Physiological and Molecular Plant Pathology (2001) 58, 209±228doi:10.1006/pmpp.2001.0330, available online at http://www.idealibrary.com on

Inhibition of Blumeria graminis germination and germling developmentwithin colonies of oat mildew

T. L.W. CARVER1*, P. C. ROBERTS1, B. J . THOMAS1 and M. F. LYNGKáR2

1Institute of Grassland and Environmental Research, Aberystwyth, Ceredigion SY23 3EB, U.K. and 2Risù National Laboratory,

DK-4000 Roskilde, Denmark

(Accepted for publication April 2001)

eventually d

0885-5765/0

* To who

Germination by Blumeria graminis DC Speer �. spp. avenae, hordei and tritici, was greatly suppressed whenconidia fell within colonies of �. spp. avenae or hordei established on susceptible oat or barley, respectively.On healthy oat or barley, and when distant from powdery mildew colonies, all �. spp. formed normalappressoria. This was also true when conidia germinated within established barley mildew colonies. Withinbarley mildew colonies, appressoria of f. sp. hordei penetrated epidermal cells ( formed haustoria) morefrequently than appressoria distant from colonies. Similarly, �. spp. avenae and tritici, normally unable toinfect barley, frequently penetrated epidermal cells subtending established barley mildew colonies. Thus,colony establishment induced barley epidermal cell accessibility, even to non-pathogenic �. spp. Incontrast, when all three �. spp. germinated within established oat mildew colonies, most formed abnormal,hypha-like germ tubes. Since they did not form appressoria, and were thus unable to attempt penetration,it was impossible to determine whether oat mildew colonies induced accessibility of underlying oatepidermal cells. However, when super®cial structures of established colonies were removed, germlings of all�. spp. formed appressoria freely on cells containing oat mildew colony haustoria. Furthermore, these cellsshowed high level induced accessibility not only to f. sp. avenae but also to the normally non-pathogenic �.spp. This indicated that factors disrupting germination and further development by conidia lying withinoat mildew colonies were produced from the super®cial colony structures and not by haustorium-containing plant cells. The factors appear to have limited mobility. *c 2001 Academic Press

Keywords: Blumeria graminis; Erysiphe graminis; oat mildew; barley mildew; germination; germ tubes;

biocontrol.

INTRODUCTION

Blumeria graminis DC Speer (syn. Erysiphe graminis DC)causes powdery mildew of cereals and other grasses. Thedisease is spread mainly by wind-blown asexual conidia.Following deposition on a host, conidia germinate andgermling morphogenesis proceeds through a series ofdistinct steps. Emergence of the primary germ tube(PGT) provides the ®rst visible sign of germination(within 30±120 min). The PGT remains short (approx.5±10 mm) and aseptate, and is thought to have at leastthree functions, namely, attachment to the host surface[5±7, 11 , 26], gaining access to host water [2], andrecognising host surface features [4 , 6 , 7, 38]. The featuresrecognized probably include hydrophobicity and cutinand cellulose breakdown products released by fungalenzymes [5 , 6 , 10 , 14 , 32 , 35 , 36]. This recognition engagesprocesses that lead to elongation of a second germ tube[4 , 5 , 38] that emerges shortly after the PGT [5, 6 , 26 , 38].Elongation of the second germ tube is pre-requisite to it

i�erentiating a specialized infection structure,

1/050209+20 $35.00/00

m all correspondence should be addressed.

the appressorium. As it grows epiphytically, the elongat-ing appressorial germ tube forms a single septum and itselfrecognizes host surface features [4±6 , 17]. This recognitionleads to swelling of the distal cell and eventually todi�erentiation of a hooked appressorial lobe at the tubeapex (Fig. 1). At 208C, appressoria are formed by 10±12 h after inoculation. Occasionally, ®rst- or later-formedgerm tubes fail to contact, or recognize contact with theleaf, in which case they remain as short, non-functional``subsidiary germ tubes'' [7 , 26 , 38]. In this case, reservespermitting, a subsequent germ tube will take on the PGTfunctions and the next-formed germ tube will elongate.Similarly, elongated germ tubes sometimes fail to contact,or respond to, the host surface, in which case they remainas undi�erentiated, hypha-like structures incapable ofinfection [4 , 5 , 38].

When a functional appressorium is formed, an infec-tion peg emerges from beneath the appressorial lobe andattempts to penetrate the underlying epidermal cell (12±16 h after inoculation) [1]. The attacked cell respondsand papilla deposition is initiated. However, in a com-

patible relationship, the infection peg often penetrates the

*c 2001 Academic Press

(a)

(b)

PGT

PGT

S

S

App

App

H

H

10 mm

10 mm

C

C

FIG. 1. (a and b) Light and LTSEM micrographs, respectively,of ``normal'' B. graminis f. sp. avenae germlings 24 h afterinoculation. Each conidium (c) has formed a short, aseptateprimary germ tube (PGT) and a second appressorial germ tube(App) that has elongated (to approx. 40 mm), swollen and has adistinct, hooked apical lobe (H). A single septum (S) is evidentwithin the appressorial germ tube (approx. 10±15 mm from the

conidium).

210 T. L. W. Ca

response site, enters the cell and an immature primaryhaustorium is formed (15±24 h). The haustorium thendevelops digitate processes from each end of theellipsoidal haustorial body and absorbs nutrients fromthe infected cell [1 , 3 , 18]. These are exported to theappressorium, hyphae emerge from the appressorial germtube or from the mother conidium, and an epiphytichyphal weft grows at accelerating rate over the leafsurface [3 , 18]. After 2±3 days, nutrients supplied by thegrowing primary haustorium become insu�cient tosupport the young colony, and a second generation ofhyphal appressoria is initiated in response to darknessduring the daily cycle [3, 18]. These hyphal appressoriaare usually sited either on the cell containing the primaryhaustorium or within one or two cells distance of it.Penetrations from these appressoria form secondaryhaustoria (usually between two and six) and hyphal

growth accelerates further. Successive generations of

haustoria (with increasing numbers in each generation)are formed nightly. By 4±5 days the ®rst conidiophoresdevelop and sporulation commences shortly afterwards.

Even in compatible host±B. graminis interactions wheresome primary penetration attempts succeed, many fail topenetrate the papilla defence. It is now well establishedthat the ability of B. graminis to penetrate cereal epidermalcells is dramatically a�ected by the outcome of priorattacks by the fungus [8, 20 , 21 , 23 , 24, 27±29 , 33 , 34]. Thishas implications for the ®eld situation where individualcells may be confronted by sequential attacks as conidiaare deposited continually from the aerial pool. Thus, if aninitial attack is successful, the cell shows inducedaccessibility (sensu Ouchi et al. [34]) so that later attackson that cell will almost invariable succeed. By contrast,failed initial attacks induce ``inaccessibility'' (sensuKunohet al. [21]) within the cell so that subsequent attacks almostinvariably fail. While these induced changes are essen-tially localized phenomena, being expressed most stronglyin the single cell directly subjected to earlier attack, thereis some transmission of e�ects to immediately adjacentepidermal cells although the e�ects are often undetectableat two cells distance [8, 21 , 23 , 27±29 , 33].

Lyngkjñr and Carver's previous studies [8 , 27±29]involved double inoculation procedures whereby the ®rst,``inducer'' inoculum was allowed to develop and then itssuper®cial fungal structures (conidia, germ tubes,hyphae) were removed before the second, ``challenge''inoculum was applied. The inducer inoculum was given amaximum of 48 h incubation before removal of super®cialstructures i.e. where primary penetration had succeeded,a primary haustorium was formed but no hyphalappressoria or second generation haustoria were initiated.The success (or failure) of challenge attack was thendetermined in relation to the presence of inducer primaryhaustoria or papillae remaining as intracellular evidenceof successful or failed inducer attack, respectively.

We undertook the present work to extend Lyngkjñr andCarver's earlier studies [8, 27±29]. Here, however, weaimed to allow colonies from inducer inoculum to developsecond generation haustoria before applying challengeinoculum. We intended that germling appressoria of thechallenge inoculum would attack host cells at the timethat inducer colonies were initiating their third generationhaustoria. Thus, our intention was to determine whetherthe presence of multiple haustoria (second and thirdgeneration) produced by a developing colony enhancedthe expression of induced accessibility in nearby, unat-tacked cells.

We executed our ®rst experiments using oat (Avenasativa L.) inoculated with a compatible isolate ofB. graminis f. sp. avenae Marchal. However, we found itimpossible to gather su�cient data to ful®l our objective.Where challenge conidia were deposited within the

rver et al.

boundaries of an established f.sp. avenae colony derived

conditions and a minimum temperature of 128C.

a

from inducer inoculation, very few of these conidiaseemed to germinate. Furthermore, very few of thosethat did germinate formed an appressorium. Themajority of germlings produced highly abnormalhypha-like germ tubes and did not appear to attemptpenetration of the host cells.

The work reported here was therefore redirectedtowards exploring the e�ects that established coloniesmight have on germination and germling developmentby B. graminis conidia deposited within establishedB. graminis colonies. We investigated these e�ects withrespect to f. sp. avenae colonies on susceptible oats and f.

Inhibition of B. gr

sp. hordei colonies on susceptible barley.

MATERIALS AND METHODS

Pathogens and plants

An isolate of B. graminis f. sp. avenae race 5, was multipliedon plants of oat cv. Selma grown in a spore-proofglasshouse under natural lighting conditions and mini-mum temperature of 128C. Similarly, isolate CC1 ofB. graminis f. sp. hordei and an isolate of B. graminis f. sp.tritici (originally supplied by A. Daniels, Aventis, U.K.),were multiplied on barley cv. Pallas and wheat cv.Riband, respectively. One day before conidia wererequired for experimentation, infected leaves were shakento remove older conidia and ensure a supply of vigorousyoung conidia for inoculum.

For all experiments, fully expanded second-formedleaves of cereal plants were used. Studies of oat (A. sativaL.) used cvs Selma and Maldwyn. Neither has knownmajor gene resistance to B. graminis f. sp. avenae race 5.Studies of barley (Hordeum vulgare L.) used cv. Pallas,which is susceptible to B. graminis f. sp. hordei isolate CC1.Since the second leaves of these cvs are susceptible tovirulent fungal isolates, penetration attempts from manyappressoria are successful and relatively few attacks fail inassociation with papilla deposition by living epidermalcells, and few result in death of attacked cells ([8],Lyngkjñr and Carver, unpublished).

