In vitro leaf inoculation studies as an indication of tree foliage susceptibility to Phytophthora ramorum in the UK


*To whom correspondence should be addressed.


Leaves of 11 coniferous and 23 broad-leaved tree species important to UK forestry were tested for their susceptibility to the quarantine pathogen Phytophthora ramorum using a detached leaf assay. Two European and two USA isolates were used. Wounded and unwounded leaves were dipped in zoospore suspensions during summer; conifers were also tested in winter. Successful infection of tissue and amount of necrosis were assessed. Highly susceptible broad-leaved hosts included Aesculus hippocastanum, Fraxinus excelsior, Quercus ilex, Ulmus procera and, to a lesser extent, Castanea sativa, Q. cerris and Q. petraea, together with Umbellularia californica and rhododendrons. Acer pseudoplatanus, Alnus glutinosa, Carpinus betulus, Corylus avellana, Fagus sylvatica, Prunus avium, Q. robur, Q. rubra and Q. suber showed consistently low susceptibility. Conifer species including Abies procera, Picea abies, P. sitchensis, Pseudotsuga menziesii, Sequoia sempervirens and Tsuga heterophylla were also susceptible. Pseudotsuga menziesii and A. procera were severely affected. Pinus contorta, P. nigra var. maritima and P. sylvestris were virtually resistant, while Taxus baccata was only slightly affected. Increased necrosis was apparent on leaves that were wounded prior to inoculation. These results extend the known range of trees that P. ramorum is able to attack and confirm its relative host-nonspecificity.


The chromistan pathogen Phytophthora ramorum (order: Pythiales, family: Pythiaceae) causes high mortality of native trees, especially Lithocarpus densiflorus (tan oak) and species of black oak [viz. Quercus agrifolia (coastal live oak), Q. kelloggii (Californian black oak) and Q. parvula var. shrevei (Shreve's oak)] in the coastal fog belts of California and parts of Oregon (Rizzo et al., 2002). On mature trees, especially oaks in the USA, this pathogen mostly causes bleeding stem cankers that can lead to tree death; hence the disease is popularly known as sudden oak death (SOD) in America. However, P. ramorum has also been found on understorey shrubs and herbaceous plants in California (Davidson et al., 2003; Garbelotto et al., 2003), causing foliage necrosis and shoot tip dieback. In Europe and the UK, infected ornamental shrubs with foliage and shoot symptoms have been found mostly in nursery environments (Werres et al., 2001; Delatour et al., 2002; Orlikowski & Szkuta, 2002; De Merlier et al., 2003; Lane et al., 2003; Moralejo & Werres, 2003; Pintos Varela et al., 2003). It is feared that this very destructive pathogen may spread to local indigenous hosts and ecosystems. To date, records show that species of Rhododendron and Viburnum are most commonly affected, but Camellia, Hamamelis, Kalmia, Leucothoë, Pieris and even a few containerized Taxus baccata (yew) have been recorded as hosts (Anonymous, 2003; Brokenshire, 2003; Brown, 2003; Lane et al., 2003, 2004; Werres & De Merlier, 2003; Beales et al., 2004). Infected rhododendrons are also found in woodland areas in the Netherlands and the UK. Until recently, P. ramorum had not been reported on any mature trees in Europe or the UK (Moralejo & Werres, 2003). However, the first discoveries of this pathogen on trees growing in woods or gardens in the UK and the Netherlands were made in October 2003 (Brasier et al., 2004). In the UK, bleeding cankers have been found on Aesculus hippocastanum (horse chestnut), Fagus sylvatica (beech), Quercus cerris (Turkey oak) and Q. falcata (southern red oak) (Brasier et al., 2004), while in Holland an infected Quercus rubra (northern red oak) has been found (Netherlands Plant Protection Services, personal communication).

Foliar infections of certain understorey plants play a crucial role in the development of disease epidemics (Garbelotto et al., 2003). In California, P. ramorum sporulates on shrubs such as Umbellularia californica (Californian bay laurel) and Arbutus menziesii (madrone) providing key sources of inoculum that drive epidemics (Davidson et al., 2002a; Garbelotto et al., 2003; Davidson, 2004). However, in Oregon, sporulation on foliage of tan oaks is thought to be a main source of inoculum (E. M. Hansen, personal communication). Phytophthora ramorum has not yet been found sporulating on tree trunks with bleeding cankers.

