Soil229: 225–234, 2001.© 2001Kluwer Academic Publishers. Printed in the Netherlands.

225

Drought survival in Dactylis glomerataand Festuca arundinaceaundersimilar rooting conditions in tubes

F. Volaire1 & F. LelievreLaboratoire d’Ecophysiologie des Plantes sous Stress Environnementaux, INRA (Institut National de RechercheAgronomique), 2 Place Viala, 34060 Montpellier, Cedex 1, France.1Corresponding author∗

Received 9 November 1999. Accepted in revised form 3 October 2000

Key words:cocksfoot, dehydration tolerance, drought resistance, fescue, membrane stability, soil water potential

Abstract

Drought survival in perennial forage plants involves different adaptative responses such as delay of dehydrationthrough water uptake, limitation of water loss and tolerance of tissues to dessication. To compare the importance ofthese responses in contrasting cultivars of forage grasses at the whole plant level, we carried out two experimentsunder glasshouse conditions. Plants of cocksfoot (Dactylis glomerataL.) cultivars, cvs. Currie, Medly (both ofMediterranean origin) and Lutetia (of continental origin), and of tall fescue (Festuca arundinaceaL.) cv. Centurion(Mediterranean) were grown in 60 cm-deep cylinders to eliminate the effect of differences of root depth on wateravailability whilst allowing severe drought to be imposed at a realistic rate. In both experiments, the cvs. wereranked similarly for plant survival, with high mortality for Centurion, low for the Mediterranean cocksfoots Currieand Medly, and intermediate for Lutetia. These differences could not be ascribed to water use during most of thedrought period since water uptake and decrease in leaf extension were not significantly different between speciesand cultivars. However, resistant cvs. of cocksfoot were able to extract water for a longer period and at a lower soilwater potential (9s) than other cvs. The critical9s at plant death was−3.8 and−3.6 MPa for Medly and Currieand−3.0-,−2.6 MPa for Lutetia and Centurion. Moreover, at a low soil water reserve (15–2%), membrane stabilityand water content were maintained for longer in enclosed immature leaf bases of cocksfoots cultivars, whereas thefescue Centurion exhibited accelerated lamina senescence and steady increase of membrane damage in survivingtissues. Therefore, it is proposed that the drought resistance of tall fescue in the field can mainly be ascribed to itsability to develop a deep root system. In cocksfoot, dehydration tolerance in surviving tissues and the ability ofroots to extract water at low soil water potentials may, in addition to root depth, contribute significantly to plantsurvival under severe drought.

Introduction

How plants respond to drought is a subject of highinterest since water deficit is one of the major lim-iting factor to plant production in many parts of theworld (Boyer, 1982). The nature and intensity of waterdeficits can be very variable, but drought resistanceis usually assessed as the ability of plants and cropsto maintain a certain level of production under watershortage. This criterion is relevant for most species of

∗ FAX No: +46-75-22-116. TEL No: 49-96-12-666.E-mail: volaire@ensam.inra.fr

agronomic interest (mainly annuals such as cereals)subjected to periods of drought during their cycle ofproduction (Passioura, 1996). However, for other spe-cies, such as perennial forage plants grown in Medi-terranean and semiarid rain-fed areas where intensesummer droughts stop all production, the most relev-ant criterion of plant resistance is drought survival, i.e.the ability of plants to remain alive during summerand recover when rehydration occurs. Although themechanisms involved in maintenance of productionduring moderate drought have been analysed extens-ively in many species (Blum, 1996; Turner, 1997),those contributing to drought survival have received

226

less attention. In addition, the metabolic bases of pro-ductivity on the one hand and survival on the otherare likely to be very different and are difficult to com-pare (Jones et al., 1981). Therefore, it is necessaryto identify the physiological adaptative features mostinvolved in survival to prolonged drought and hencecontribute to the creation of varieties better adapted forpasture improvement and soil conservation in regionssubjected to long and intense summer water deficits.