Neither oat cv. nor Pallas barley is susceptible toB. graminis f. sp. tritici, Pallas is not susceptible toB. graminisf. sp. avenae, and the oat cvs are not susceptible toB. graminis f. sp. hordei. As expected from previous studies[13], preliminary observations suggested that failure ofinfection by inappropriate �. spp. was explained by acombination of two factors: either papilla responsesprevented penetration or attacked cells died rapidly inresponse to attack. Importantly, even on inappropriatecereal spp., conidia of all fungal �. spp. appeared togerminate and develop normally up to the stage of

appressorial di�erentiation.

To produce disease-free experimental plants, seeds weresown singly in paper tubes (15 � 120 mm) containingsoil : peat : sand (60 : 30 : 10; v : v). Plants were grown tofull expansion of the second-formed leaf (approx. 18 days)in a spore-proof glasshouse under natural lighting

minis development 211

Experimental designs and procedures

Double inoculation experiments: established colonies in place.All of these experiments involved using ®rst a low densityinoculation to establish young colonies on test leaves andthen, 72 h later, applying a second, high densityinoculation. In this way, conidia of the second inocu-lation were deposited both within established colonieswhere they lay touching or between the super®cialhyphae, or distant from colonies when they lay on cellsthat had no contact with hyphae of established colonies.

For the ®rst inoculation, in all cases nine leaves of Selmaor Maldwyn oat, or of Pallas barley were inoculated withconidia of the appropriate f. sp. (avenae or hordei,respectively). Paper tubes containing the plants werelaid on a bench, and their second-formed leaves werearranged adaxial surface up (and held ¯at using smallweights at leaf apex and base), before placing a sporesettling tower over them. Conidia were blown frominfected leaves into the tower and a slide placed amongthe leaves was used to trap conidia and monitor sporedensity which was adjusted to approx. 2±5 conidia mmÿ2.The inoculated plants and four additional healthy plantswere then transferred to a growth cabinet at 208C andunder a 12 h light/12 h dark cycle, with 200 mmolmÿ2 sÿ1 photon ¯ux density during the light period.The ®rst light period commenced immediately afterinoculation.

Under these conditions, previous studies [3 , 9 , 12]allowed us to anticipate the pattern of colony develop-ment. Following establishment of a functional primaryhaustorium at around 20 h after inoculation, secondaryhyphal elongation and branching was expected toaccelerate up to the dark period beginning 60 h afterinoculation, when, in the 60±72 h dark period, hyphalappressoria and then second generation haustoria wouldbe initiated. One of the nine inoculated leaves was cut,®xed, cleared and microscopically examined 71 h afterinoculation to con®rm that colonies were developing asexpected. When this was con®rmed, four of the inocu-lated leaves were ®xed 72 h after inoculation forreference.

At 72 h, a second inoculation was applied to theremaining four inoculated leaves, and the second-formedleaves of the four healthy plants were inoculated simulta-neously to provide a control. The second inoculation wasadjusted to approx. 50 conidia mmÿ2. According to

requirement, conidia for the second inoculation were

a

either of the f. sp. appropriate to the host (i.e. the sameisolate as used for establishing colonies via the ®rstinoculation) or of an inappropriate f. sp. normally unableto infect the test leaves. Following the second inoculation,leaves were returned to the growth cabinet and incubatedfor a further 24 h before all were ®xed.

Double inoculation experiments: established colonies removed.Thirteen second-formed leaves of Selma oat wereinoculated with approx. 2 conidida mmÿ2 of B. graminisf. sp. avenae race 5. These, and 12 healthy (uninoculated)leaves, were incubated as described above, but for 96 h,although one inoculated leaf was ®xed, cleared andmicroscopically examined 95 h after inoculation. Micro-scopy of this leaf con®rmed the anticipation [3] that thethird generation of haustoria had been initiated by thistime. At 96 h after the ®rst inoculation, the adaxialsurface of all leaves (including the healthy ones) waspainted with cellulose acetate/acetone solution which wasallowed to dry (approx. 2±5 min) before the dried acetate®lm was peeled away. This removed all super®cial fungalstructures due to the ®rst inoculation (colony hyphae,failed germlings and conidia) and the epicuticular leafwaxes from inoculated and healthy leaves [10 , 27]. Asexpected [27], epidermal cells containing haustoriaformed by colonies on leaves treated with the ®rstinoculum remained alive (as judged by lack of cyto-plasmic granulation and whole-cell auto¯uorescence).Four leaves that had received the ®rst inoculum were theninoculated for a second time using 50 conidia mmÿ2 ofeither f. sp. avenae, f. sp. hordei or f. sp. tritici, and di�erentsets of four healthy leaves were also inoculated simul-taneously with the di�erent �. spp. All were incubated for

212 T. L. W. C

a further 36 h before ®xation.

Preparation for light microscopy

As described previously (e.g. [13]) leaves were ®xed andcleared by a procedure that avoids displacing ungermi-nated conidia and loosely attached germlings. Brie¯y,central 30 mm segments of leaves were excised, and laidinoculated surface up on tissue paper pads moistenedwith 3 : 1 ethanol : glacial acetic acid (v : v) to ®x andbleach the leaves. They were then transferred to padsmoistened with water and subsequently to pads mois-tened with lactoglycerol (lactic acid : glycerol : water,1 : 1 : 1, v : v) to clear. For studies of fungal developmentwhere leaves are ®xed before large colonies are allowed todevelop, it is possible to attain good resolution ofpathogen and host using unstained leaves mountedwithout coverslips, and viewed with no-coverslip objec-tives. However, because of di�raction around hyphae ofcolonies established by the ®rst inoculum, it was

impossible to use this approach to resolve details of

second inoculum development within established colo-nies. We needed, therefore, to stain specimens.

The ®rst experiment involved double inoculatingSelma oat with B. graminis f. sp. avenae, leavingestablished colonies from the ®rst inoculation in place.Here, staining was achieved by placing a smear ofTrypan blue in lactoglycerol (0.1 g per 100 ml) on acoverslip, lowering the adaxial (inoculated) surface of theleaf segment onto the coverslip, and then inverting themount onto a microscope slide smeared with lactogly-cerol. This procedure was devised to allow stainingwithout displacing loosely attached germlings. We usedthis to gather data on the developmental characteristicsof germlings formed by conidia of the second inoculumin relation to their proximity to established colonies.However, we found that even with care some ungermi-nated conidia were displaced; they could be seen movingover relatively long distances if specimens were viewedimmediately after mounting when the liquid media were¯owing over segments. Therefore, for this experiment wehad limited con®dence in data for germination frequencyby conidia of the second inoculum in relation to theirproximity to established colonies.

To allow assessment of germination frequency as wellas the morphological characteristics of germlings formedby the second inoculum, we devised a staining techniquethat caused no detectable displacement of ungerminatedconidia. This was used for all further experiments. Asbefore, leaves were ®xed, bleached, and cleared, butdishes containing the segments supported on tissue paperwere held on a slant to avoid accumulation of freeliquids around segments and the possibility that freeliquid would move onto the segment surface. To stain,segments supported on slanted dishes were sprayed verylightly using a hand-pumped atomizer (Schott,Germany) from a distance of approx. 40 cm withalcoholic methyl blue stain (0.2 g methyl blue in100 ml 95% ethanol). A small amount of spray (sixpumps) was applied and then spraying was suspendedfor 1 min to allow the ethanol to evaporate. In this way,no visible droplets of liquid accumulated. A furtherspray was then applied. This process was repeated untilmicroscope inspection showed colonies, germlings andungerminated conidia to be stained su�ciently (usuallythree cycles). Repeated observation (before and afterspray staining, without coverslip) of spores locateddistantly from established colonies, showed that theprocedure did not displace ungerminated conidia.Following staining, specimens were mounted without acoverslip and viewed by transmitted light and di�eren-tial interference contrast microscopy using 20 � or40 � objectives. Observations were made immediatelybecause although staining remained good for approx.

rver et al.

24 h, it faded thereafter.

a

Data collection

Double inoculation experiments: established colonies in place.Our aim in these experiments was to determine the e�ectcolonies established from the ®rst inoculation might haveon conidial germination and the development of germl-ings from the second inoculation.

For the ®rst experiment where coverslips were ®tted, oneach control leaf (inoculated once, 24 h before ®xation)50 randomly selected, germinated conidia wereexamined. Conidia were classi®ed as germinated if atleast one short germ tube had been produced. For each ofthese germlings, note was taken of whether they haddeveloped abnormally or had developed normally todi�erentiate an appressorial germ tube (illustrated inFig. 1). Appressorial germ tubes were recognized by: thepresence of a single septum within the tube, swelling

Inhibition of B. gr

towards the tube apex, and the presence of a hooked

Inner zoneOuter zone

(c) Inner and outer zones of 96 h colony

(b) Colony 96 h after inoculation

(a) Colony 72 h after inoculation

100 µm

100 µm

La1

Lo1 Lo2

La2

Max. longitudinal distanceC

Max. lateral distance

FIG. 2. Drawings of B. graminis f. sp. avenae colonies ®xed 72 h(a) and 96 h (b) and (c) after inoculating oat cv. Selma. (a)Positioning of co-ordinates for measurement of maximumlongitudinal and lateral measurements. C � conidium. (b)Positioning of co-ordinates marking the last branches on leaderhyphae. Measurement of Lo1 to Lo2 gives the longitudinaldistance between last branches; La1 to La2 gives lateral distancebetween last branches. (c) The inner and outer zones of a 96 hcolony [same as shown in (b)]. The inner zone lies within animaginary boundary line linking the points at which the lasthyphal branch occurred on the colony's major runner hyphae.The outer zone (shaded) lay between the inner boundary andthe outer boundary which was de®ned by an imaginary line

joining the tips of the colony's major hyphal branches.

apical lobe. Note was also made of whether appressoriawere associated with successful penetration as evidencedby the presence of a haustorium. Abnormal germlingswere those which had either failed to form an elongatedgerm tube (at least 15 mm), or where the elongated germtube had not di�erentiated into an appressorial germtube. On double-inoculated leaf segments ®xed 24 h afterapplication of the second inoculum, observations weremade for 50 germinated conidia that lay on leafepidermal cells located ``distant'' from the nearest colony,and germlings were classi®ed in the same way as oncontrols. Distant cells were de®ned as being at least fourepidermal cells away from the nearest cell in contact withhyphae belonging to colonies formed by the ®rstinoculum. Similar classi®cations were made for 50germlings arising from conidia located within the outer``boundaries'' of colonies established by the ®rst inocu-lum. The colony boundary was de®ned by an imaginaryline joining the tips of a colony's leader hyphae (Fig. 2).

