Analysis of the natural infection of European horse chestnut (Aesculus hippocastanum) by Pseudomonas syringae pv. aesculi

Authors


E-mail: sarah.green@forestry.gsi.gov.uk

Abstract

Pseudomonas syringae pv. aesculi (Psa) is an emerging bacterial pathogen responsible for a recent epidemic of bleeding canker of European horse chestnut (Aesculus hippocastanum) in northwest Europe. Very little is known about the infection biology of this pathogen, which can cause lethal cankers in the branches and stem of its host. In this study, branches and whole trees of European horse chestnut naturally infected with Psa were subjected to detailed morphological and histological examination to identify the primary infection sites, the time of infection, and the patterns of subsequent lesion expansion within the host. Lesions developed during the host dormant season on the 2003–2009 extension growth increments and were centred mainly on lenticels, leaf scars and nodes. The oldest lesion developed in the 2004/2005 dormant season and the number of new lesions increased in each subsequent year. The lesions developed in the cortex and phloem and extended into the cambium to cause cankers, but there was no evidence of necrosis in the xylem. All lesions on the branches were discrete and apparently contained by a necrophylactic periderm, although there was evidence that Psa could survive within such periderms and subsequently breach them. Examination of two whole 30-year-old trees revealed extensive, continuous cankers in the phloem and cambium which had formed within a single growing season. Thus, the success of Psa as a tree pathogen and the causal agent of a large-scale epidemic may in part reflect an ability to infect the aerial woody parts of its host directly.

Introduction

Bleeding canker of European horse chestnut (Aesculus hippocastanum) is a new disease which has established rapidly and become widespread since 2002/2003 in several countries of northwest Europe, including Britain, the Netherlands and Belgium (Webber et al., 2008; Green et al., 2009). Thousands of horse chestnut trees across Britain now exhibit symptoms associated with this disease (Forestry Commission, 2008), which include rust-coloured liquid oozing from cracks in the bark located on the main stem and branches, necrotic phloem underlying the outer bark, and dieback often leading to tree death (Green et al., 2009). The causal agent of this disease has been identified as the Gram-negative bacterium, Pseudomonas syringae pv. aesculi (Psa). This identification was based on a partial sequence for its gyrase B gene, which was identical to that of the Psa type strain isolated in 1969 from leaf lesions on Indian horse chestnut (Aesculus indica) in India (Durgapal, 1971; Durgapal & Singh, 1980).

Pseudomonas syringae pv. aesculi is highly aggressive on European horse chestnut and apparently very mobile, since the pathogen can spread rapidly between and within infected trees, causing dieback and mortality when it kills a large proportion of the phloem in the branches and stem. Psa can be isolated from lesions in the phloem within bleeding cankers both on the main stems of trees and on branches of various ages (Green et al., 2009), but the initial stages of infection by this pathogen have not been elucidated.

Other closely-related pathogens of woody hosts include P. syringae pv. syringae and P. syringae pv. morsprunorum, which cause bacterial canker of stone fruit trees (Prunus spp.), and P. avellanae, the causal agent of bacterial canker of European hazelnut (Corylus avellana). These pathogens typically initiate new infections in the spring and summer on the soft tissues of their hosts such as leaves, young shoots and fruits, and do not invade the perennial organs of the tree until the autumn or winter, with canker extension typically occurring in the early spring (Crosse, 1966; Scortichini, 2002; Kennelly et al., 2007).

In contrast, there are no reports in the literature of Psa killing the tissues of the leaves or flowers of European horse chestnut, nor of its epiphytic survival on these organs. Observations of a number of diseased trees in Scotland and southwest England showed that lesions and cankers occurred on the stem, branches and young, extending shoots (Green et al., 2009), suggesting that the sites at which initial infection occurs may be located on the woody parts of the tree. This interpretation is supported by experiments which indicated that Psa was able to infect host tissues directly when inoculated onto young, unwounded branches of European horse chestnut (Green et al., 2009). It is possible that Psa gains access to the tissues within stems and branches via discontinuities in the outer bark such as leaf scars, growth cracks at branch junctions, dormant buds or wounds since these are entry points for other bacterial tree pathogens; all of which require natural openings or unprotected surfaces for infection (Crosse, 1955; Scortichini, 2002; Kennelly et al., 2007).