A number of authors (Levitt, 1980; Ludlow, 1989;May and Milthorpe, 1962) have identified three maincategories of drought resistance: Drought escape,drought avoidance by delay of tissue dehydrationand tolerance of dehydration. In perennial temper-ate grasses, some species such asPoa bulbosaL.exhibit a drought escape strategy through a total sum-mer dormancy involving complete dehydration of theplants (Ofir, 1986). However, most common species(Dactylis glomerata, Lolium perenne), although qui-escent (Villers, 1975) during the dry summer (nogrowth and most aerial tissues senescent), are notreally dormant since they grow when fully irrigated(Volaire et al., 1998a). During drought, those plantscannot tolerate extreme dehydration and tend to avoidit by maintaining water uptake and limiting water loss.Previous experiments showed that delay of tissue de-hydration through root development and water uptake,was higher in tolerant cvs. of cocksfoot (Volaire andThomas, 1995; Volaire et al., 1998a), as found inother major grasses (Carrow, 1996a; McWilliam andKramer, 1968). Two other factors appearing to con-tribute to survival are reduced transpiration throughlamina mortality and dehydration tolerance of mer-istematic tissues, which are the surviving tissues re-maining alive for the longest period (Volaire et al.,1998b). In particular, the membrane stability of thosetissues seems to be a critical component of droughttolerance (Blum and Ebercon, 1981) and this is asso-ciated with their ability to survive dehydration (Croweet al., 1987; Leopold et al., 1981). However, differ-ences between contrasting cvs. are difficult to analysein field experiments because of the overriding effectof rooting depth on water status (Volaire and Lelièvre,1997; Volaire et al., 1998b). Therefore, to eliminatethe effects of rooting pattern and depth, plants weregrown in containers of restricted depth (60 cm), anddrought started only when the soil volume was fullyexplored by the roots. We compared contrasting cvs.of cocksfoot and a variety of Mediterranean tall fes-cue known for its high drought survival (Lelièvre andDesplobins, 1994). The main objectives of this study

were: (i) to compare the cvs. for water extraction andcritical soil water potential at plant death, (ii) to ana-lyse the response of mature and immature aerial partsand, hence, (iii) to assess, at whole-plant level, the im-portance of adaptative mechanisms in forage grassessubjected to long and severe drought.

Materials and methods

Plant material

Three cvs. of cocksfoot (Dactylis glomerataL.) weregrown: cv Medly (syn. KM2), of Mediterranean origin(INRA Montpellier), cv Currie, of Australian ori-gin (only in one experiment, Exp97), both of whichare drought resistant in the field, and cv Lutetia, ofcontinental origin (INRA, Lusignan), a poor droughtsurviver. They were compared with tall fescue(Fes-tuca arundinaceaL.) cv Centurion (syn. LGM), ofMediterranean origin (INRA, Montpellier) and gooddrought surviver in field conditions.

Experimental design and conditions

Two experiments were carried out in a greenhouse atINRA Montpellier successively in summers–autumnsof 1997 and 1998 (Exp97, Exp98). In both cases,plants were sown in the previous autumn, grown attemperature over 20◦C in winter (to prevent ver-nalisation and subsequent flowering) and regularlydefoliated. In early June, plants were transplanted intoPVC tubes 60 cm deep× 5.5 cm diameter (2 and 3plants/tube respectively for Exp97 and Exp98) filledwith the same quantity of substrate (80% sand, 10%clay, 10% loam), fully irrigated every 2 days and fer-tilised for 3 weeks before starting the experiments. Atthe beginning of July, when roots of all cvs. reachedthe bottom of the tubes, irrigation was stopped for alltubes in Exp97 and for 70% of tubes in Exp98 (30%were maintained fully irrigated as control). Duringthe first 2 months of drought period (July and Au-gust), the mean vapour pressure deficit was 0.88 KPain Exp97 and 1.1 KPa in Exp98. The mean daily airtemperatures were 23◦C (min 19.5◦C max 27◦C)in Exp97 and 25.7◦C (min 21.9◦C, max 30.3◦C)in Exp98. Additional measurements of soil temper-ature in Exp98 showed that in drought conditions,at the depth of apices (−1 cm), the mean temper-ature was around 29.6◦C (min 27 ◦C, max 37◦C)with extremes reaching 42◦C. The temperatures re-corded in a former summer field experiment (Volaire

227

Figure 1. Soil water retention relationship obtained by the pres-sure-plate technique and fitted curve between soil water potential9s(MPa) and soil water content (%):9s = −104.66∗ exp (−1.439∗SW) – 0.00277.Means± SD (n=12). When no SD bar is shown, itis included in the width of the symbol.

et al., 1998a) showed similar daily means but higherday/night amplitude and extremes that could reach47 ◦C at apice depth. Therefore, thermal conditionsfor our greenhouse experiment can be regarded as lesssevere than those usually observed in Mediterraneanfield conditions.