Of course, on double inoculated leaves, some germlingsderiving from the ®rst inoculation failed to penetrate andestablish colonies, and we could not distinguish betweenthese and germlings from the second inoculation. Some ofthese failed germlings would inevitably be included inour samples. However, since the density of the secondinoculation (appox. 50 conidia mmÿ2) was 10±25 timesgreater than the ®rst (approx. 2±5 conidia mmÿ2), andsince many germlings from the ®rst inoculation formedcolonies, the great majority of data would necessarilyrelate to germlings derived from the second inoculation.Therefore, we introduced only small errors due toinclusion of observations of germlings from the ®rstinoculation.

We wished to gauge the growth made by colonies inthe period 72±96 h after inoculation i.e. from the time ofapplying second inoculum to the time of ®xation fordouble inoculated leaves. Drawings of two ``typical''B. graminis f. sp. avenae colonies are shown in Fig. 2(a) and(b) ( for 72 h and 96 h colonies, respectively). Asmeasures of colony size, the maximum longitudinal andlateral distances from tip to tip of the longest hyphae[Fig. 2(a) and (b)] were measured for 10 colonies on eachleaf segment ®xed 72 h after inoculation, and for 10colonies ®xed 96 h after inoculation. For the 96 hsegments, measurements were also made of the maximumlongitudinal and lateral distances between the lastbranches on leader hyphae [Fig. 2(b)]. From thesemeasurements, it was deduced that the lengths andbreadths of 72 h colonies were approximately equivalentto the last-branch length and breadth distances seen in96 h colonies. Thus, when observing 96 h colonies, theposition of the last branch on leader hyphae o�ered anapproximate, but reasonable, indication of the location of

minis development 213

leader hyphal tips at 72 h when the second inoculum was

prior to analysis of variance.

shop2 to maximize details of areas of interest.

a

applied. This information allowed us to modify datacollection criteria for subsequent experiments.

In remaining experiments where the fungus was spray-stained, additional data were collected. As describedabove, on controls (inoculated once, 24 h before ®xation)50 germlings were classi®ed for normality of develop-ment, but in addition, 100 randomly selected conidiawere examined to determine whether they had germi-nated. Similarly, on double-inoculated leaves ®xed 24 hafter the second inoculation, 50 germlings lying onepidermal cells distant from colonies were classi®ed fornormality of development, and 100 conidia on such cellswere examined for germination. In relation to establishedcolonies, data were collected from two di�erent zoneswithin colonies. The inner zone was de®ned by animaginary boundary line linking the points at which thelast hyphal branch occurred on the colony's major runnerhyphae [Fig. 2(c)]. The outer zone lay between the innerboundary and the outer boundary which was, as before,de®ned by an imaginary line joining the tips of a colony'smajor hyphal branches [Fig. 2(c)]. On each leaf segment,many randomly selected colonies were examined anddata were collected from within each zone until we hadaccumulated observations from 100 conidia to determinewhether they had germinated, and from 50 germinatedconidia for normality of germling development.

As considered above, on double inoculated leaves, onlysmall errors would have been introduced due to inclusionof data deriving from the ®rst inoculum in the data sets.

For all leaf segments ®xed 24 h after the secondinoculation and in all experiments, mean percentageswere calculated, as appropriate, for spore germination,abnormal germling development, and haustorium for-mation. Data sets were gathered from each location ondouble inoculated leaves (distant from colonies, and, asappropriate, within colony outer and inner zones) and forcontrols inoculated once 24 h before ®xation. Percentageswere transformed to arcsine square roots (transformedvalue � 180/p � arcsine [Z/(100)]) to normalize dataand stabilize the variance throughout the data range,before being subjected to analysis of variance (usingGenstat [15]).

Double inoculation experiments: established colonies removed.Our aim here was to determine whether any e�ects of the®rst inoculation on subsequent performance of conidiaand germlings arising from the second inoculation,depended: (1) upon the presence of colony hyphaederived from the ®rst inoculation; or (2) whether thepresence of haustoria within epidermal cells was su�cientto mediate these e�ects.

On each control leaf segment, data to assess germina-tion and normality of germling development weregathered as described above. On leaves inoculated

214 T. L. W. C

twice, germination was determined for 50 conidia located

on epidermal cells at least four cells away from the nearestcell containing a haustorium formed by colonies arisingfrom the ®rst inoculum (distant cells), and 25 germlingson distant cells were scored for their developmentalcharacteristics. Similarly where conidia of the secondinoculation lay on epidermal cells containing haustoriaformed by colonies of the ®rst inoculation, germinationwas determined for 50 conidia and germling developmentwas assessed for 25 germinated conidia.

As described above, mean data from each leaf segmentwere calculated as percentages that were transformed

rver et al.

Low temperature scanning electron microscopy (LTSEM)

To complement light microscopy and aid imageinterpretation, sub-samples of double inoculated leaveswere examined by LTSEM. Thus, when the central30 mm leaf segments were excised and ®xed for lightmicroscopy, the adjacent basipetal 10 mm leaf bladesegment from some samples was taken for LTSEM.Following established procedure [11], these segmentswere attached to a copper SEM stub with colloidalgraphite and frozen rapidly by introducing the stub to apre-cooled stub holder (approx. ÿ1908C) under anargon ¯ush, and transferring the assembly to the pre-cooled (ÿ1908C) stage of an Emscope SP2000A sputter-cryo system. Cryo-®xation in this way avoided ice crystalformation on specimens. The specimen was then sputtercoated with gold before detailed observation on the coldstage (approx. ÿ1608C) of a JEOL JSM 840A SEM,with 3.0 or 5.0 kV electron accelerating voltage. SEMimages were recorded using a digitizer board andsoftware (SemAfore2, JEOL) running under a Win-dows2 operating system. The ®gures presented here arefrom digitized images enhanced using Adobe Photo-

RESULTS

Selma oat double-inoculated with f. sp. avenae, coverslipsapplied

Data for spore germination will not be described becauseapplying coverslips to specimens was seen to displacemany ungerminated spores. Fig. 3 shows that depositionwithin the boundary of an established colony had adramatic e�ect on germling development. On controlleaves (inoculated only once, 24 h before ®xation) andwhen germlings were distant from established colonies ondouble-inoculated leaves, the great majority of germlingsdeveloped normal appressoria. By contrast, when conidiawere located within the boundary of an established

colony, more than 70% of germlings were abnormal

germlings formed haustoria.

20

40

60

80

(a)100

0

(%)

aa

b

10

20

30

40

(b)50

0

(%)

a

ab

b

Control

Distant

Within

Conidium location

relative to colony

FIG. 3. Development of B. graminis f. sp. avenae germlings on oatcv. Selma second-formed leaves 24 h after inoculation ontohealthy control leaves or onto leaves bearing B. graminis f. sp.avenae colonies derived from a ®rst inoculation 72 h earlier. Allmaterial was ®xed 24 h after the second inoculation. (a)Percentage of morphologically abnormal germlings that failedto develop an appressorium. (b) Percentage of germlings withapparently normal appressoria that penetrated plant epidermalcells successfully to form a haustorium. On controls ( ), dataare based on counts of 50 randomly selected germlings, while onleaves bearing colonies, counts were made for 50 germlingslocated distant from colonies ( ) and a further 50 germlingswere located within outer colony boundaries (h), on each offour replicate leaves in all cases. Percentage data weretransformed to angles for analysis of variance. Where columnsare headed by di�erent letters, LSD tests applied to angle-transformed data indicated a signi®cant di�erence (P5 0.05)

Inhibition of B. gra

[Fig. 3(a)]. Sometimes more than one short germ tubewas formed, and none elongated, but this was relativelyrare. Most often, conidia that germinated formed a shortPGT and a second germ tube that grew to a length fargreater than normal for an appressorial germ tube(Fig. 4). These long tubes failed to form a hookedappressorial lobe at their apex. Sometimes they appearedto be slightly swollen at the point where an appressoriallobe would be expected (30±50 mm from the conidium),but the germ tube then narrowed and extended as a

between means within a data set.

hypha-like tube with parallel walls and a simple, rounded

apex. When germlings were within the boundary ofestablished colonies, they frequently made direct contactwith hyphae of the established colony [Fig. 4(a), (b), (d)and (e)]. Contact could be via the conidium, the germtubes, or both. However, abnormal germlings wereformed even where no direct contact was made[Fig. 4(c) and ( f)].

On controls and when germlings were distant fromestablished colonies on double inoculated leaves, approx.15±30% of germlings successfully penetrated leaf epi-dermal cells to form a haustorium [Fig. 3(b)]. Bycontrast, when germlings were located within theboundary of an established colony, less than 5% of

minis development 215

Experiments where specimens were viewed without coverslips,and hyphae of established colonies remained in place

Oats inoculated ®rst with f. sp. avenae and secondly with �.spp. avenae, hordei or tritici

Germination. When spray stained specimens weremounted without coverslips to avoid displacing ungermi-nated conidia we felt con®dent that data gathered todescribe spore germination were valid. Here, therefore,we considered percentage germination of conidia locatedon control leaves, and, where leaves bore coloniesestablished from the ®rst inoculation, we considered thegermination of conidia deposited distant from coloniesand lying within the outer and inner zones of colonies[Fig. 2(c)].

Fig. 5(a) and (b) show that when f. sp. avenae conidiawere inoculated onto Maldwyn or Selma oat leavesbearing established f. sp. avenae colonies, germinationrates were signi®cantly depressed if conidia lay within theinner zone of an established colony. Here, germinationpercentages were approximately halved compared withlocations distant from colonies or on controls inoculatedonly once. This was true for experiments involving bothoat cvs. There was no di�erence between controls andlocations distant from colonies on double-inoculatedleaves. In both experiments, there was a tendency forconidia within the outer zone of colonies to show slightlylower germination than at distant locations or oncontrols, but this e�ect was small and signi®cant onlyfor the experiment with cv. Maldwyn.