The horse chestnut bleeding canker epidemic is having a major impact on urban and rural landscapes throughout affected countries. It is important to elucidate the processes of infection by Psa in order to understand how the bacterium can colonize its host so rapidly and cause such severe disease symptoms. This information may also be of relevance in relation to other bacterial diseases of trees, for which the infection processes are generally poorly understood (Kennelly et al., 2007). Studies of host infection by other bacterial canker pathogens have largely been conducted following artificial inoculation of host trees (e.g. Shanmuganathan, 1962; Scortichini & Lazzari, 1996; Temsah et al., 2007, 2008). However, these studies may not be representative of natural infection, since the artificially inoculated trees are either wounded or exposed to very high numbers of bacterial cells.

In this study, detailed anatomical and histological examinations were carried out on a number of European horse chestnut trees naturally infected with Psa to determine how infections on the woody parts of the trees developed. The main objectives of the study were to locate the primary infection sites, determine when lesions developed, and investigate the subsequent patterns of spread of Psa within the tree. The results are discussed in terms of their implications for understanding the Psa epidemic on European horse chestnut.

Materials and methods

Branch assessments

To locate the primary infection sites of Psa, assessments were carried out on branches from naturally infected European horse chestnut trees showing typical bleeding canker symptoms of defoliation, dieback, bark cracking and stem bleeding. Based on the scars left by each year’s terminal bud, the extension growth increment for each year could be identified on young branches. Branch lengths containing the extension growth increments formed in 2003–2009 were sampled from trees at two sites near Edinburgh, Scotland.

At site one, located near Penicuik, Midlothian, 13 trees showing established symptoms of infection by Psa were selected for future observation and analysis in March 2009. These were grafts originating from material collected from Castle Howard, Yorkshire; Kempton Park, Hereford; Thetford, Norfolk; and Florence, Italy; and had been planted in 1996. The trees were examined for lesions on leaves, buds, flowers and woody branches at monthly intervals from June to October 2009, when lesions on the 2009 extension growth increment were first observed. In October 2009, two branches were selected per tree on the basis of the presence of numerous, sunken, blackened and/or cracked bark lesions typical of infection by Psa (Fig. 1). Both branches were labelled and one was left on the tree, while the second branch was excised close to its junction with the main stem and brought to the laboratory for assessment. For one of the 13 trees at this site, two branches containing lesions were excised to provide additional material for confirmation of the presence of Psa.

Figure 1.

 Cracked, sunken lesion typical of Pseudomonas syringae pv. aesculi located on an internode on a young branch of European horse chestnut (Aesculus hippocastanum).

At site two, located near Peebles, Scottish Borders, infected branches with visible bark lesions were excised in October 2009 from two mature, large-diameter European horse chestnut trees which had been deteriorating in condition for several years. Three branches were collected from one tree and two from the second tree.

The total number of discrete lesions on each extension growth increment, the apparent point of origin of each lesion with respect to specific anatomical features of the host (lenticels, nodes, leaf scars, buds, etc.), and lesion length (mm) and tangential spread as a percentage of the branch circumference were recorded for all selected branches, i.e. both those that were excised and those that were left in situ. The time of lesion development was also recorded for all lesions on the 2009 extension growth increments and for lesions which had killed the cambium on extension growth increments dating from 2003–2008. This was done by dissecting all lesions on branches excised from trees at site one and eight lesions on branches excised from trees at site two. The time of cambial killing was determined by cutting through the branch at the centre of each lesion to expose a transverse view of the lesion. The freshly cut branch segments (approximately 10–15 mm thick) were then immediately examined using a stereo microscope. The year in which the cambium had been killed was determined by calculating the age of the radial growth increment in the xylem immediately adjacent to the killed portion of the cambium. Lesion development was also examined in these branch segments to determine which host tissues were killed during the progression of the disease. Several branch segments containing lesions were fixed in a standard solution of formalin-acetic acid-alcohol (FAA) and processed in a ThermoFisher Pathcentre (Thermo Fisher) in which they were dehydrated in an industrial methylated spirit and ethanol series, and embedded in Paraplast. Sections (4 μm thick) were cut through the embedded branch segments using a Finesse ME microtome (Thermo Fisher) and stained using either thionin and orange G (Stoughton, 1930) or Gram stain (Gurr, 1965). The stained sections were examined using a compound microscope at ×100 and ×400 magnification and photographs taken of representative lesions using a microscope-mounted camera (Olympus E-300).