In view of the damage occurring in the USA, and the recent discoveries of this pathogen on trees in the UK, there is concern that it may become widespread in local woodlands. Therefore, information on the potential host range is needed to make an assessment of the risk that could be posed by P. ramorum. Detached-foliage assays have been used by Hansen et al. (2002, 2005), Moralejo & Hernandez (2002) and Parke et al. (2002), amongst others, to assess susceptibility to P. ramorum in vitro in order to get an indication of the tree species most likely to suffer shoot and foliar infection. Thus, in the present study, leaves of 11 coniferous and 23 broad-leaved tree species were tested for infection and degree of necrosis caused by isolates of P. ramorum from both the USA and Europe. Although the USA isolates are morphologically similar to those from Europe (Garbelotto et al., 2003), there is evidence of genetic and adaptive differences which indicate that the European and USA isolates comprise different subpopulations (Brasier, 2003; Ivors, 2004). These differences could affect the host range and severity of the disease; hence the need to include isolates from both sources. The results of these tests are discussed and related to recent field observations.

Materials and methods

The foliage of coniferous and broad-leaved hosts was tested using a detached-leaf-dip assay adapted from a method devised by Parke et al. (2002). Phytophthora ramorum is a quarantine pathogen in the UK and all the work described in this study was carried out in licensed quarantine laboratories. Conditions in which the work was carried out are described in Defra quarantine licence number PHL 182/4547(08/2003). All inoculum and inoculated plant material were destroyed by autoclaving, and equipment was decontaminated in the quarantine laboratory immediately after use and then sterilized for re-use.

Isolates and inoculum production

Four isolates of the pathogen were used in this study and control leaves were dipped in sterile water. Isolate P1376 (from Viburnum tinus with basal stem cankers and wilt) and P1399 (from Rhododendron cv. Albert Schweitze exhibiting leaf necrosis and stem-tip dieback), both A1 mating type, originated from diseased nursery plants in the UK. The other two were American isolates: P1349 from Rhododendron in California and P1403 from Vaccinium ovatum in Oregon, both of the A2 mating type. Isolates were cultured on carrot agar (Brasier, 1967, 1969) and incubated at 18–20°C in continuous day light (60-W bulbs; Daylight Company, UK) suspended 30 cm above the plates) for 14 days to produce sporangia. To obtain zoospores, cultures were flooded with 5 mL of sterile water and sporangia were dislodged into the water by rubbing the surface of the culture with a sterile bent glass rod. The liquid was poured off the plates and collected in a sterile beaker which was placed firstly in a refrigerator at 7°C for 1 h, then returned to room temperature (20°C) for a further 75 min to induce zoospore release (Parke et al., 2002). The resulting suspension for each isolate was vacuum-filtered through a 20-µL millipore filter using a Buchner funnel attached to a 0·5-L vacuum flask to remove residual pieces of mycelium, chlamydospores and empty sporangia, thereby obtaining pure zoospore suspensions which were transferred to sterile beakers. A 0·3-mL subsample of each pure zoospore suspension was placed in a separate microtube and shaken on a Vortex mixer (90 s) to initiate zoospore encystment so that zoospore counts could be made using a haemocytometer. The concentration of each suspension was adjusted to 2–4 × 105 zoospores mL−1. Thirty millilitres of the corrected suspension of each isolate was dispensed into individual sterile 50-mL glass beakers and plant material was inoculated as described below. To determine viability of inoculum, three aliquots each of 1 and 10 µL of the suspension of each isolate were spread onto Phytophthora-selective agar (SMA) (Brasier & Kirk, 2002) before dipping the leaves. In case there was an effect of leaf dipping on inoculum viability or dilution of inoculum, spread plates were made as described above after all the leaves had been dipped. Plates were incubated at 20°C for 3–4 days. The number of colony-forming units was counted and zoospore viability checked.

Host plants and preparation of plant material

Host plants are listed in Table 1 and the same genotypes were used, where possible, for the experiments. For the conifer species, the current season's growth shoots approximately 10 cm long and bearing needles were used, whereas fully expanded leaves of the broad-leaved trees were selected. All plant material was obtained from mature healthy trees (except Californian bay laurel, where nursery trees were used) at the Alice Holt Research Station in Surrey, UK, and was brought to the laboratory immediately. The cut ends of the stems were placed in water overnight. Before inoculation, leaves were rinsed with sterile water and placed on sterile paper towels to air-dry. The midpoint of each shoot (conifers) or leaf (broad-leaved species) was marked with a permanent ink marker pen. Rhododendrons were included as positive controls in both the broad-leaved and conifer experiments to confirm the pathogenicity of the isolates.