When f. sp. tritici conidia were inoculated ontoMaldwyn oat leaves bearing f. sp. avenae colonies, e�ectsof location on germination by f. sp. tritici conidia werevery similar [Fig. 5(c)] to those seen in Fig. 5(a) and (b).Germination frequency by conidia located within theinner zone of f. sp. avenae colonies was about half that seenfor conidia distant from colonies or on controls, butgermination was not depressed if conidia were within the

outer zone of colonies.

(a)

(b)

(c)

(d)

(e)

(f)

50 µm 30 µm

30 µm

30 µm50 µm

50 µm

CuAb

PGT

PGTC

C

Ab

Hy

Cu

Hy Ab

PGT

C

C

C'

Hy

Ab

PGT

AbHy

C

PGT

C

Ab

Ab

Hy

Hy

SGT

C

PGT

FIG. 4. Light (a)±(c) and LTSEM (d)±( f) micrographs showing morphologically abnormal germlings of B. graminis f. sp. avenaegrowing within the outer boundary of B. graminis f. sp. avenae colonies on Selma oat leaves. Colonies were established from a ®rstinoculation applied 72 h earlier than the second inoculation from which the germlings derive. All material was ®xed 24 h after thesecond inoculation. Often, germlings derived from the second inoculum had contact with hyphae of established colonies (a), (b), (d)and (e) via either the conidium, germ tubes or both, although in some cases no contact was made (c) and ( f). Sometimes, conidiawere ungerminated (a) and (b) or had formed only one [bottom spore, (b)] or more short germ tubes. Where a long germ tube wasproduced, a short germ tube with the appearance of a typical primary germ tube (a)±(d) and ( f) was almost invariably visible. [In(e) it is probable that a primary germ tube is hidden beneath the conidium]. Occasionally germlings with a long germ tube hadmore than one short germ tube (d) in which case one of these made contact with the host surface and had the characteristics of aprimary germ tube, while the other(s) was typical of a redundant subsidiary germ tube. Frequently, abnormal long germ tubesbecame distinctly swollen towards a region approx. 30±50 mm from the conidium before the tube narrowed and continued growthas a hypha-like tube with parallel walls and a simple rounded apex [germling on right in (a), central germling in (b) and (e)],although in some cases no sign of swelling was evident ( f). Ab � abnormal long germ tube; C � conidium; Cu � ungerminatedconidium; C0 � conidium with short germ tube only; Hy � hypha of established colony; PGT � primary germ tube;

SGT � subsidiary germ tube.

216 T. L. W. Carver et al.

For f. sp. hordei, the suppression of germinationassociated with deposition within f. sp. avenae coloniesappeared even greater than for conidia of other �. spp.[Fig. 5(d)]. When f. sp. hordei conidia were depositedwithin the inner zone of colonies, germination wasreduced by nearly 70% (to approx. 17%) comparedwith locations distant from the colonies (appox. 53%

germination). Again, suppression of germination was less

marked for conidia deposited within the outer zone off. sp. avenae colonies, although even here a signi®cantlylower percentage germinated than if located distant fromcolonies. Interestingly, there was also a small butsigni®cant reduction in germination for f. sp. hordeiconidia located distant from f. sp. avenae colonies ondouble inoculated leaves compared with controls. This

e�ect was not seen in any other combination.

20

40

60

80

(a)100

0

aa

b

c

20

40

60

80

100

0

Conidia

thatgerm

inated(%

)

(c)

a a

a

b

Control

Distant

Outer

zone

Inner

zone

Conidium location

relative to colony

Control

Distant

Outer

zone

Inner

zone

Conidium location

relative to colony

(b)

aa

ab

(d)

a

b

c

d

FIG. 5. Percentages of B. graminis �. spp. avenae, tritici and hordei conidia that had germinated 24 h after inoculation onto second-formed oat leaves of healthy controls, or onto leaves bearing B. graminis f. sp. avenae colonies derived from a ®rst inoculation 72 hearlier. All material was ®xed 24 h after the second inoculation. Data are for f. sp. avenae conidia on oat cvs Maldwyn (a) and Selma(b), and for �. spp. tritici (c) and hordei (d) on cv. Maldwyn. On controls ( ) data are based on counts of 100 randomly selectedconidia, while on leaves bearing colonies, counts were made for 100 conidia located distant from colonies ( ), 100 within the outerzone of colonies ( ), and 100 located within the inner zone of colonies ( ), on each of four replicate leaves in all cases. Percentagedata were transformed to angles for analysis of variance. Where columns are headed by di�erent letters, LSD tests applied to angle-

transformed data indicated a signi®cant di�erence (P5 0.05) between means within a data set.

Inhibition of B. graminis development 217

Germling development. Fig. 6(a) and (b) show that when f.sp. avenae conidia germinated on Maldwyn or Selma oatleaves bearing established f. sp. avenae colonies, a veryhigh percentage (474%) developed abnormally ifconidia lay within the inner zone of an establishedcolony. As described for the ®rst experiment (Fig. 4),abnormal germlings were those that failed to form a longgerm tube, or, where a long germ tube was formed, it didnot di�erentiate an appressorium. In general, abnormal-ities were infrequent (appox. 4±12%) for germlings oncontrols or located distant from colonies. They were alsorelatively infrequent when germinated conidia lay withinthe outer zone of established colonies (518%).

When f. sp. tritici germlings developed on Maldwyn oatleaves bearing f. sp. avenae colonies, e�ects of location ongermling development by f. sp. tritici [Fig. 6(c)] were verysimilar to those seen in Fig. 6(a) and (b). On controls andin locations distant from colonies, very few germlings

developed abnormally (57%). Within the outer zone

of f. sp. avenae colonies, the abnormality frequency wasslightly but signi®cantly higher (approx. 16%).However, when f. sp. tritici conidia were within theinner colony zone, the majority of germlings developedabnormally (77%). The abnormalities were similar tothose shown in Fig. 4.

As with e�ects on germination, the e�ects of germlinglocation on the frequency of abnormal germling devel-opment appeared greater for f. sp. hordei than other �.spp. As in other cases, when f. sp. hordei germlings were oncontrol leaves or distant from f. sp. avenae colonies, a verylow frequency of abnormality was seen [59%,Fig. 6(d)], and when located within the inner zone ofcolonies abnormalities were extremely frequent [82%,Fig. 6(d)]. However, in contrast to other combinations, arelatively high abnormality frequency [appox. 49%,Fig. 6(d)] was seen when germinated f.sp hordei conidiawere located in the outer zone of f. sp. avenae colonies.

These data indicated that f. sp. hordei was more sensitive

20

40

60

80

(a)100

0

ab b

c

20

40

60

80

100

0

Germ

inatedconidia

developingabnorm

ally(%

)

Control

Distant

Outer

zone

Inner

zone

Conidium location

relative to colony

Control

Distant

Outer

zone

Inner

zone

Conidium location

relative to colony

(c)

a a

b

c

(b)

aab

b

c

(d)

a a

b

c

FIG. 6. Percentages of B. graminis �. spp. avenae, tritici and hordei conidia that germinated but formed abnormal germ tube structures( failed to form appressoria) 24 h after inoculation onto second-formed oat leaves of healthy controls, or onto leaves bearingB. graminis f. sp. avenae colonies derived from a ®rst inoculation 72 h earlier. All material was ®xed 24 h after second inoculation.Data are for f. sp. avenae conidia on oat cvs Maldwyn (a) and Selma (b), and for �. spp. tritici (c) and hordei (d) on cv. Maldwyn. Oncontrols ( ), data are based on counts of 50 randomly selected germlings, while on leaves bearing colonies, counts were made for 50germlings located distant from colonies ( ), 50 within the outer zone of colonies ( ),and 50 within the inner zone of colonies ( ),on each of four replicate leaves in all cases. Percentage data were transformed to angles for analysis of variance. Where columns areheaded by di�erent letters, LSD tests applied to angle-transformed data indicated a signi®cant di�erence (P5 0.05) between

means within a data set.

218 T. L. W. Carver et al.

to the in¯uences of f. sp. avenae colonies than either f. sp.avenae itself or f. sp. tritici.

In addition to being more sensitive to the in¯uence of f.sp. avenae colonies, f. sp. hordei germlings appeared to showa greater range of morphological abnormalities (Fig. 7)than germlings of the other �. spp.

Haustorium formation. Germlings of f. sp. hordei neverformed haustoria in oat cv. Maldwyn irrespective ofwhether they attacked control leaves or leaves previouslyinfected with f. sp. avenae. Haustorium formation by f. sp.tritici was also extremely rare; no germlings formedhaustoria in control leaves or on cells distant fromestablished f. sp. avenae colonies. Within the outer zoneof f. sp. avenae colonies, only 1% of f. sp. tritici germlingsformed a haustorium and within the inner zone, only

1.5% did so.

There were no di�erences in the percentages of f. sp.avenae germlings forming haustoria in epidermal cells ofcontrol leaves and cells distant from established coloniesof f. sp. avenae (Fig. 8). On Maldwyn [Fig. 8(a)], therewas no di�erence between germlings distant from coloniesand those within the outer zone of colonies, althoughwithin the inner zone of colonies signi®cantly fewergermlings formed haustoria than in any other location. Inthe experiment using Selma [Fig. 8(b)], similar percen-tages of germlings formed haustoria in controls, in cellsdistant from colonies and in the inner zone of colonies.However, when germlings were located within the outerzone of colonies, a signi®cantly higher percentage formedhaustoria than in any other location.

Barley cv. Pallas inoculated ®rst with f. sp. hordei and

secondly with �. spp. hordei, avenae or tritici.