To confirm the presence of Psa in lesions, samples consisting of a combination of cortex and phloem were taken from 58 lesions centred on different branch features (lenticels, leaf scars, nodes, buds, etc.) on 18 branches (13 from site one and five from site two). To test whether Psa infected a range of tissues, samples of cortex/phloem, cambium, xylem, pith and petiole tissues were taken from the margins of 30 lesions located on an additional branch collected from site one and from nine lesions located on a branch collected from site two. For all samples, tissue pieces approximately 5 mm2 were cut from the lesion margins, placed on Nutrient agar amended with 5% w/v sucrose, crystal violet (2 mg L−1) and actidione (cycloheximide) (50 μg mL−1), to inhibit fungal growth, and incubated at room temperature overnight. Any bacterial growth which developed was then tested for fluorescence under UV light, and cells were collected from fluorescent growth using a sterile loop. The bacterial cells were placed into 100 μL of sterile distilled water, and 2 μL of this suspension were used in a real-time PCR assay which amplified DNA in whole cells of Psa as described by Green et al. (2009).

Whole tree assessment

The distribution of lesions throughout the branching structure of entire horse chestnut trees with typical bleeding canker symptoms was determined. Two diseased trees, which were approximately 30 years old, were felled at a single location in Glasgow, Scotland. One tree was felled in November 2005 and the other in May 2006. The aerial part of each whole tree was sawn into five segments, consisting of the upper crown, mid-crown, upper stem, mid stem and lower stem. Each segment was completely stripped of outer bark to reveal continuous cankers (i.e. lesions which had killed the cambium) which were present throughout the stem and branches. To determine the time of canker formation in each tree segment, transverse sections (35 μm thick) were cut using a microtome through samples of wood and bark excised from each canker, stained with safranin and examined using a compound microscope. The time at which the cankers had formed was established by examination of the number of radial growth increments in the xylem of the calluses formed at their margins. Sections through a total of 18 cankers were examined in this way, 13 from one tree and five from the other. Necrotic pieces of phloem were taken from the margins of tangentially spreading lesions, placed onto 2% malt agar or Nutrient agar, and incubated at room temperature in the dark in an attempt to isolate the causal pathogen.

Results

Branch assessments

No lesions were observed on leaves, buds or flowers during the 2009 growing season. All lesions occurred on the bark of the woody branches and occasionally on leaf petioles. Data were pooled for all branches since the patterns of infection were similar across trees at both sites. Observations were made on a total of 186 lesions typical of Psa, of which 130 were measured. Lesions were found in branch extension growth increments dating from 2003 to 2009 and there was a steady increase in the number of lesions on extension growth increments from 2003 to 2006 (Fig. 2). The lesions ranged in length from 2–128 mm and occupied from 5 to 100% of the branch circumference. The majority of these lesions (122) were less than 51 mm long and 117 had extended tangentially to occupy <30% of the branch circumference. Generally, the largest lesions were found on the 2006, 2007 and 2008 extension growth increments, although there was considerable variation in lesion size on all ages of material (Table 1a).

Figure 2.

 Total number of lesions on different extension growth increments on 32 European horse chestnut (Aesculus hippocastanum) branches naturally infected with Pseudomonas syringae pv. aesculi. Data are pooled for all branches that were excised and those left in situ. Note that, with the exception of the 2009 extension growth increment, lesions may have been initiated in later years than the year of extension growth.

Table 1.   Number of lesions measured and their mean length (mm) and tangential extension (as a percentage of branch circumference) in relation to: (a) branch extension growth increment and (b) specific anatomical features of European horse chestnut (Aesculus hippocastanum) branches naturally infected with Pseudomonas syringae pv. aesculi that are closest to the centre of each lesion. Data in parentheses are the standard deviation of the individual lesions
Lesion positionNumber of lesions measuredMean lesion length in mmMean tangential lesion extension as a % of branch circumference
(a) Branch extension growth increment
 2003411 (3·8)11 (2·5)
 20041015 (8·5)14 (10·2)
 20051814 (10·7)21 (22·7)
 20062225 (26·1)25 (19·1)
 20072027 (30·2)26 (19·3)
 20083023 (17·9)29 (21·4)
 2009268 (11·9)10 (9·2)
(b) Specific anatomical feature closest to the centre of each lesion
 Internode5924 (25·5)22 (17·1)
 Lenticel274 (1·9)6 (2·2)
 Leaf scar2219 (10·9)31 (23·1)
 Node820 (13·4)26 (19·8)
 Side shoot633 (22·1)29 (16·9)
 Branch axil516 (9·5)36 (36·6)
 Axillary bud323 (12·5)25 (8·7)