Table 1.  Host species used to assess foliage susceptibility to Phytophthora ramorum in detached leaf assays
Botanical nameCommon nameEcological status in the UK
  • a

    Tested in the first series of broad-leaf experiments (experiments 1a and 1b).

  • b

    Tested in all experiments.

Broad-leaf hosts
 Acer pseudoplatanusaSycamoreNaturalized
 Aesculus hippocastanumHorse chestnutNaturalized
 Alnus glutinosaAlderNative
 Betula pendulaaSilver birchNative
 Carpinus betulusaHornbeamNative
 Castanea sativaaSweet chestnutNaturalized
 Corylus avellanaHazelNative
 Eucalyptus gunniiCider gumExotic
 Fagus sylvaticaaBeechNative
 Fraxinus excelsioraAshNative
 Ilex aquifoliumHollyNative
 Populus tremulaaAspenNative
 Prunus aviumWild cherry or GaeanNative
 Quercus cerrisaTurkey oakNaturalized
 Quercus ilexaHolm or holly oakNaturalized locally
 Quercus petraeaSessile or Durmast oakNative
 Quercus roburaEnglish or common oakNative
 Quercus rubraaRed oakExotic
 Quercus suberCork oakExotic
 Rhododendron catawbiensebRhododendronExotic
 Tilia cordataSmall-leaved lime treesNative
 Ulmus proceraEnglish elmNative
 Umbellularia californicaCalifornian bay laurelExotic
 Abies proceraNoble firExotic
 Chamaecyparis lawsonianaLawson cypressExotic
 Picea abiesNorway spruceExotic
 Picea sitchensisSitka spruceExotic
 Pinus contortaLodgepole pineExotic
 Pinus nigra var. maritimaCorsican pineExotic
 Pinus sylvestrisScots pineNative
 Pseudotsuga menziesiiDouglas firExotic
 Sequoia sempervirensCoastal redwoodExotic
 Taxus bacataYewNative
 Tsuga heterophyllaWestern hemlockExotic

Inoculation technique

Plant material was divided into two groups, one for the nonwounded treatments and the other for the wounded treatments.

Non-wounded treatments

Care was taken not to expose the cut end of the shoot or petiole to the zoospore suspension. Each shoot or leaf was dipped in the zoospore suspension, apical end first, up to the midway mark, and swirled gently for 1 min. On removal, excess liquid was allowed to drop off.

Wounded treatments

The distal end of each shoot or leaf was dipped after freshly cutting the base and wounding the leaf. The conifers were wounded by trimming approximately 5 mm off the tips of five needles on the distal part of the shoots. Two 5-mm-deep V-shaped incisions were made on the broad leaves, one on either side of the distal part of the leaves.

Incubation and reisolation of the pathogen

Inoculated material was placed on raised sterile metal grids in plastic storage boxes. The base of each box was lined with thoroughly wet sterile paper towels, creating a moist incubation chamber. The sides of the chambers were sprayed with sterile water and the chambers were sealed with a layer of cling film. Plant material was incubated at 20°C with 8 h cool white fluorescent light for 6 days. The moist chambers were opened and the sides were damped down with sterile water daily. At the end of the experiments, four to six pieces of plant tissue per leaf or four to six needles per shoot (and one to four shoot pieces where shoot necrosis was evident) were plated on SMA to confirm that lesions were indeed caused by the pathogen. Isolations were made from the dead–live junction of lesions when present; otherwise, pieces of tissue were selected randomly from the inoculated area.

Disease estimation, data collection and reisolation

Broad-leaved trees and conifers required slightly different assessment methods because of the differences in foliage structure. Three parameters were used to evaluate disease development 6 days after inoculation. For all hosts, the presence or absence of necrosis was recorded for percentage incidence of disease (parameter 1), based on visual inspection. Secondly, on coniferous hosts, the number of necrotic needles and the total number of needles inoculated per shoot were counted and disease severity expressed as a percentage of needles infected (parameter 2). On broad-leaved hosts, digital images of the infected leaves were made and disease severity was expressed as a percentage of necrotic surface area (parameter 2). Infection was confirmed by calculating percentage reisolation (number of positive reisolations out of four to six tissue pieces plated per leaf or shoot) of the pathogen (parameter 3).