(a) (c)

(b) (d)

Hy

Hy

Hy

Hy

Cu

CuPGT

PGT

PGT

PGT

30 µm 30 µm

30 µm 30 µm

FIG. 7. Light micrographs showing examples of abnormal germling structures formed by B. graminis f. sp. hordei when conidiagerminated within the boundary of f. sp. avenae colonies growing on Maldwyn oat leaves. Colonies were established from a ®rstinoculation applied 72 h earlier than the second inoculation from which the germlings derive. All material was ®xed 24 h after thesecond inoculation. (a) The elongated germ tube is a simple, hypha-like structure showing little sign of swelling or apical hooking,and two septa are evident. The ®rst is close to the spore, at the position where the single septum forms in morphologically normalappressorial germ tubes, and the second is approximately 35 mm from the conidium i.e. approximately where the apical appressorialhook is formed by a normal appressorial germ tube. Note: three ungerminated conidia are located close to hyphae (top right ofmicrograph). (b) As in (a), two septa are present in the elongated germ tube. The central cell of this tube is swollen, but a furtherhypha-like cell has been formed. Small protruberances are evident on the central cell, the most distal of which (directeddownwards) is reminiscent of a poorly developed appressorial lobe. (c) Again two septa are present in the elongated germ tube. Thecentral cell is swollen and shows distinct protruberances, but these are clearly too small to be mistaken for appressorial lobes. Thedistal cell of the germ tube is slightly swollen with a club-like apex. (d) When ®rst seen, this germling was mistakenly thought tohave formed a normal appressorium with a hooked apical lobe (double arrowhead). The branch cell, separated from the hooked cellby a septum and growing left from the central portion of the hooked cell, was thought to be secondary hyphal growth. However,careful examination showed that no haustorium was present beneath the hooked structure. Therefore the cell growing leftrepresented abnormal branching of the elongated germ tube and the hooked cell cannot be interpreted as a normal appressorialstructure. Several examples of similar germlings were seen. Cu � ungerminated conidium; Hy � hypha of established colony;PGT � primary germ tube; arrowheads indicate septa in elongated germ tube; double arrowheads indicate appressorium-like

structure.

Inhibition of B. graminis development 219

Germination. Data from barley showed trends verysimilar to those from oat (Fig. 9). Conidia of all �. spp.germinated well on Pallas barley control leaves, andequally well on cells distant from established colonies of f.sp. hordei. There was a slight tendency for germination tobe reduced when conidia were located within the outerzone of f. sp. hordei colonies, but the reduction was smalland signi®cant only in the case of f. sp. avenae. In all cases,germination was signi®cantly reduced where conidia werelocated within the inner zone of colonies. However, thedegree of reduction in germination was relatively smallerthan that seen within the inner zone of f. sp. avenae

colonies on oat (where germination rates were at least

halved; Fig. 5). The maximum reduction was seen with f.sp. avenae where on distant cells approx. 69% germinatedcompared to approx. 39% within the inner zone of f. sp.hordei colonies (i.e. a reduction of 43%) while there wereonly 32 and 31% reductions in germination of f. sp. hordeiand tritici, respectively.

Germling development. Extremely low frequencies ofgermling abnormality were seen when conidia of any �.spp. germinated within f. sp. hordei colonies growing onbarley (Fig. 10). This was in complete contrast to thesituation seen when conidia germinated within colonies of

f. sp. avenae growing on oat. Even when the germlings

observation were not collected.

10

20

30

40

50

0

Germ

inatedconidia

form

inghaustoria(%

)

Control

Distant

Outer

zone

Inner

zone

Conidium location

relative to colony

Control

Distant

Outer

zone

Inner

zone

Conidium location

relative to colony

(a)a

ab

b

c

(b)

a

ab

c

b

FIG. 8. Percentages of B. graminis f. sp. avenae conidia that had germinated on second-formed leaves of oat cvs Maldwyn (a) andSelma (b), and that formed morphologically normal appressoria from which plant epidermal cells were penetrated successfully,resulting in haustorium formation, 24 h after inoculation onto healthy control leaves or onto leaves bearing B. graminis f. sp. avenaecolonies derived from a ®rst inoculation 72 h earlier. All material was ®xed 24 h after the second inoculation. On controls ( ), dataare based on counts of 50 randomly selected germlings, while on leaves bearing colonies, counts were made for 50 germlings locateddistant from colonies ( ), 50 within the outer zone of colonies ( ) and 50 within the inner of colonies ( ), on each of four replicateleaves in all cases. Percentage data were transformed to angles for analysis of variance. Where columns are headed by di�erentletters, LSD tests applied to angle-transformed data indicated a signi®cant di�erence (P5 0.05) between means within a data set.

220 T. L. W. Carver et al.

were intimately associated with hyphae of an establishedcolony they almost always formed normal appressoria(Fig. 11). For f. sp. hordei conidia, there was no di�erencein abnormality frequency between germlings locatedwithin the inner zone of colonies compared with controls,germlings located distant from colonies or those locatedwithin the colony outer zone. For �. spp. avenae and tritici,there was a signi®cant increase in the frequency ofabnormality within the inner zone of f. sp. hordei colonies.However, even here only approx. 10% of germlingsappeared abnormal (as compared with 70±80%abnormality for any �. spp. germinating within theinner zone of f. sp. avenae colonies on oat).

Haustorium formation. When f. sp. hordei germlingsattacked control barley leaves, epidermal cells distantfrom established colonies, or were located within theouter zone of colonies, 13±14% of appressoria wereassociated with successful penetration attempts as evi-denced by the presence of a haustorium beneath theirappressoria (Fig. 12). However, when germlings werelocated within the inner zone of colonies, the percentageof appressoria with haustoria was signi®cantly increasedand almost doubled in frequency (22.5%).

On controls and on cells distant from f. sp. hordeicolonies, germlings of �. spp. avenae and tritici neverformed haustoria (Fig. 12). By contrast, when germlingsof these inappropriate �. spp. were located within theinner zone of f. sp. hordei colonies, they frequently formeda haustorium (17.5 and 14.5% for avenae and tritici,

respectively, Figs 12 and 13), and even if they were

located within the outer zone of colonies, a few germlingsformed haustoria (4 and 2%, respectively).

From casual observation, it was apparent that whengermlings of the inappropriate �. spp. located within a f.sp. hordei colony formed a haustorium, this haustoriumwas most frequently formed in an epidermal cell that alsocontained a haustorium of the established colony(illustrated in Fig. 13). However, data to con®rm this

Double inoculation of Selma oat: hyphae of established coloniesremoved

Removing the super®cial hyphae of B. graminis f. sp.avenae colonies that had been previously allowed todevelop for 96 h before re-inoculation, allowed us todetermine whether e�ects on germination and germlingdevelopment depended upon the presence of hyphae orwhether the presence of haustoria within epidermal cellswas su�cient to mediate these e�ects. Where colonyhyphae were removed, haustoria formed by the colonywere clearly visible, and epidermal cells containing thosehaustoria appeared to remain alive as evidenced by lackof whole-cell auto¯uorescence and absence of cytoplasmicgranulation or disorganization. This was as expectedfrom earlier studies [8 , 27±29] where haustorium-containing cells also survived removal of super®cialstructures. Cells containing colony haustoria obviouslylay within the outer boundary of colonies as de®nedearlier by an imaginary line linking tips of leader hyphae.

Furthermore, from our observations and Hirata's [18]

20

40

60

80

100

0

Conidia

thatgerm

inated(%

)

Control

Distant

Outer

zone

Inner

zone

Conidium location

relative to colony

20

40

60

80

100

0

20

40

60

80

100

0

(a)

a a a

b

(b)

a ab

c

(c)

aa

ab

b

FIG. 9. Percentages of B. graminis �. spp. hordei (a), avenae (b)and tritici (c) conidia that had germinated 24 h after inoculationonto second-formed cv. Pallas barley leaves of healthy controls,or onto leaves bearing B. graminis f. sp. hordei colonies derivedfrom a ®rst inoculation 72 h earlier. All material was ®xed 24 hafter the second inoculation. On controls ( ), data are based oncounts of 100 randomly selected conidia, while on leaves bearingcolonies, counts were made for 100 conidia located distant fromcolonies ( ), 100 within in the outer zone of colonies ( ), and100 within the inner zone of colonies ( ), on each of fourreplicate leaves in all cases. Percentage data were transformed toangles for analysis of variance. Where columns are headed bydi�erent letters, LSD tests applied to angle-transformed dataindicated a signi®cant di�erence (P5 0.05) between means

within a data set.

5

10

15

20

0

Germ

inatedconidia

developingabnorm

ally(%

)

Control

Distant

Outer

zone

Inner

zone

Conidium location

relative to colony

(c)

a

aa

b

5

10

15

20

0

(b)

ab

aa

b

5

10

15

20

0

(a)

aa

a a

FIG. 10. Percentages of B. graminis �. spp. hordei (a), avenae (b)and tritici (c) conidia that germinated but formed abnormalgerm tube structures ( failed to form appressoria) 24 h afterinoculation onto second-formed cv. Pallas barley leaves ofhealthy controls, or onto leaves bearing B. graminis f. sp. hordeicolonies derived from a ®rst inoculation 72 h earlier. Allmaterial was ®xed 24 h after the second inoculation. Oncontrols ( ), data are based on counts of 50 randomly selectedgermlings, while on leaves bearing colonies, counts were madefor 50 germlings located distant from colonies ( ), 50 within theouter zone of colonies ( ), and 50 within the inner zone ofcolonies ( ), on each of four replicate leaves in all cases.Percentage data were transformed to angles for analysis ofvariance. Where columns are headed by di�erent letters, LSDtests applied to angle-transformed data indicated a signi®cant

di�erence (P5 0.05) between means within a data set.

Inhibition of B. graminis development 221

colony haustoria.

(a) (b)

PGT

PGT

C C

C

H

H

App

S

HyApp

App

Hy

H

20 µm 20 µm

FIG. 11. LTSEM micrographs of normal B. graminis f. sp. hordei germlings that developed when in intimate contact with hyphae ofB. graminis f. sp. hordei growing on a barley leaf. The germlings grew from conidia inoculated 24 h earlier, while the colonies derivedfrom inoculation made 96 h earlier. In (b) the conidium on the left has formed two subsidiary germ tubes, probably because it wasdeposited upon the hyphae of the colony and failed to perceive the underlying host leaf surface [38]. The primary germ tube of thisconidium is presumably hidden from view. App � appressorial germ tube; C � conidium; H � hooked apical lobe of appressorium;

Hy � hyphae of colony; PGT � primary germ tube; S � subsidiary germ tube.

222 T. L. W. Carver et al.

detailed studies, it is most likely that haustorium-containing cells lay within the inner zone of colonieswhose hyphae had been removed. On the other hand,epidermal cells four cells distant from haustorium-containing cells probably lay outside ( for acropetal andbasipetal cells), or only just within ( for lateral cells), theboundaries of colonies that had been removed. Thus,these epidermal cells were regarded as comparable tocells previously de®ned as distant from colonies.