Soil measurements

We weighed 10 tubes (Exp97) and 15 tubes (Exp98)per cv. two or three times a week and the mass of drysoil mixture was measured (at the end of the exper-iment) after drying at 80◦C for 72h. The mean soilwater content (SWM) was calculated for every tubeas a function of time. The lowest mean soil moisture(SWN) was recorded at the end of Exp97 when the lastsurviving plant of the last cv. alive died (SWN of 2.0%in cocksfoot Medly). The highest soil water content(SWX) was 10.0% after 48 h following full irriga-tion and end of water leaching (field capacity of thesubstrate). The Fraction of Soil Water reserve Avail-able for Plants at day n (FSWAP), was calculated as:[((SWM - SWN) / (SWX - SWN))∗100)]. In addition,as Exp97 aimed to measure the critical (lethal) soilwater potential related to plant death, soil water con-tent was measured gravimetrically at four soil depths(5, 15, 35 and 55 cm) once the last plant had died ineach tube. The relationship between soil water content(SW) and soil water potential (9s) was obtained atINRA Orleans by the pressure plate technique (Fig-ure 1). The values of the fitted curve between soil

water potential (MPa) and soil water content (%) are:

9 = −104.66∗exp(−1.44∗SW) − 0.003

Plant measurements

Leaf extensionIn both experiments, lamina length was measuredevery 2–3 d on the fastest growing leaves of 20 tillersper cv. until leaf growth ceased.

Plant death and survivalPlant death is defined as the ultimate stage when thelast living bud of a plant loses its aptitude to regrowwhen rehydrated. It occurs after all visible aerial partshave become senescent, when enclosed leaf bases pro-tecting the meristems turn yellow and stiff. In Exp97,no destructive measurement from the 10 tubes percv. was possible since water use during the droughtperiod and lethal9s had to be measured accurately.The percentage of senescent aerial tissue was assessedvisually (scale 0–100%) for each plant every 3–5 days.When it reached 95–97% on a plant, a prevision ofregrowth was assessed (scale from 4 to 0), based onthe external aspect (colour, texture) of the bases oftillers. When the last surviving plant in a tube reachedthe mark 0 on this scale (estimated plant death), soilwater content was immediately measured at differ-ent depths to estimate the lethal9s. The previsionof regrowth was validated in an additive experimentwith 200 plants of dactylis rehydrated after drought intubes, that showed a strong correlation between visualmarks and regrowth biomass, and an error of±5 dayson individual death date (corresponding to±0.12 MPafor lethal9s). In addition, the death of the plants wasproved by the observation of no subsequent regrowthof the plants carefully transplanted and rehydrated. InExp97, drought survival of the cvs was estimated asthe number of days to reach death for 75% of theplants. In Exp98, it was estimated as the percentageof plants recovering (aerial leaf growth observed visu-ally) after 3 weeks of full irrigation following eightweeks of drought in 150 plants per cv.

Biomass, water content and membrane stabilityThese parameters were studied in Exp98 by destruct-ive measurements, carried out by sampling two tubes(two replicates) per cv. every week for 8 weeks for thedroughted plants, and once at the end of the droughtfor the fully irrigated control. The six plants weredivided into two fractions (1) the first 20 mm above

228

root insertion (mainly sheaths and leaf bases) and (2)the remaining upper tissues (mainly mature lamina).This second fraction was divided into green and sen-escent tissues. Photocopies were made of green laminafor subsequent measurement of leaf area using imageanalyser software by Optimas. Fresh weight and dryweight (after 48 h at 80◦C) were measured on thesetissues to assess green and senescent biomass and drymatter concentration of green lamina.

In the fraction containing the lowest 20 mm partof the tillers, the enclosed bases of immature leaveswere dissected out and split into two subsamples. Onesubsample was immediately weighed and then dried(48 h at 80◦C) to determine water content. The othersubsample was used to measure membrane stability.The rate of injury to cell membranes by drought maybe estimated through measurement of electrolyte leak-age from the cells. Membrane damage was measuredas follows: After incubation in 4 ml of deionized waterat 20◦C in the dark for 22 h, conductivity (C1) wasmeasured with a seed analyser (G2000, Wavefront,Inc, Ann Arbor, USA). Samples were then boiled in4 ml deionized water for 1 h, cooled for 90 min andthe conductivity measured again (C2). According toHowarth et al. (1997), the coefficient of membranedamage (CMD) was calculated as:

CMD = (C1/(Cl+ C2)∗100.