The position of the centre of each lesion in relation to specific anatomical features of the host was recorded for 154 lesions on 31 branches from trees at sites one and two, including branches that were excised and those left in situ (Table 2). Lesions occurred at a range of identifiable positions on the branches where there was some discontinuity or irregularity in the bark including, most frequently, lenticels (Fig. 3a,b), leaf scars (Fig. 3c) and nodes (Fig. 3d) (Table 2). Lesions that were recognisably centred on lenticels were the smallest (Table 1b) but this may be due to their recent origin, as all of those which clearly met this description were initiated in 2009. Otherwise, the size of lesions was not affected by their specific location in relation to branch features (Table 1b). Approximately half of all lesions were located on branch internodes and were not associated with leaf scars, nodes or any other obvious anatomical feature (Table 2); (Fig. 1). It is possible that a number of these lesions were originally centred on lenticels but a point of origin could not be discerned due to subsequent lesion extension and degradation of tissues.

Table 2.   Total number of lesions, number of lesions sampled for positive identification of Pseudomonas syringae pv. aesculi (Psa) and number of lesions positive for Psa on specific anatomical features of European horse chestnut (Aesculus hippocastanum) branches naturally infected with Psa
Location of lesionTotal number of lesionsNumber of lesions sampled for confirmation of PsaNumber of lesions positive for Psa
Lenticels291511
Leaf scar26135
Node1141
Side shoot630
Branch axil421
Petiole222
Axillary bud300
Internode: no discernable point of entry731910
Figure 3.

 Young branches of European horse chestnut (Aesculus hippocastanum) with lesions caused by Pseudomonas syringae pv. aesculi. (a) Four lesions centred on lenticels. (b) Close up view of one of four lesions centred on a lenticel. (c) Lesion centred on a leaf scar. (d) Lesion centred on a node.

One lesion which was centred on a lenticel, and which had been initiated on a 2009 extension growth increment, was examined in more detail to confirm the presence of Psa. The lesion, which was approximately 4 mm in length, included necrotic cortex which was surrounded by a necrophylactic periderm (Fig. 4a). Bacterial cells were observed among the disrupted cortical tissues within the lesion (Fig. 4b) and Psa was positively identified from this lesion using the real-time PCR assay. Other lenticel-centred lesions which were examined, and which were also located on the 2009 extension growth increments, included necrotic phloem (e.g. Fig. 4c), and one had reached the cambium. Several healthy lenticels were examined on a 2009 extension growth increment collected in October. Their structure was found to consist of a raised surface with a central opening located above a cluster of loosely packed cells underlain by a layer of suberized or lignified cells (Fig. 4d). Of the 27 lesions discernibly centred on lenticels, 22 were located on the 2009 growth increment and the rest, which were probably also initiated in 2009 based on the condition of the necrotic tissues, were located on the 2004, 2005, 2006 and 2008 extension growth increments.

Figure 4.

 Photomicrographs of transverse sections taken through lenticel regions on young branches of European horse chestnut (Aesculus hippocastanum). (a) Cellular degradation within the cortex of a lesion centred on a lenticel. Arrows indicate two layers of necrophylactic periderm. Bar = 200 μm. (b) Bacterial cells in the cortex of the same lesion shown in (a). Bar = 10 μm. (c) Lesion centred on a lenticel (arrow) which has reached the phloem (ph). Bar = 200 μm. (d) Healthy lenticel (arrow) in the same branch as (c). Bar = 200 μm.