Germination and germling development. Overall, slightly butsigni®cantly (P5 0.05) fewer conidia germinated oncontrol leaves (approx. 63%) than on double inoculatedleaves. However, unlike experiments where colonyhyphae were left in place, there was no di�erence ingermination between conidia distant from cells contain-ing colony haustoria (germination � approx. 68%), andconidia overlying epidermal cells containing colonyhaustoria (germination � approx. 69%). There was asigni®cant (P5 0.001) di�erence in germinationbetween the three fungal �. spp., with f. sp. hordei conidiagerminating slightly less well (approx. 62%) than either�. spp. avenae (approx. 70%) or tritici (approx. 68%).These small di�erences presumably re¯ect uncontrolledvariation in inoculum quality between �. spp. However,there was no interaction between conidium location and�. spp. indicating consistency of these small e�ects withinthe experiment.

For germinated conidia, there was only a very lowfrequency of germling abnormality in all locations andmost germlings formed apparently normal appressoria.For all �. spp. and all locations, abnormality was neverseen for more than 6% of germlings even if the germlingswere in contact with epidermal cells containing haustoria

formed by f. sp. avenae colonies that had been removed.

Thus, in complete contrast to experiments wherecolonies were undisturbed, the location of conidia onepidermal cells that had been within the boundaries ofcolonies established by the ®rst inoculum had no e�ect oneither their germination or on subsequent germlingdevelopment.

Haustorium formation. Although location had no e�ect onappressorium formation, it had a dramatic e�ect on thepercentage of germlings that penetrated epidermal cellssuccessfully to produce haustoria (Fig. 14). When f. sp.avenae germlings attacked control leaves or epidermal cellsdistant from cells containing a haustorium formed earlierby an f. sp. avenae colony, approx. 42% formed haustoria.However, when attack was on a cell containing ahaustorium formed earlier, 92% did so. Thus, cells thatcontained colony haustoria were conditioned to a highlevel of ``accessibility'' to subsequent attack by f. sp. avenae.

The presence of f. sp. avenae colony haustoria inepidermal cells of oat also conditioned them to a highlevel of accessibility to subsequent attack by germlings of�. spp. tritici and hordei. Thus, 90% of f. sp. tritici and58% of f. sp. hordei germlings penetrated successfully andformed haustoria when attacked cells contained f. sp.avenae colony haustoria. By contrast, neither of theseinappropriate �. spp. formed haustoria in control oatleaves or in cells distant from cells containing f. sp. avenae

DISCUSSION

We were very surprised by our initial ®nding from Selmaoat that germination and germling development were

markedly and adversely a�ected when B. graminis conidia

Germ

inatedconidia

form

inghaustoria(%

)

Control

Distant

Outer

zone

Inner

zone

Conidium location

relative to colony

10

20

30

50

0

40

(a)

a a a

b

10

20

30

50

0

40

(b)

NA NAa

b

(c)

NA NAa

b

10

20

30

50

0

40

FIG. 12. Percentages of B. graminis �. spp. hordei (a), avenae (b)and tritici (c) conidia that had germinated on second-formed cv.Pallas barley leaves, and that formed morphologically normalappressoria from which plant epidermal cells were penetratedsuccessfully, resulting in haustorium formation, 24 h afterinoculation onto healthy control leaves or onto leaves bearingB. graminis f. sp. hordei colonies derived from a ®rst inoculation72 h earlier. All material was ®xed 24 h after the secondinoculation. On controls ( ), data are based on counts of 50randomly selected germlings, while on leaves bearing colonies,counts were made for 50 germlings located distant from colonies( ), 50 within the outer zone of colonies ( ) and 50 within theinner zone of colonies ( ), on each of four replicate leaves in allcases. Percentage data were transformed to angles for analysis ofvariance. Where columns are headed by di�erent letters, LSDtests applied to angle-transformed data indicated a signi®cantdi�erence (P5 0.05) between means within a data set. NA:

data � zero, therefore excluded from analyses.

Hat

Hat

Ct

CtAppt Appt

Apph

Ch

Hh

FIG. 13. Light micrograph of two B. graminis f. sp. triticigermlings located within the inner zone of a B. graminis f. sp.hordei colony. The colony was established from a f. sp. hordeiconidium deposited 72 h earlier than the f. sp. tritici conidia.The specimen was ®xed 24 h after deposition of the f. sp. triticiconidia. Both f. sp. tritici germlings produced normal appres-soria from which haustoria were formed. Apph � primaryappressorium of f. sp. hordei colony; Appt � appressorium of f.sp. tritici germling; Ch � conidium of f. sp. hordei, Ct � conidiumof f. sp. tritici. Hh � primary haustorium of f. sp. hordei colony;

Hat � primary haustorium of f. sp. tritici germling.

Inhibition of B. graminis development 223

were deposited within established oat mildew colonies.This was because a considerable body of evidenceindicates that deposition of B. graminis conidia, andconidia of other powdery mildew fungi, close to anestablished B. graminis colony can greatly enhance thechances of these conidia producing successful infections.

This evidence comes from investigations of wheat andbarley infected with their appropriate f. sp. and sub-sequently inoculated with an inappropriate f. sp. ofB. graminis or with less closely related Erysiphaceae. ThusMoseman and his co-workers [30 , 31] found that wheatcould be attacked by B. graminis f. sp. hordei if leaves werealready infected with a virulent isolate of B. graminis f. sp.tritici or if the two fungi were inoculated simultaneously.Tsuchiya and Hirata [37] inoculated B. graminis f. sp.hordei-infected barley with f. sp. tritici, a B. graminis isolatefrom Agropyron, or with powdery mildew fungi from 49di�erent dicotyledonous species; the fungi includedspecies from the genera Erysiphe, Sphaerotheca, Podosphaera,Microsphaera and Uncinula. Of the 51 fungi tested, 45 wereable to infect the barley and 30 formed sporulatingcolonies. Tsuchiya and Hirata [37] noted that theseinfections almost invariably occurred close to established f.sp. hordei colonies and detailed histological studies of theB. graminis isolate from Agropyron and of Sphaerothecafuliginea showed that successful infections by these fungiwere almost invariably established in barley epidermalcells containing B. graminis f. sp. hordei haustoria or inneighbouring cells. Similarly, Ouchi et al. [33] concluded

that induced `accessibility' of barley to attack by wheat

Germ

inatedconidia

form

inghaustoria(%

)

Control

Distant

Overlying

Location of appressorium in relation to

epidermal cells containing haustoria formed by

colony

20

40

60

100

0

80

Overlying

Overlying

avenae

B. graminis f.sp.

tritici hordei

FIG. 14. Percentages of B. graminis f. sp. avenae, tritici and hordei germlings that formed appressoria and penetrated Selma oatepidermal cells to form a haustorium 36 h after inoculation. Data are for germlings on healthy control leaves, or on leaves that hadsupported f. sp. avenae colonies derived from a ®rst inoculation applied 96 h before test conidia were inoculated, but from which allsuper®cial fungal structures (conidia, hyphae) were removed immediately before inoculation with test conidia. On controls ( )data are based on counts of 50 randomly selected germlings, while on leaves that had supported colonies data are based on counts of25 germlings whose appressoria were on epidermal cells that were distant from cells containing haustoria formed earlier by colonies( ), and 25 germlings with appressoria overlying cells containing haustoria formed earlier by colonies (h), on each of four replicateleaves in each case. Data for �. spp. tritici and hordei on healthy oat control leaves and on epidermal cells distant from haustorium-

containing epidermal cells were zero in all cases (not shown).

224 T. L. W. Carver et al.

mildew and S. fuliginea was a phenomenon expressedlocally in the immediate vicinity of established barleymildew colonies. More recent and detailed studies byKunoh and his co-workers [22 , 24 , 25] again showed thatinduced accessibility of barley coleoptile cells to attack bythe non pathogen Erysiphe pisi, is an extremely localizede�ect dependent on prior or near simultaneous infectionby B. graminis f. sp. hordei. Thus, many studies have shownthat barley and wheat can be rendered susceptible topowdery mildew fungi that are normally unable to infectthem, if they are previously infected by a pathogenicpowdery mildew fungus. However, this depends upon thenormally non-pathogenic fungus being in close proximityto established infections of the pathogenic B. graminis. Thisin turn clearly suggests that proximity to an establishedB. graminis colony does not greatly impede germination ordevelopment of later-applied propagules and in factfavours their ability to infect. Hence, we were surprisedto ®nd that this was not the case when B. graminis conidiawere deposited within oat mildew colonies.

Nevertheless, our observations of the behaviour of threedi�erent B. graminis �. spp. deposited within the innerzone of established barley mildew colonies were broadly inline with expectation. Although germination by conidia ofall three �. spp. was somewhat depressed within thesecolonies, and the frequency of abnormal long germ tubes

was slightly increased within the inner colony zone, the

majority of germinated conidia di�erentiated morpho-logically normal appressoria. The functionality of theseappressoria was demonstrated by the fact that a higherpercentage were associated with successful penetrationand haustorium formation than where conidia wereremote from established colonies. Indeed, our ®ndingswere comparable to earlier studies of powdery mildewedbarley and wheat [30 , 31 , 37] in showing that inappro-priate �. spp., normally unable to infect healthy barley,frequently formed haustoria if they attacked within theinner zone of barley powdery mildew colonies. The abilityof the inappropriate �. spp. to infect barley, and increasedpenetration frequency by B. graminis f. sp. hordei, waslargely lost if attacks were made within the outer zone ofcolonies or if they were distant from them. Theseobservations support the conclusion from earlier histo-logical studies that prior infection of barley by acompatible isolate decreases epidermal cells' ability toresist attack [27±29]. However, also in agreement withthese studies, this e�ect appears to be localized to cellscontaining pre-formed haustoria and to neighbouringcells.

When conidia of all three �. spp. were deposited withinestablished oat mildew colonies germination percentageswere depressed. This was similar to, but even moremarked than the situation seen within barley mildew

colonies. We know from previous work [8] that the

a

presence of a haustorium within an oat epidermal cellshould have induced accessibility to subsequent attack byB. graminis f. sp. avenae. However, we could not be surefrom our observations of conidia deposited withinestablished oat mildew colonies whether the plant cellshad been a�ected. This was because very few germinatedconidia produced morphologically normal appressoria,and even where appressorial germ tubes formed anapparently normal, hooked apical lobe, we could not besure if the structure was functional. Thus, very fewgermlings were capable of attempting infection and thisalone could explain the very low frequency of haustoriumformation seen within oat mildew colonies.