Experimental design and statistical analysis

Twelve hosts were tested in any one experiment. The conifer experiments were performed on summer and winter material and experiments were repeated (i.e. replicated) usually 2–3 weeks after commencement of the first experiment. The summer experiments were carried out in July and repeated in August 2003, and the winter experiments carried out in March 2003 and repeated towards the end of March The factorial design of the experiments for conifers was therefore made up of two experiments of 12 hosts (11 conifers plus rhododendron) × five isolates (two × EU plus two × USA plus control) × two treatments (wounded and nonwounded) × two seasons (summer and winter), giving a total of 480 units. An experimental unit comprised one shoot per combination of host, isolate, treatment, season and experiment per season.

In the case of the broad-leaved hosts, only summer material was used, and experiments were repeated (i.e. replicated). Bearing in mind that only 12 hosts could be tested at a time, two series of experiments had to be carried out to test all 23 broad-leaved hosts. Thus the first series of broad-leaf experiments were carried out in June (experiment 1a) and repeated in July (experiment 1b) 2003 (hosts listed in Table 1). The second series of experiments was carried out in July (experiment 2a) and repeated in August (experiment 2b) 2003 (hosts listed in Table 1). The statistical design for each of the broad-leaf series of experiments was therefore made up of two replicates of 12 hosts (rhododendron was used in each series as a positive control) × five isolates × two treatments (wounded/nonwounded), giving a total of 240 units per experiment series. An experimental unit comprised a single leaf per host–isolate–treatment–experiment combination.

Data from the following parameters were measured for the conifers: percentage incidence necrosis (based on visual presence or absence of necrosis) (parameter 1) (an indication of susceptibility based on percentage incidence necrosis was made, such that 0–25% infection was considered to indicate low susceptibility, 26–50% intermediate and < 50% high susceptibility); disease severity was based on the percentage of inoculated needles with necrosis (parameter 2); and infection was based on the presence or absence of reisolation (parameter 3). All the above data were analysed using a binomial generalized linear model with a logit link. The effects of the different factors were tested using an analysis of deviance.

In the broad-leaf species, percentage incidence necrosis data (parameter 1 – presence or absence of necrosis) were analysed using a binomial generalized linear model with a logit link. Disease severity data (parameter 2 – percentage surface area necrosis) were analysed using a log + 1 transformation and a general linear model. The effects of the different factors were tested using an analysis of deviance. Infection data (parameter 3 – percentage reisolation) were grouped into four classes – zero, 1–39% (low), 40–74% (moderate) and 75–100% (high) – and then analysed using an ordinal logistic model (McCullagh & Nelder, 1989).


There were no significant differences among the P. ramorum isolates for any of the parameters measured in either conifers or broad-leaf hosts.

Conifer hosts – disease incidence (parameter 1)

There was a significantly lower incidence of needle necrosis on the controls (2%) than on the inoculated shoots (47%) (P < 0·001) and P. ramorum was not isolated from any of the controls. Inoculated rhododendrons showed 100% incidence of necrosis, confirming the virulence of all four isolates. Controls and rhododendrons were consequently removed from all further analyses so that differences between coniferous hosts could be elucidated. Hosts with low susceptibility included Corsican pine, lodgepole pine and Scots pine (Table 2). On the other hand, coastal redwood, Douglas fir, noble fir, Norway and Sitka spruces and western hemlock had high susceptibility (Table 2). Lawson cypress and yew were intermediate. Further analyses using these clusters, as well as season and wounded or nonwounded treatment data, indicated that there were significant interactions (P < 0·01). Most hosts developed a greater degree of necrosis during the summer inoculations, with the exception of coastal redwood, which had more necrosis on winter shoots and was the main cause of the interaction. Wounding did not significantly increase the incidence of necrosis on plants from the low-susceptibility group, but it did in plants from the other two groups (P < 0·01) (Table 2). There were no significant differences in levels of necrosis caused by the various isolates. Only 39% of all inoculated conifer shoots developed needle necrosis, compared with 82% of the broad-leaved species.