Statistical analysis

The data were analysed using the appropriate ana-lysis of variance models in the SAS package. Figureswere drawn and curves fitted with the Fig. P. package(Biosoft, Cambridge, UK).

Results

Soil water extraction and plant survival rate

In both years, roots of all plants were allowed to reachthe bottom of the tubes before drought was imposed.The rate of soil moisture loss was not significantly dif-ferent (P>0.05) between cvs. whilst plants were stillalive (Figure 1). In Exp97, the summer was relativelycool so that plants were allowed to survive on storedsoil moisture for a long period (90–159 days), untillate in the autumn. Soil water content dropped from10% to below 3% in 30 to 38 days in summer andthen decreased slowly until plant death that occurred

at different dates for the cvs. The associated criticalthreshold of soil water content was significantly dif-ferent (P<0.001) between the cvs. (Table 1). The cvs.were ranked Centurion< Lutetia< Curie = Medlyfor both total water extraction and days of droughtsurvived. Because of surface evaporation, the upper15 cm soil horizon was significantly drier than lowerlayers, the lowest soil water potential achieved by rootextraction alone was calculated from the soil moisturecontent measured at the 15–60 cm on the death of thelast surviving plant. This potential was significantlydifferent between cvs. (Table 1). It ranged from−2.6MPa and−3 MPa for Centurion and Lutetia, to−3.6MPa and−3.8 MPa for Currie and Medly.

Due to higher evaporative conditions in the sum-mer of 1998 (higher mean air temperature and vapourpressure deficit than in summer 1997) and higher plantdensity in tubes, mean soil water content (SWM)dropped more quickly in Exp98, from 10% (field capa-city) to below 3% in 23 days. Just before rehydrationon day 54, SWM was 2.40% in Centurion and around2.25% in both cocksfoots. Following full rehydrationfor 21 days, plant survival ranked as follows: 14%for Centurion, 27% for Lutetia and 46% for Medly.The fraction of soil water reserve available for plants(FSWAP) was rapidly used to 85% in around 25 daysand then the last 15% to a very low rate, as in Exp97.

Mature lamina

Leaf extensionIn both summers, leaf extension decreased as soon aswater deficit was imposed (Figure 2) and then stoppedat a similar date for all cvs. (d32–d35 in Exp97 andd17–d20 in Exp98). Leaf elongation rate expressed asa percentage of initial rate, decreased linearly with thefraction of soil water reserve available for plants. Leafextension ceased when FSWAP was 15.3% (±3.8)corresponding to a mean soil water potential of around−0.2 MPa, not significantly different between cvs. andyears.

Biomass, leaf area and senescenceIn Exp98, the aerial green biomass of all cvs. (ex-pressed as a percentage of the initial value) was stableuntil FSWAP reached 15% and then declined sharply(Figure 4a). When FSWAP decreased below 20%, theleaf area declined significantly and was reduced earlierin fescue than in cocksfoots (Figure 4b). In Exp97,the fraction of green leaf area evolved similarly for alldactylis but fescue Centurion senesced slightly earlier

229

Figure 2. Decline in soil water content (%) with number of days without irrigation, in two experiments (1997: closed symbols; 1998: opensymbols) for cultivars ofDactylis glomerata(�,�: Lutetia; ,#: Medly;N: Currie (in 1997 only) andFestuca arundinacea(�,♦: Centurion).Means± SD (n=15). The insert shows mean soil water content between days 45 and 160, period covering the end of both experiments. For1997, means were calculated from tubes with at least one plant alive and represent data for a decreasing number of tubes along time.

Figure 3. Relationship between leaf elongation rate (as a proportion of initial rate) and the fraction of soil water reserve available for plants, intwo experiments (1997: closed symbols; 1998: open symbols) for cultivars ofDactylis glomerata(�, �: Lutetia; , #: Medly; N: Currie (in1997 only) andFestuca arundinacea(�, ♦: Centurion). Means± SD (n=15).