A total of 115 lesions on branches excised from trees at sites one and two were dissected to determine when they had developed. In the majority of lesions there was disruption and apparent degradation of the cortex and phloem and the lesions were at least partially enclosed by one or more layers of necrophylactic periderm (e.g. Fig. 4a). All lesions appeared to have extended into the cambium during the host dormant season, i.e., during a period between the end of one growing season and the start of the next. This conclusion stems from a consistent observation that the vascular cambium had been killed after the cessation of cell division. In each case, the last annual xylem ring underlying the dead cambial layer was complete and undegraded (Fig. 5), although the intact xylem often exhibited an area of dark staining adjacent to the region of dead cambium (Fig. 5). Psa had evidently remained active in many lesions despite being encircled by necrophylactic periderm. This was indicated by cases in which tangential spread into regions of the phloem and cortex beyond the necrophylactic periderm associated with the lesion’s original boundaries had occurred. In the case of the 2008 initiated lesion shown in (Fig. 5), this enlargement of the lesion apparently occurred following the cessation of secondary growth in 2009 (Fig. 5). This pattern of perennation was observed in the majority of older, discrete lesions which were examined on the branches and suggests that Psa frequently survives within the tissues of established lesions.

Figure 5.

 Transverse section through a young branch of European horse chestnut (Aesculus hippocastanum) with a discrete lesion caused by Pseudomonas syringae pv. aesculi. The cambium was killed after the 2008 growing season (indicated by the upper, continuous white arrow) with staining (s) in the xylem under the original site of the killed cambium. The lesion has extended tangentially (as indicated by the lower, dashed white arrow) into the phloem (ph) and cortex (co) following secondary growth of the xylem (x) and phloem during the 2009 growing season.

The earliest instance of cambial killing occurred in the 2004/05 dormant season in a single lesion located on a 2004 extension growth increment on a branch from site two. At site one, the earliest instance of cambial killing occurred in the 2005/06 dormant season in a single lesion located on a 2003 extension growth increment. Over both sites, the number of lesions which had killed the cambium in subsequent years were five in 2006/07, 26 in 2007/08 and 27 in 2008/09. In 2009/10, 27 lesions developed at the end of the 2009 growing season on the 2009 extension growth increments. Seven of these lesions had already killed the cambium at the time of assessment but in each case the 2009 xylem ring was undisrupted. Several of the remaining lesions which had developed in 2009 and had not penetrated as far as the cambium were apparently confined to the phloem by a wound periderm. A total of 28 superficial lesions were also recorded on extension growth increments from 2003 to 2008. However, the year of initiation of these lesions could not be determined with any accuracy because they had not reached the cambium.

Psa was positively identified from 30 of the 58 lesions sampled from the different branch features, and from 17 of the 18 sampled branches. Psa was identified from lesions centred on lenticels and internodes, as well as leaf scars, petioles, a node and a branch axil (Table 2). Of the 30 lesions positive for Psa, 17 developed at the end of the 2009 growing season, nine had killed the cambium in the 2008/09 dormant period, one in the 2007/08 dormant period, one in the 2006/07 dormant period and two in the 2004/05 dormant period. The older lesions (pre-2009 growing season) from which Psa was identified also showed evidence of perennation as illustrated in (Fig. 5).

A total of 89 samples consisting of cortex/phloem, cambium, xylem, pith and petiole tissues located at the margins of 39 lesions were tested for Psa. The bacterium was confirmed to be present in 23 of the 39 lesions, including 44% of cortex/phloem samples (17/39), 57% of cambium samples (8/14), 14% of xylem samples (3/21), 8% of pith samples (1/12) and 100% of petiole samples (3/3). The 23 lesions in which Psa was confirmed had been initiated in all years from 2004/05 to 2009, with the single exception of 2005/06.

Whole tree assessment

Examination of the two whole trees in 2006 revealed extensive lesions in the phloem which had spread tangentially and were connected longitudinally by continuous, strip-like cankers in the phloem and cambium (Fig. 6a). These strip cankers extended from the young branches in the upper crown to the base of the main stem (Fig. 6b). On the main stem, phloem lesions which had spread tangentially were frequently centred on branch insertion points, where the bark was heavily ridged. The time at which the lesions had caused death of the cambium was established at various points along their lengths on both trees. In all cases, cambial death had occurred during the course of the 2004 growing season. This was indicated both by continued radial growth of the xylem beyond the margins of the lesions where living cambium was still present and by the formation of a traumatic ring which had formed more widely within the xylem of the stems in mid- to late- 2004. The 2003 xylem ring was consistently complete and undamaged, indicating that the cankers had extended during a single growing season. No necrosis of parenchyma within the xylem underlying lesions was noted and xylem growth was only disrupted adjacent to regions of dead cambium. Psa was isolated from only one of the trees. However, the lack of success in isolation of Psa from the second tree was almost certainly due to the use of a sub-optimal method for isolation of the pathogen from both trees at a time when a PCR-based assay for the pathogen had not yet been developed.