Table 2.  Overall disease incidence and effects of season and wounding on susceptibility of inoculated conifers to Phytophthora ramorum, based on visual inspection only
HostOverall disease incidence (% necrosisa,b) and susceptibility groupcDisease incidence (% necrosisa)Disease incidence (% necrosisa)
  • a

    Percentage calculated as the number of shoots bearing necrotic needles divided by the total number of shoots inoculated (eight shoots inoculated per host per wounded or nonwounded treatment, 16 shoots in total per experiment) × 100.

  • b

    Overall mean percentage of shoots bearing necrotic needles irrespective of season and wound effects.

  • c

    Susceptibility group based on visible response of foliage.

Coastal redwood56 (high)50633875
Corsican pine25 (low)50 01931
Douglas fir59 (high)75443881
Lawson cypress31 (intermediate)3825 063
Lodgepole pine13 (low)19061313
Noble fir63 (high)81443194
Norway spruce59 (high)75444475
Scots pine13 (low)25 0 619
Sitka spruce53 (high)75313175
Western hemlock53 (high)75312581
Yew41 (intermediate)50311369

Conifer hosts – disease severity (parameter 2)

Few pine and Lawson cypress needles had necrosis so they were excluded from the analysis. The analysis of the remaining data showed significant host × wound-treatment interaction (P < 0·001) attributable to the Douglas fir and noble fir data. In the absence of wounding they were the most susceptible species (34 and 26%, respectively) of needles showing necrosis (Table 3), but these levels dropped in the wounding treatment. However, with the other conifers (coastal redwood, Norway spruce, Sitka spruce, western hemlock and yew), wounding increased susceptibility. Most hosts had a similar degree of disease severity in the wounded treatment; the exception was yew, which had more necrotic needles than the other species (Table 3).

Table 3.  Effects of wound treatment on disease severity and infection potential of needles, and effect of season on infection potential of conifers inculated with Phytophthora ramorum
HostDisease severity (% necrotic needles)aInfection potential (reisolation %)b
  • a

    Calculated as the percentage of the total number of needles inoculated per shoot which were necrotic.

  • b

    Calculated as the percentage of needles that gave positive reisolation of P. ramorum.

  • c

    Not analysed.

Coastal redwood 519 6815038
Corsican pinec 0503813
Douglas fir3417253156 0
Lawson cypress 02519 6
Lodgepole pine131925 6
Noble fir2616193856 0
Norway spruce 213 02519 6
Scots pine 0635013
Sitka spruce 515 63138 0
Western hemlock 316 0 6 6 0
Yew 32813635619

Conifer hosts – infection potential (parameter 3)

Overall, only 23% of inoculated conifer needles yielded the pathogen on reisolation, compared with 91% of broad-leaved species. There was a significant effect of season (P < 0·001) with higher levels of reisolation from summer material than from winter material (Table 3). However, there was also a significant host × wound-treatment interaction (P < 0·001). Wounding the leaves increased the level of reisolation in all cases (Table 3), but the extent of the increase differed between hosts. For example, only a small increase was evident in lodgepole pine, Douglas fir and western hemlock, whereas a greater increase was evident in coastal redwood, Corsican pine, Scots pine and yew.

Broad-leaf hosts – disease incidence (parameter 1)

In all experiments, a significantly lower percentage of control leaves than inoculated leaves showed signs of necrosis on most hosts (82% of all inoculated leaves developed some degree of necrosis). Additionally, the reisolations from all control leaves were negative and so the controls were excluded from further analyses. There were no significant treatment interactions with any of the factors tested. By dropping the control and the nonsignificant experiment and isolate effects, on reanalysis it was evident that wounding significantly (P < 0·001) increased the percentage of inoculated leaves of alder, beech, birch, cider gum, common oak, cork oak, holly, red oak and wild cherry that developed necrosis (Table 4).

Table 4.  Disease incidence (mean percentage of nectrotic leaves) on wounded and nonwounded leaves of broad leaf species 6 days after inoculation with Phytophthora ramorum based on visual inspection only
Experiments 1a and 1bExperiments 2a and 2b
HostNon-wounded (%)aWounded (%)HostNon-wounded (%)Wounded (%)
  • a

    Percentages based on the number of necrotic leaves out of 16 leaves per treatment.