230

Table 1. Number of days without irrigation to reach death of 75% of plants, extreme soil moisture and lethal soilwater potential when the last plant died in each tube for one cultivar of tall fescue (Centurion) and three cultivars ofcocksfoot (Lutetia, Currie and Medly) subjected to 160 days of water deficit in summer/autumn 1997. Differencesbetween cultivars were significant atP<0.01 at 8 cm-depth and atP< 0.001 for other soil depths and means.a, b,c: means followed by different letters are significantly different in Duncan’s multiple range test.

Cultivar Days Soil water content at four soil depths Means Soil water

to reach death (%) potential

of 75% of plants (MPa)

Means

5 cm 15 cm 35 cm 55 cm 0–60 cm 15-60 cm 15–60 cm

Centurion 101 1.58a 2.49a 2.63a 2.60a 2.33a 2.58a −2.59a

Lutetia 141 1.46a 2.40ab 2.55a 2.52a 2.23b 2.49a −2.96a

Currie 155 1.49ab 2.28bc 2.40b 2.33b 2.13c 2.34b −3.65b

Medly 155 1.35b 2.23c 2.38b 2.31b 2.07c 2.30b −3.82b

Figure 4. Relationships between the fraction of soil water reserve available for plants (%) and in Experiment 1998 (open symbols): aerial greenbiomass, as a percentage of initial value (a); green leaf area as a percentage of initial value (b) and percentage of water content in lamina (c);in Experiment 1997 (closed symbols): the fraction of green biomass (d); for cultivars ofDactylis glomerata(�, �: Lutetia; , #: Medly; N:Currie (in 1997 only) andFestuca arundinacea(�, ♦: Centurion) subjected to, respectively, 54 and 160 days of water deficit in summer 1998and summer/autumn 1997.

231

Figure 5. Water content (a, b) and coefficient of membrane damage of immature leaf bases as a function of the number of days without irrigation(a, c) and of the fraction of soil water reserve available for plants (b, d) for cultivars ofDactylis glomerata(�: Lutetia;#: Medly); andFestucaarundinacea(♦: Centurion) subjected to 54 days of water deficit in controlled conditions (summer 1998). Data obtained on day 51 are not fittedin the models (a, c).

(Figure 4d). In Exp98, at similar FSWAP, the watercontent of lamina was lower in cocksfoot Medly (15%less than in Lutetia at FSWAP 20%). At low FSWAP,when leaf growth had stopped, the water content oflamina of Medly was stable (at around 38%) but thatof Lutetia and Centurion were more variable and gen-erally decreased more rapidly to reach the value ofsenescent tissue, below 35% for Centurion (Figure 4c).Fully irrigated plants had similar water content (andsimilar to initial values of 75%) after 8 weeks of fullirrigation. Their biomass and leaf area were also notsignificantly different between cvs. (data not shown).

Immature leaf bases (Exp98)

As drought increased, water content and membranedamage in immature leaf bases evolved similarly (Fig-

ure 5). For the fescue, water content and CMD inimmature leaf bases varied linearly with time and werelower (water content) and higher (CMD) than those ofdactylis after day 30, along with earlier mortality. Forboth dactylis, those parameters varied with time in asigmoïd way, with a significant variation between day10 and day 20, followed by a stabilisation around 40%of water content and a CMD of 40% between days 20and 44 (Figure 5 a, c). This stabilisation was observedwhen FSWAP fell from 20% to 5% (Figure 5b, d). Wa-ter content in bases of immature leaves of Medly wasconstantly lower (−5% on average) than this of Lutetiabut their CMD was similar. Around day 50, the tissuesremaining alive exhibited a sudden decrease of watercontent (20%) and increase of CMD (70%) along withincreasing mortality. In control plants, mean water

232

content and CMD of immature leaf bases were, re-spectively, 68% and 14% and were not significantlydifferent between cvs. (data not shown).

Discussion

The experiments we carried out in a temperature-controlled glasshouse aimed to mimic the progressivedrought that adult forage plants experience in Mediter-ranean summers. In both years, the water deficit lastedat least 8 weeks and was even prolonged in 1997 dueto low evaporative demand and plant density in thetubes. This allowed us to date the death of all plantsin autumn 1997, with a precision impossible when afaster progression of water deficit triggers plant mor-tality in a few days as in Exp98. All cvs. showed bothsimilar behaviour during the first part of the droughtand differential patterns of responses only at higherwater deficits.