Nevertheless, there was some evidence from Selma oat(Fig. 8) that accessibility to B. graminis f. sp. avenae mayhave been induced within the outer zone of establishedcolonies, and some evidence of induced accessibility to f.sp. tritici within oat mildew colonies established onMaldwyn oat. However, clear evidence for inducedaccessibility in oat epidermal cells was provided onlywhen we removed the super®cial structures of the oatmildew colonies prior to applying the second inoculum.Thus, when germlings of all three �. spp. attackedepidermal cells containing a haustorium formed earlierby a oat mildew colony, a substantial proportion weresuccessful in penetrating the cells and establishing afunctional haustorium. For f. sp. avenae, this was aquantitative e�ect where the frequency of successfulpenetration was increased relative to control leaves orepidermal cells distant from haustorium-containing cells.For �. spp. tritici and hordei the e�ect was qualitative inthe sense that these fungi were virtually unable to infectoat cells that did not contain a f. sp. avenae haustorium.

The available evidence clearly shows that priorinfection of cereal epidermal cells by B. graminis hasdramatic, albeit localized, consequences for the ability ofcells to resist subsequent attack by the fungus. Thesee�ects a�ect not only the outcome of attacks bycompatible isolates of the appropriate f. sp., but extendto permit invasion of cells by inappropriate �. spp. andnon pathogenic species. The basis of induced changes inaccessibility remain unknown.

At present, we have no idea what causes suppression ofgermination by conidia deposited within establishedpowdery mildew colonies. However, it seems that similarfactors operate in colonies of both barley and oat powderymildew, and the factors were e�ective against all three �.spp. that we tested. It is a truism that powdery mildewconidia do not germinate while they are withinthe developing spore chain attached to the motherconidiophore, but that they germinate rapidly afterrelease. It seems possible that the same factor(s) suppressgermination by conidia deposited within an establishedcolony. The factor(s) are not particularly mobile as the

Inhibition of B. gr

suppressive e�ect was usually greater when conidia lay

within the colony inner zone. The inner zone recognizedat the time of ®xation probably represented (approxi-mately) the limits of hyphal development extant at thetime when second inoculations were made (Fig. 2)suggesting that conidia were close to, or in contactwith, hyphae from the time of their deposition. It is notclear however, whether spores that failed to germinatewere actively and rapidly killed by factors released by thecolony or whether they survived for some time (perhapsuntil ®xation) although processes leading to germ tubeemergence were inhibited. Further studies should be ableto distinguish these possibilities and in turn shed light onthe likelihood that similar factor(s) suppress the germina-tion of conidia attached to the spore chain whichobviously can remain alive for some time before release.

Abnormal germling development was extremely fre-quent when conidia of all B. graminis �. spp. germinatedwithin oat mildew colonies. By comparison, abnormal-ities were rarely seen within barley mildew colonies, evenwhere germlings of the second inoculum were in veryclose contact with hyphae (Fig. 11). The latter ®ndingappears to contradict Hirata's [18] observations of thebarley mildew system. Hirata [18] stated that . . . ``when(barley) leaves bearing old (barley) mildew colonies areinoculated (with B. graminis f. sp. hordei conidia), conidiathat fall on or near the peripheral part of the coloniesgerminate and develop long septate germ tubes. Thegerm tube does not generally form an appressorium andthe primary haustorium''. Hirata's drawings of thesegerm tubes resemble the abnormal germlings we sawwithin the inner zone of oat mildew colonies. However,Hirata did not quantify the frequency of this occurrence,and he did not indicate exactly the age of the coloniesinvolved in his observations. It is possible that the abilityof powdery mildew colonies to generate/release chemicalsthat cause abnormal germling development is a functionof their age. Thus, although our oat mildew colonies hadobviously released the factor(s) within 4 days ofinoculation, barley mildew colonies of this age had notdone so. If we had incubated our barley mildew coloniesfor longer before we applied the second inoculum, wemay have detected e�ects on germling development.

Hirata [18] also found that anastomosis occasionallyoccurred when abnormal long germ tubes encounteredhyphae. We never saw this. He went on to show that suchanastomosis could occur between conidial germ tubes ofpowdery mildew from Agropyron (a non-pathogen ofbarley) and hyphae of established barley mildew colonies.From Tsuchiya and Hirata's [37] further discussion ofthese studies, it appears that following anastomosis withbarley powdery mildew, the powdery mildew fromAgropyron derived nutrients from the barley mildewcolony. This allowed the Agropyron fungus to develophyphae, conidiophores and conidia capable of re-

minis development 225

infecting Agropyron. Clearly, Hirata and his co-workers'

a

studies have important implications with respect to theintimacy of genetic and physiologic association resultingfrom anastomoses formed between fungal individualsboth within and between �. spp. of B. graminis andpossibly between genera of the Erysiphales. There isimmense scope to pursue these studies.

Presumably the e�ective factor(s) produced by oatmildew colonies that caused abnormal germling devel-opment in our experiments must be di�erent from thosethat suppress germination. This conclusion arises fromour ®nding that germination frequency was consistentlyreduced within colonies of both f. sp. hordei and f. sp.avenae, whereas our barley powdery mildew colonies hadlittle if any detectable e�ect on germling development.Again, however, we have no idea what factor(s) maycause abnormal germling development. We believe thefactor(s) has limited mobility, because, as with suppres-sion of germination, greater e�ect was evident within theinner than the outer colony zone. Nevertheless, directcontact with a hypha was not necessary for abnormalgermling development (Figs 4 and 7). Therefore, thefactor(s) had some ability for movement over the leafsurface. Where conidia germinated, the factor(s) did notappear to inhibit PGT function since germlings oftendeveloped subsequent germ tubes that elongated sub-stantially. Such elongation occurs only if PGTs functioncorrectly in recognising inductive features of the plantsurface [4±7 , 38]. The factor's major e�ect was to inhibitthe later stages of appressorial germ tube di�erentiation.Thus, elongating germ tubes remained hypha-like, failedto swell and di�erentiate a hooked apical lobe, and oftendeveloped peculiar branches and multiple septa. Suchabnormalities could result from failure of the elongatingtube itself to recognize plant surface features [4±7]. Thisseems unlikely, however, because the e�ective factor(s)did not a�ect the PGT's ability to recognize suchfeatures. Alternatively, it may be that the oat mildewcolonies release factor(s) that block intracellular signal-ling or signal transduction pathways that regulateappressorial formation. Temporal regulation of suchpathways is known to be associated with maturationand hooking of the appressorial germ tube; pharmaco-logical inhibitors of these pathways can prevent appres-sorium formation and lead to the development ofundi�erentiated, hypha-like elongated germ tubes [16 ,17 , 19]. Perhaps, our oat powdery mildew coloniesproduced chemicals with similar e�ects. Whatever theexplanation, it is intriguing that the e�ective factor(s)disrupt germling development but are presumablyharmless to the established colony.

Our experiments suggest strongly that the factorssuppressing B. graminis germination and causing abnor-mal germling development were directly associated withpresence of oat mildew hyphae, and are not due to

226 T. L. W. C

indirect e�ects mediated via modi®cation of the host oat

plant's physiological or biochemical status. Thus, whensuper®cial hyphae were removed from oat leaves, sporesdeposited on haustorium-containing cells germinated andnormal, functional appressoria developed with at leastthe same frequency as on healthy leaves. The inducedaccessibility of haustorium-containing cells showed thatthey remained radically altered, at least in terms of theiraccessibility, by the presence of pre-formed haustoria. Iffactors inhibitory to germination and germling develop-ment were generated within haustorium-containing cells,it seems highly likely that removal of the epicuticular waxlayer (that inevitably occurs during cellulose acetatestripping [7 , 10 , 11, 27]) would have facilitated theirmovement to the cell surface and increased inhibitorye�ects. This was clearly not so. These ®ndings againappear to contradict Hirata's [18] observations. Hiratafound that abnormal germlings often developed whenbarley mildew conidia were inoculated onto a leaf fromwhich older colonies had previously been ``scraped''away. We assume that ``scraping'' involved a physicalprocedure that could have ruptured the cell walls ofsuper®cial fungal structures (hyphae, conidiophores, etc.)releasing their cell contents onto the leaf surface prior toapplication of test conidia. If the factor(s) that causegermling abnormality are generated within cells of thecolony, these may well have been released to remain asresidue on the leaf surface following ``scraping'' inHirata's experiment. Our technique would have removedall such residues along with the leaf epicuticular wax. Wesuggest this explains the apparent di�erence between ourobservations and Hirata's.

Our current work indicates that when B. graminisconidia are deposited within an established colony theirgermination is suppressed. Our experiments and Hirata'sseminal studies show that where conidia do germinatewithin a colony, their subsequent development can bedisrupted so that they do not form appressoria orhaustoria, and are therefore incapable of independentexistence and reproduction. Why should establishedcolonies produce factors that inhibit germination anddisrupt germling development? In situations with geneti-cally diverse host and pathogen populations, there couldbe obvious selective advantages for an established colonyin preventing potential competitors from infecting nearbyplant epidermal cells which have been rendered highlyaccessible to penetration by the presence of haustoriaformed by the established colony. These competitors maybe less-well adapted to parasitize a particular plantgenotype, and may even be avirulent. If this were so,avirulent individuals could kill attacked cells and deprivethe adapted colony of potential feeding sites. On the otherhand, some conidia produced by a colony are likely to bedeposited close to that colony; suppressing their develop-ment would prevent competition for nutrients in ®nite

rver et al.

local supply within a niche already occupied by that

a

pathogen genotype. Similar e�ects are likely to applywithin agricultural crops although their selectiveadvantage may be less marked. According to Hirata's[18] observations, conidia that germinate and formabnormal germ tubes at the periphery of an establishedcolony, may be able to form anastomoses with hyphae ofthat colony and, in e�ect, parasitize the colony by using itsresources. It remains to be determined how frequently thishappens, but it seems unlikely that the frequency ofanastomosis compensates in any signi®cant way for thesuppressive e�ects established colonies have on germina-tion and appressorium formation by potential competi-tors.