Beech 25·0 62·5Cork oak  0·0 37·5
Birch 37·5 75·0Holly 12·5 75·0
Hornbeam 75·0 75·0Hazel 50·0 37·5
Red oak 75·0 87·5Alder 50·0 87·5
Common oak 75·0100·0Small-leaved lime 75·0 62·5
Ash100·0100·0Cider gum 87·5100·0
Aspen100·0100·0Wild cherry 87·5100·0
Holm oak100·0100·0Californian bay laurel100·0100·0
Sweet chestnut100·0100·0Horse chestnut100·0100·0
Turkey oak100·0100·0Sessile oak100·0100·0

Broad-leaf hosts – disease severity (parameter 2)

There were significant interactions in both sets of experiments. In experiments 1a and 1b, the host–experiment interaction was significant (P < 0·001). Data from each experiment were analysed separately. Consistently (across both experiments 1a and 1b) high levels of necrosis were produced on ash and rhododendron. However, disease was more severe in experiment 1a than in experiment 1b for birch, holm oak and turkey oak (Table 5). In experiments 2a and 2b there was significant (P < 0·001) three-way interaction between experiment, host and treatment, making interpretation complex. Investigation of the interaction using mean values on a loge scale revealed that consistently low levels of necrosis were obtained in cork oak and hazel, but high levels were found in elm, horse chestnut, rhododendron and sessile oak (Table 5). Wounding increased necrosis significantly on certain hosts, viz. Californian bay laurel, cider gum, holly and wild cherry in experiment 2a.

Table 5.  Disease severity on inoculated leaves of broad-leaf species expressed as loge scale mean percentage necrotic leaf area 6 days after inoculation with Phytophthora ramorum
HostExperiment 1aExperiment 1bHostExperiment 2aExperiment 2b
  • a

    Loge value of the means. Means based on eight leaves per experiment. The lack of host × treatment interaction in experiments 1a and 1b enabled treatment data to be pooled. In experiments 2a and 2b, the three-way interaction prevented pooling the data.

  • SE for experiments 1a and 1b = 0·37; LSD = 0·96.

  • SE for experiments 2a and 2b = 0·41; LSD = 1·08.

Aspen2·271·78Californian bay laurel1·482·772·713·41
Beech0·190·43Cider gum0·922·793·793·64
Birch2·670·35Cork oak0·000·690·060·17
Common oak0·651·03Elm2·872·652·612·44
Holm oak4·44a2·65Hazel0·100·351·310·40
Red oak0·740·75Horse chestnut2·552·543·412·84
Sweet chestnut2·081·53Sessile oak1·912·252·242·97
Sycamore1·170·89Small-leaved lime2·613·190·200·17
Turkey oak2·811·24Wild cherry0·381·472·032·03

Broad-leaf hosts – infection potential (parameter 3)

Phytophthora ramorum was never reisolated from the controls, but it was isolated from 91·4% of inoculated leaves, indicating a very high infection level. Analyses were therefore conducted on inoculated material only. A significant host–treatment interaction was evident for experiments 1a and 1b, but upon further investigation of the category of high levels of reisolation only, it was clear that the pathogen was reisolated consistently from ash, holm oak, hornbeam, rhododendron and sweet chestnut (Table 6). Wounding increased levels of reisolation in some hosts (e.g. birch and common oak). In experiments 2a and 2b there was host–experiment interaction (P < 0·001), which made interpretation of results difficult, but in general P. ramorum was infrequently reisolated from alder and wild cherry, while being obtained moderately frequently from holly. The other hosts yielded high levels of reisolation (Table 6).

Table 6.  Infection potential of Phytophthora ramorum based on levels of reisolation from inoculated leaves of broad-leaf species
HostExperiment 1a and 1ba (proportion of leaves with > 75% reisolation)cHostExperiment 2a and 2bb (number of leaves in each reisolation category)c
  • a

    Host × treatment interaction in experiments 1a and 1b.

  • b

    Overall host effect overlooking possible replicate effects in experiments 2a and 2b.

  • c

    Zero, pathogen not reisolated; low, 1–39% of reisolations yielded the pathogen; moderate, 40–74% of reisolations yielded the pathogen; high, P. ramorum obtained in > 75% of reisolations. Percentages based on the number of positive reisolations out of four to six pieces of tissue plated per leaf.