From the end of irrigation to the end of leafelongation

While the fraction of soil water available for plantswas over 15%, water uptake was not significantly dif-ferent between the cvs. Hence, during this period,we can compare cvs. without the confounding effectof differences in soil water use or plant water status(Hanson and Hitz, 1982). In addition, leaf extensiondeclined as soon as drought was imposed and stoppedat similar dates and soil water potential for all speciesand cvs. In earlier work, no difference either betweencocksfoots and ryegrass was observed in the field, forthe response of leaf elongation to prolonged drought(Volaire et al., 1998a). As leaf growth decreased, leafarea declined, which is a common response to waterdeficits (Ludlow and Muchow, 1990) and essential ad-aptation for plant survival (Johnson and Assay, 1993).Over this period, leaf area of the cocksfoot cvs. de-clined at the same rate, but was reduced more rapidlyin tall fescue which rolled its leaves. This adaptativeresponse of fescue contributed neither to slow its wateruptake nor to delay dehydration of its surviving tis-sues. However, as the fescue tended to senesce earlieras well, a specific role to leaf rolling would be difficultto assess in these experiments.

From the end of leaf elongation to the plant death orsurvival

During this period, most aerial tissues became pro-gressively senescent, the water content of lamina de-creasing quickly below 50%, threshold proposed byHsiao (1973) as characteristic of the onset of a dessica-tion stage. An important consequence of this reductionin leaf area is the associated reduction in water lossobserved markedly for all cvs. in the second stage ofthe drought, thereby reducing the rate of water use anddelaying the onset of more severe stress (Turner andBegg, 1981).

Exp97 showed that under extreme drought, just be-fore the death of the last surviving plants, soil moistureand9s evolved asymptotically to a limit value, withlow variability within each cv. (confidence intervalsfrom ±0.14 to±0.23 MPa,P<0.05). Consequently,the various sources of this variability were also small.In particular, the error on death date determination, es-timated at±5 days, only added a limited error (±0.12MPa) on the determination of lethal9s because ofthe slow evolution of this parameter at the end of thedrought. The lethal soil water potential achieved by thetested cvs. ranged−2.6 to−3.8 Mpa and are muchlower than the−1.6 MPa threshold often related toplant permanent wilting point, although no sharplydefined lower limit for availability of water to plantscan be defined (Kramer and Boyer, 1995). On theother hand, our data are lower to those given by Rigaand Vartanian (1999) that found 28% of recovery forNicotiana tabacumsubjected to extreme levels of soildessication (9s lower than−5.3 MPa).

Comparison between populations

In limited and similar rooting conditions, cv. survivalranked similarly for both years. The tall fescue Cen-turion exhibited the earliest mortality in spite of itshigh resistance in the field. In comparison with cocks-foot, the leaf area of the fescue declined at higher soilmoisture and the hydration of its lamina and immatureleaf bases decreased constantly along with increasingmembrane damage in those tissues. In addition, asdeath of tall fescue was associated to higher soil wa-ter potential (9s =−2.6 MPa) than those of dactylis(−3.0-,−3.8 MPa), it is also arguable that either waterextraction was less efficient in fescue or that deathoccurred earlier both in its aerial and root systems.Therefore, tall fescue showed no strong expressionof a mechanism that would enable the plant to pro-tect the water status of critical tissues such as the

233

apical meristems (Bray, 1993). For this species, adeep and extensive rooting and associated water up-take have been emphasized as an important componentof drought resistance (Carrow 1996b; Ervin and Koski,1998; Qian et al., 1997). As shown in this study, thismajor drought avoidance mechanism which delays therate of dehydration is questioned as beneficial underroot limiting conditions (Bonos and Murphy, 1999;Qian et al., 1997) since the limitation of root growthin shallow soils can be cited as a causal factor ofdrought stress and resulting stand decline of tall fescue(Torbert et al., 1990).