Currently we are attempting to isolate the factorsresponsible for inhibiting spore germination and germlingdevelopment within oat mildew colonies. It is conceivable

Inhibition of B. gr

that this may reveal novel means of disease control.

We formally acknowledge the pioneering research con-tribution made by Professor K. Hirata (latterly Amano)that laid the foundations for the studies reported here,and his many other painstaking works that havecontributed so much to our current understanding ofpowdery mildew biology. The research reported here wassupported by MAFF (U.K.) under Project CE0154 andin part by the Danish Agricultural and VeterinaryResearch Council. The Institute of Grassland andEnvironmental Research is supported by the Biotechnol-

ogy and Biological Sciences Research Council.

REFERENCES

1. Aist JR, Bushnell WR. 1991. Invasion of plants bypowdery mildew fungi, and cellular mechanisms ofresistance. In: Cole GT, Hoch HC, eds. The Spore andDisease Initiation in Plants and Animals. New York, U.S.A.:Plenum Press, 321±345.

2. Carver TLW, Bushnell WR. 1983. The probable role ofprimary germ tubes in water uptake before infection byErysiphe graminis. Physiological Plant Pathology 23: 229±240.

3. Carver TLW, Carr AJH. 1978. The early stages of mildewcolony development on susceptible oats. Annals of AppliedBiology 89: 201±209.

4. Carver TLW, Ingerson SM. 1987. Responses of Erysiphegraminis germlings to contact with arti®cial and hostsurfaces. Physiological and Molecular Plant Pathology 30:359±372.

5. Carver TLW, Ingerson SM, Thomas BJ. 1996. In¯uencesof host surface features on development of Erysiphe graminisand Erysiphe pisi. In: Kerstiens G, ed.. Plant CuticlesÐAnIntegrated Functional Approach. Oxford, U.K.: Bios Scienti®cPublishers, 255±266.

6. Carver TLW, Ingerson-Morris SM, Thomas BJ, ZeyenRJ. 1995. Early interactions during powdery mildewinfection. Canadian Journal of Botany 73(suppl. 1):

S632±S639.

7. Carver TLW, Kunoh H, Thomas BJ, Nicholson RL.1999. Release and visualization of the extracellular matrixof conidia of Blumeria graminis. Mycological Research 103:547±560.

8. Carver TLW, Lyngkjñr MF, Neyron L, Strudwicke CC.1999. Induction of cellular accessibility and inaccessibilityand suppression and potentiation of cell death in oatattacked by Blumeria graminis f. sp. avenae. Physiological andMolecular Plant Pathology 55: 183±196.

9. Carver TLW, Phillips M. 1982. E�ects of photoperiod andlevel of irradiance on production of haustoria by Erysiphegraminis f. sp. hordei. Transactions of the British MycologicalSociety 79: 207±211.

10. Carver TLW, Thomas BJ. 1990. Normal germling devel-opment by Erysiphe graminis on cereal leaves freed ofepicuticular wax. Plant Pathology 39: 367±375.

11. Carver TLW, Thomas BJ, Ingerson-Morris SM. 1995.The surface of Erysiphe graminis and the production ofextracellular material at the fungus-host interface duringgermling and colony development. Canadian Journal ofBotany 73: 272±287.

12. Carver TLW, Williams O. 1980. The in¯uence ofphotoperiod on growth patterns of Erysiphe graminis f. sp.hordei. Annals of Applied Biology 94: 405±414.

13. Carver TLW, Zeyen RJ, Robbins MP, Dearne GA. 1992.E�ects of the PAL inhibitor, AOPP, on oat, barley, andwheat cell responses to appropriate and inappropriateformae speciales of Erysiphe graminis DC. Physiological andMolecular Plant Pathology 41: 397±409.

14. Francis S, Dewey M, Gurr SJ. 1996. The role of cutinasein germling development in Erysiphe graminis. Physiologicaland Molecular Plant Pathology 49: 201±211.

15. Genstat 5: Procedure Library Manual Release PL 10.1997. Payne RW, Morgan GW and Arnold GM, eds.Oxford, U.K.: Numerical Algorithms Group, 371.

16. Hall AA, Bindslev L, Rouster J, Rasmussen SW, OliverR, Gurr SJ. 1999. The involvement of cAMP and PKA inappressorial di�erentiation in Erysiphe graminis f. sp. hordei.Molecular Plant Microbe Interactions 12: 960±968.

17. Hall AA, Gurr SJ. 2000. Initiation of appressorial germtube di�erentiation and appressorial hooking: distinctmorphological events regulated by cAMP-signalling inBlumeria graminis f. sp. hordei. Physiological and MolecularPlant Pathology 56: 39±46.

18. Hirata K. 1967. Notes on haustoria, hyphae and conidia ofthe powdery mildew fungus of barley. Memoirs of theFaculty of Agriculture, Niigata University 6: 207±259.

19. Kinane JS, Dalvin S, Bindslev L, Hall AA, Gurr SJ,Oliver R. 2000. Evidence that cAMP ¯uxes control earlydevelopment of barley mildew. Molecular Plant±MicrobeInteractions 13: 494±502.

20. Kunoh H, Hayashimoto A, Harui M, Ishizaki H. 1985.Induced susceptibility and enhanced resistance at thecellular level in barley coleoptiles. I. The signi®cance oftiming of fungal invasion. Physiological Plant Pathology 27:43±54.

21. Kunoh H, Katsuragawa N, Yamaoka N, HayashimotoA. 1988. Induced accessibility and enhanced inaccessi-bility at the cellular level in barley coleoptiles. III. Timingand localization of enhanced inaccessibility in a singlecoleoptile cell and its transfer to an adjacent cell.Physiological and Molecular Plant Pathology 33: 81±93.

22. Kunoh H, Komura T, Kobayashi I, Yamaoka N. 1990.Induced accessibility and enhanced inaccessibility at the

minis development 227

cellular level in barley coleoptiles. VII. Enhancement of

a

inaccessibility at the prepenetration stage of a nonpatho-gen. Physiological and Molecular Plant Pathology 37: 399±407.

23. Kunoh H, Kuroda K, Hayashimoto A, Ishizaki H. 1986.Induced susceptibility and enhanced resistance at thecellular level in barley coleoptiles. II. Timing and locationof induced susceptibility in a single coleoptile cell and itstransfer to an adjacent cell. Canadian Journal of Botany 64:889±895.

24. Kunoh H, Kuroda K, Toyoda K, Yamaoka N, Kobaya-shi I. 1991. Induced accessibility and enhanced inaccessi-bility at the cellular level in barley coleoptiles. IX.Accessibility induced by Erysiphe graminis which promotesthe subsequent infection of challenging E. graminis. Annalsof the Phytopathological Society of Japan 57: 57±60.

25. Kunoh H, Toyoda K, Yamaoka N, Kobayashi I. 1989.Induced accessibility and enhanced inaccessibility at thecellular level in barley coleoptiles. V. Duration of stimulusby a non-pathogen in relation to enhanced inaccessibility.Physiological and Molecular Plant Pathology 35: 507±518.

26. Kunoh H, Tsuzuki T, Ishizaki H. 1978. Cytologicalstudies of early stages of powdery mildew in barley andwheat leaves. IV. Direct ingress from super®cial primarygerm tubes and appressoria of Erysiphe graminis hordei onbarley leaves. Physiological Plant Pathology 13: 327±333.

27. Lyngkjñr MF, Carver TLW. 1999a. Induced accessibilityand inaccessibility to Blumeria graminis f. sp. hordei in barleyepidermal cells attacked by a compatible isolate. Physio-logical and Molecular Plant Pathology 55: 151±162.

28. Lyngkjñr MF, Carver TLW. 1999b. Modi®cation of mlo5resistance to Blumeria graminis attack in barley as aconsequence of induced accessibility and inaccessibility.Physiological and Molecular Plant Pathology 55: 163±174.

29. Lyngkjñr MF, Carver TLW. 2000. Conditioning of cerealleaf cellular defence responses to powdery mildew by prior

228 T. L. W. C

attack. Molecular Plant Pathology 1: 41±50.

30. Moseman JG, Greeley LW. 1964. Predisposition of wheatby Erysiphe graminis f. sp. tritici to infection by Erysiphegraminis f. sp. hordei. Phytopathology 54: 618.

31. Moseman JG, Scharen AL, Greeley LW. 1965. Propa-gation of Erysiphe graminis f. sp. tritici on barley and Erysiphegraminis f. sp. hordei on wheat. Phytopathology 55: 92±96.

32. Nielsen KA, Nicholson RL, Carver TLW, Kunoh H,Oliver RP. 2000. First Touch: An immediate response tosurface recognition in conidia of Blumeria graminis.Physiological and Molecular Plant Pathology 56: 63±70.

33. Ouchi S, Oku H, Hibino C. 1976. Localization of inducedresistance and susceptibility in barley leaves inoculatedwith the powdery mildew fungus. Phytopathology 66:901±905.

34. Ouchi S, Oku H, Hibino C, Akiyama I. 1974. Inductionof accessibility and resistance in leaves of barley by someraces of Erysiphe graminis. Phytopathologische Zeitschrift 79:24±34.

35. Pryce-Jones E, Carver TLW, Gurr SJ. 1999. The roles ofcellulase enzymes and mechanical force in host penetra-tion by Erysiphe graminis f. sp. hordei. Physiological andMolecular Plant Pathology 55: 175±182.

36. Suzuki S, Komiya Y, Mitsui T, Tsuyumu S, Kunoh H,Carver TLW, Nicholson RL. 1998. Release of cell walldegrading enzymes from conidia of Blumeria graminis onarti®cial substrata. Annals of the Phytopathological Society ofJapan 64: 160±167.

37. Tsuchiya K, Hirata K. 1973. Growth of various powderymildew fungi on barley leaves infected preliminarily withthe barley mildew fugus. Annals of the PhytopathologicalSociety of Japan 39: 399±403.

38. Wright AJ, Carver TLW, Thomas BJ, Fenwick NID,Kunoh H, Nicholson RL. 2000. The rapid and accuratedetermination of germ tube emergence site by Blumeriagraminis conidia. Physiological and Molecular Plant Pathology

rver et al.

57: 281±301.