Ash1·0001·000Alder630 7
Aspen0·1250·125Californian bay laurel00016
Beech0·2500·750Cider gum10213
Birch0·0000·875Cork oak00511
Common oak0·1250·625Elm00016
Holm oak0·8751·000Hazel21211
Hornbeam0·8750·875Holly242 8
Red oak0·2500·500Horse chestnut00016
Sweet chestnut0·7501·000Sessile oak00214
Sycamore0·7500·750Small-leaved lime00313
Turkey oak0·8750·875Wild cherry256 3


The main aims of this study were to test the ability of P. ramorum to infect foliage on a range of tree hosts found in the UK, and to compare the relative amounts of damage caused. The reisolation data were indicative of infection ability, while necrosis was a measure of damage. Indications of host susceptibility cannot be based on the degree of necrosis alone; reisolation must confirm that necrosis is caused by P. ramorum and not by another agent. It is therefore essential to consider the necrosis and reisolation data together. The information obtained from parameter 1 (disease incidence) was useful in providing an idea of general trends, but data from parameters 2 (disease severity) and 3 (infection potential) were essential for comparing host susceptibilities.

In general, foliage of conifers was much less susceptible to infection than broad-leaf species. This was evident in both lower levels of necrosis and less frequent reisolation of the pathogen. Pinus and Lawson cypress needles were generally unaffected, and P. ramorum could not be reisolated, except in a few instances from lodgepole pine or from wounded tissue, suggesting that they are probably nonhost species. Douglas fir and noble fir were highly susceptible hosts, and the decrease in needle necrosis after wounding was difficult to explain, especially when all the other conifers showed an increase in disease after wounding. Although the visual assessment of necrosis on spruces was high, the levels of reisolation were fairly low. Thus foliage of these species is probably not at risk of being infected.

In California, infection of Douglas fir occurs in forests, and needle necrosis, small branch cankers (0·5–1 cm diameter), wilted new shoots and defoliation and dieback of branches are indicative of P. ramorum infection (Davidson et al., 2002b; Garbelotto et al., 2003). In the present study, needle necrosis on Douglas fir could be attributed to shoot infections rather than needle infections. Apparently, the pathogen gained access to shoot tissue and progressed along the shoots causing the needles to die, rather than the needles themselves being infected. Hansen et al. (2005) found that shoot-tip dieback resulted only from infections occurring near buds or at the bases of cones, confirming that needles are not the main point of entry for P. ramorum in Douglas fir. Young shoot tissue that was present during the summer inoculations in the present study was far more susceptible than older tissue found on winter shoots, and Hansen et al. (2005) give a detailed account of the susceptibility of Douglas fir at different phenological stages, which corroborate the present findings.

At a nursery in the UK, infected foliage was found on potted yew trees (Lane et al., 2004), but no further reports of P. ramorum on this host have been made. In the present study, yew was moderately susceptible, although wounding increased infection and necrosis dramatically. Native yew is therefore unlikely to be an important host for P. ramorum.

Infected foliage on basal sprouts and lower branches of young coastal redwoods has been reported in California (Maloney et al., 2002) and tests carried out by Hansen et al. (2005) show that infections are limited to these plant parts of coastal redwoods. In the present study, this host was also identified as a susceptible species but is not highly significant as a host because of its limited distribution in the UK.

In the broad-leaf tree species, distinct differences in susceptibility were evident. It became apparent in these experiments that the physiological condition of host tissue associated with leaf age and/or position in the canopy, amongst other things, affects disease expression. Hansen et al. (2005) also point out that leaf age affects susceptibility when using the detached leaf assay. In most hosts, leaf necrosis was more severe in the young tissue used in the first experiments of the present study and decreased when slightly older leaves were used in the repetition experiments. The significantly increased necrotic area on Californian bay laurel, cider gum and sessile oak leaves in the repeat experiments is difficult to explain, but may be attributable to leaves being sampled from the lower inner canopy, or from epicormic shoots arising from the trunk or tree base. Leaves growing in this protected environment would have thinner cuticles and softer tissues, and disease development would be more severe. Since nursery Californian bay laurels were used, a different, more susceptible genotype may have been inadvertently selected for the replicates. Clearly, the effects of leaf age and position in the canopy, as well as the effects of different host genotypes, merit further investigation.