In cocksfoot, survival was higher for the droughtresistant Medly and Currie than for the sensitive Lute-tia but to a lesser extent than in field conditions(Volaire, 1995; Volaire et al., 1998a). After leaf elong-ation ceased, membrane stability and water contentof surviving immature leaf bases were maintainedfor most of this second stage of the drought. Thisresult suggests a protection of meristems against de-hydration and some degree of membrane stabilisationthat could be involved in dehydration tolerance (Rigaand Vartanian, 1999). Although the drought sensit-ive cv. Lutetia maintained a higher water content ofits lamina and bases of immatures leaves than thecv. Medly, the membrane stability of those tissueswas similar in both cvs. and was not enhanced withprogressive drought as found in other species, suchas Triticum aestivum(Blum and Ebercon, 1981) orSorghum bicolor(Premachandra et al., 1992). Ourresults also show significant differences (0.8 MPa) inlethal soil water potential between cvs. of cocksfoot.This difference may enable resistant plants to extractfrom 5 to 15 mm of additional soil water reserve atthe end of the drought in the field, therefore delayingmortality for 10–30 days (when evaporation is only0.1–0.2 mm/d). Therefore, the differences betweencvs. of cocksfoot, which appear relatively small in ab-solute terms, might have significant consequences forsurvival of prolonged drought. In the field, a high effi-ciency for water uptake, a deep root system (Volaire etal., 1998a) and a high level of dehydration tolerance inimmature leaf bases may be combined factors contrib-uting to the striking drought survival of the resistantcocksfoots that can be used in environments with aslow as 425 mm rainfall (Reed, 1996).

Conclusions

We showed that differences for survival between cvs.of summer active forage grasses subjected to severedrought do not depend on water use during most of thedrought period when plants are grown in similar root-ing conditions. These differences depend mainly onthe ability to extract additional water at the end of thedrought when the soil water content is very low, and onthe maintenance of a critical threshold of water contentin surviving tissues, that can be due both to real dehyd-ration tolerance and/or protection of surviving tissueswithin dessicated sheaths reducing water loss (Blum,1996). We also showed that cocksfoot exhibited betterability for water uptake at very low soil water poten-tials and higher dehydration control than tall fescue. Itremains to investigate in surviving tissues of the bestgenotypes, the thresholds of dehydration tolerance andto analyse the contribution of membrane protection,osmotic adjustment and accumulation of specific pro-teins to the survival of perennial grasses submitted tosevere droughts.

Acknowledgements

Thanks to Pascal Chapon for technical support, andto Dr Harry Thomas (IGER, BBSRC, UK) for criticalreading of the manuscript and valuable suggestions.

References

Blum A 1996 Crop responses to drought and the interpretation ofadaptation. Plant Growth Regul. 20, 135–148.

Blum A and Ebercon A 1981 Cell membrane stability as a measureof drought and heat tolerance in wheat. Crop Sci. 21, 43–47.

Bonos S A and Murphy J A 1999 Growth responses and perform-ance of kentucky bluegrass under summer stress. Crop Sci. 39,770–774.

Boyer J S 1982 Plant productivity and environment. Science 218,443–448.

Bray E A 1993 Molecular responses to water deficit. Plant Physiol.103, 11–16.

Carrow R N 199a Drought resistance aspects of turfgrasses in theSoutheast: Root-shoot responses. Crop Sci. 36, 687–694.

Carrow R N 1996b Drought avoidance characteristics of diverse tallfescue cultivars. Crop Sci. 36, 371–377.

Crowe J H, Crowe L M, Carpenter J F and Aurell Wistrom C 1987Stabilization of dry phospholipid bilayers and proteins by sugars.Biochem. J. 242, 1–10.

Ervin E H and Koski A J 1998 Drought avoidance aspects and cropcoefficients of kentucky bluegrass and tall fescue turfs in thesemiarid west. Crop Sci. 38, 788–795.

Hanson A D and Hitz W D 1982 Metabolic responses of mesophytesto plant water deficit. Annu. Rev. Plant Physiol. 33, 163–203.

234

Hsiao T C 1973 Plant responses to water stress. Ann Rev. PlantPhysiol. 24, 519–70.

Howarth C J, Pollock C J and Peacock J M 1997 Developmentof laboratory-based methods for assessing seedling thermotol-erance in pearl millet. New Phytol. 137, 129–139.

Johnson D A and Assay K H 1993 Viewpoint:Selection for im-proved drought response in cool-season grasses. J. Range Man-age. 46, 194–202.

Jones M M, Turner N C and Osmond C B 1981 Mechanisms ofdrought resistance.In Physiology and Biochemistry of DroughtResistance in Plants. Eds LG Paleg and D Aspinall. pp 15–37.Academic Press, New York.