In this study, the most consistently susceptible species were ash, holm oak and rhododendron. Although there was a marked decrease in necrosis on holm oak in the repeat trial, leaves were still highly susceptible. Levels of reisolation from these hosts were always very high. Levels of leaf necrosis > ∼10% in both replicates and high levels of reisolation were considered to indicate high susceptibility.

Although birch, holly, small-leaved lime and alder had moderately high susceptibility in the first experiment, almost no necrosis developed in the replicate experiment and reisolation levels were very low for these hosts in all experiments. This suggests that some of the necrosis could have been caused by other factors and these hosts are therefore not considered highly susceptible to the pathogen.

Sweet chestnut was an interesting host. A moderate amount of necrosis developed on the leaves in both experiments, but very high levels of reisolation were recorded. While these results confirm the susceptibility of the host, leaf age or physiological condition can restrict disease development, with leaves with soft thin cuticles being more susceptible than hardened-off leaves. This phenomenon has also been observed in other tests carried out on attached foliage (SD, unpublished data). Another interesting host was aspen because it also had a moderate level of necrosis, but the pathogen could not be reisolated at all. Given that there was no blemishing on the controls, the results might indicate a hypersensitive response. Finally, a few hosts only developed a small degree of necrosis, but had a high level of reisolation (hornbeam, red oak and sycamore), with P. ramorum being isolated from symptomless leaves in some cases. The significance of this is uncertain. Species that consistently developed small areas of necrosis (< ∼8%) and which also had low levels of reisolation were considered less susceptible.

Natural infections of foliage of four broad-leaved hosts identified as susceptible in this study have now been found in woodland situations in the UK: ash, holm oak, rhododendron and sweet chestnuts (SD, unpublished data; Lane et al., 2003; Defra, 2004; Denman et al., 2005). Infection of foliage of a single ash tree was found very recently and further attention must be given to this species as a possible source of infection and inoculum production. The incidence of infected holm oaks in Cornwall is relatively high; so far 50% of samples tested have been positive (SD, unpublished data). As holm oak is evergreen, it could potentially be an important host if it generates inoculum. Furthermore, holm oak is a widespread species in Mediterranean areas, and if disease were to establish in these areas, there could be serious ecological consequences, as well as economic, social and environmental impacts. It is therefore necessary to monitor the situation in the field carefully.

Sweet chestnut foliage, on the other hand, will probably be less important as a host and source of inoculum because only a few naturally infected trees have been found in the UK so far, and infections were restricted to soft, thin-cuticled leaves on epicormic shoots beneath the canopy. As sweet chestnut is a deciduous species, infection opportunities will also be restricted.

Apart from nursery plants, naturally infected rhododendrons have been discovered at a number of sites, particularly in southwest England and Wales, but also in southeast England. Most of these infections are on naturalized Rhododendron ponticum. In the present study the ornamental R. catawbiense cv. Cunningham's White was used because it is highly susceptible to P. ramorum and genetically uniform, unlike R. ponitcum, which has a high level of genotypic variability. Plant health legislation requires infected rhododendrons to be removed and burnt since they are a major source of inoculum. At this stage, management recommendations on tree foliage infections are to prune out infected branches and monitor the trees.

The fact that there were no statistical differences in aggressiveness among the isolates used in the present study merely indicates that the two subpopulations of P. ramorum did not behave differently on the various hosts. No conclusions can be drawn about the relative aggressiveness of subpopulations from the data presented here, although other tests on logs have clearly demonstrated differences in aggressiveness of subpopulations (Brasier, 2003).

Although only a single host genotype was tested in most cases in the present experiments, further work carried out in planta has confirmed the initial results (S. D. unpublished). Furthermore, there was no dilution effect on inoculum density with successive leaf dipping (unpublished).

The results of this study (and others, see Hansen et al., 2005) demonstrate that detached leaf assays are useful as an indicator of species susceptibility, although environmental factors will affect establishment of disease in the field. Phytophthora ramorum is a ‘quarantine-listed pest’ in the UK (Defra, 2004) and the findings of this work contribute to a PRA (pest risk assessment) of the potential damage that P. ramorum could cause to British woodlands. The results will also be useful to those monitoring the spread of this pathogen.


We thank Dr Geoff Morgan for carrying out the statistical analyses. The work carried out in this study was funded by the Forestry Commission.