Kramer P J and Boyer J S 1995 Water relations of plants and soils.Academic Press, London. 495 p.

Lelièvre F and Desplobins G 1994 Pure lucerne or lucerne in asso-ciation with grasses for extensive pastures in dry Northern Medi-terranean areas.In Management and Breeding of Perennial Lu-cerne for Diversified purposes. Ed. Proceedings FAO-EucarpiaMeeting. pp 89–92. FAO, Rome.

Leopold A C, Musgrave M E and Williams K M 1981 Soluteleakage resulting from leaf dessication. Plant Physiol. 68, 1222–1225.

Levitt J 1980 Responses of plants to environmental stresses, Vol. 2.Academic Press, New York.

Ludlow M M 1989 Strategies of response to water stress.In Struc-tural and Functional Responses to Environmental Stresses. Eds.KH Kreeb, H Richter and TM Hinckley. pp 269–281. SPBAcademic Publishing bv, The Hague.

Ludlow M M and Muchow R C 1990 A critical evaluation of traitsfor improving crop yields in water-limited environments. Adv.Agron. 43, 107–153.

May L H and Milthorpe F L 1962 Drought resistance of crop plants.Field Crop Abstracts 15, 171–179.

McWilliam J R and Kramer P J 1968 The nature of the perennial re-sponse in Mediterranean grasses. I. Water relations and summersurvival inPhalaris. Aust. J. Agric. Res. 19, 381–395.

Ofir M 1986 Seasonal changes in the response to temperature ofsummer-dormantPoa bulbosaL. bulbs. Ann. Bot. 58, 81–89.

Passioura J B 1996 Drought and drought tolerance. Plant GrowthRegul. 20, 79–83.

Premachandra G S, Saneoka H and Ogata S 1992 Leaf water re-lations, osmotic adjustment, cell membrane stability, epicuticular

wax load and growth as affected by increasing water deficits inSorghum. J. Exp. Bot. 42, 739–745.

Qian Y L, Fry J D and Upham W S 1997 Rooting and droughtavoidance of warm-season turfgrasses and tall fescue in Kansas.Crop Sci. 37, 905–910.

Reed K F M 1996 Improving the adaptation of perennial ryegrass,tall fescue, phalaris and cocksfoot for Australia. J. Agric. Res.39, 457–464.

Riga P and Vartanian N 1999 Sequential expression of adaptativemechanisms is responsible for drought resistance in tobacco.Aust. J. Plant Physiol. 26, 211–220.

Torbert H A, Edwards J H and Pedersen J F 1990 Fescues with largeroots are drought tolerant. Appl. Agric. Res. 5, 3, 181–187.

Turner N C 1997 Further progress in crop water relations. Adv.Agron. 58, 293–338.

Turner N C and Begg J E 1981 Plant-water relations and adaptationto stress. Plant Soil 58, 97–131.

Villiers T A 1975 Dormancy and the survival of plants. The Instituteof Biology’s Studies in Biology no. 57. London.

Volaire F 1995 Growth, carbohydrate reserves and drought sur-vival strategies of contrastingDactylis glomeratapopulations ina Mediterranean environment. J. Appl. Ecol. 32, 56–66.

Volaire F and Thomas H 1995 Effects of drought on water rela-tions, mineral uptake, water-soluble carbohydrate accumulationand survival of two contrasting populations of cocksfoot. Ann.Bot. 75, 513–524.

Volaire F and Lelièvre F 1997 Production, persistence and water-soluble carbohydrate accumulation in 21 contrasting populationsof Dactylis glomerataL. subjected to severe drought in the southof France. Aust. J. Agric. Res. 48, 933–944.

Volaire F, Thomas H and Lelièvre F 1998a Survival and recov-ery of perennial forage grasses under prolonged Mediterraneandrought. I. Growth, death, water relations and solute content inherbage and stubble. New. Phytol. 140, 439–449.

Volaire F, Thomas H, Bertagne N, Bourgeois E, Gautier M F andLelièvre F 1998b Survival and recovery of perennial foragegrasses under prolonged Mediterranean drought. II. Water status,solute accumulations, abscisic acid concentration and accumula-tion of dehydrin transcripts in bases of immature leaves. NewPhytol. 140, 451–460.