Figure 6.

 Diseased European horse chestnut (Aesculus hippocastanum) trees which have been stripped of outer bark. (a) Segment of the main stem with tangentially spreading lesions (l) in the phloem and strip cankers (ca) in the phloem and cambium. (b) Whole tree with continuous strip cankers in the phloem and cambium which developed during 2004.

Discussion

Pseudomonas syringae pv. aesculi is an aggressive new pathogen which has spread rapidly through populations of European horse chestnut in northwest Europe. Large, mature trees can be killed by the disease due to the development of extensive cankers in stems and branches. In this study, observations of naturally infected trees have provided strong evidence that Psa initiates infection of European horse chestnut via lenticels, leaf scars and other natural openings in branches of various ages. The bacterium then colonizes the cortex, phloem and cambium and has the potential to form extensive, continuous cankers within a single growing season. Thus, part of the success of Psa as a tree pathogen and the causal agent of a large-scale epidemic is due to an apparently highly effective capacity for direct aerial infection and colonization of the woody parts of its host.

Here, it is shown that Psa can infect woody branches of varying ages via lenticels, leaf scars, nodes and branch axils. Lenticels are frequently occurring, natural openings within the periderm of stems and roots which function in gas exchange (Esau, 1965; Huang, 1986). Although lenticels may have a suberized inner surface, they also contain intercellular spaces which are continuous between the lenticel and cortex and are therefore permeable to water, oxygen and carbon dioxide, although this permeability may vary at different times of the year (Lendzian, 2006). The lenticels of European horse chestnut have not been examined in detail, and it is not known whether their structure and permeability vary seasonally. However, this study has shown that Psa was able to infect horse chestnut tissues via lenticels in late summer and autumn. It is most likely that bacteria enter the lenticel cavity through the opening in the raised lenticel surface and are able to spread into the cortex via intercellular spaces, although the underlying processes which facilitate this entry are unknown. Lenticels represent a very high number of potential infection sites and Psa appears to have a mechanism which allows it to invade these structures in branches of European horse chestnut.

Primary infection of woody tree parts via lenticels has not been demonstrated for other pseudomonad pathogens and therefore may not be typical for these diseases. For example, lenticels were suspected to be important avenues for infection of cherry trees by P. syringae pv. morsprunorum since Wormald (1930) noted that the majority of cankers occurred on stems in the absence of discernible wounds. However, this was not confirmed by the author in any subsequent studies. Crosse (1955) argued that suberization of the host tissues underlying lenticels would prevent invasion by P. syringae pv. morsprunorum and Shanmuganathan (1962) was unable to induce infection of plum by artificially spraying lenticels with inoculum of this pathogen. Serizawa et al. (1994) were also unsuccessful in their attempts to inoculate lenticels on woody vines of kiwifruit (Actinidia deliciosa) with P. syringae pv. actinidiae. It is possible that Psa has a specific biochemical or morphological adaptation which enables it to penetrate the lenticels of its woody host which is absent in other Pseudomonas tree pathogens. More detailed studies of the processes of lenticel infection by Psa are needed to investigate this further.

In contrast to lenticels, infection of branches via leaf scars is well documented for other bacterial canker pathogens, including P. syringae pv. morsprunorum on stone fruit trees and P. avellanae on European hazelnut (Crosse, 1951; Scortichini & Lazzari, 1996). The susceptibility of leaf scars is highest in the autumn immediately after leaf fall, as cortical and phloem tissues are directly exposed to inoculum before suberization of the leaf scar occurs (Crosse, 1955). Strong winds, which cause premature defoliation, combined with rainfall are the ideal conditions for leaf scar infection by P. morsprunorum (Cross, 1966). Frost cracks and other freeze injuries on woody host parts may also provide infection routes for P. syringae pv. morsprunorum, P. syringae pv. syringae and P. avellanae (Scortichini, 2002; Kennelly et al., 2007). However, injuries of this nature were not observed on European horse chestnut in this study.

Examination of discrete lesions on European horse chestnut branches showed that Psa invaded the cortex and phloem, and killed the cambium in the period between the cessation of active host growth in late summer and the onset of host growth the following spring. This is consistent with studies involving P. syringae pv. morsprunorum, P. syringae pv. actinidiae and P. avellanae which showed that woody host parts are most susceptible to primary infection during host dormancy (Wormald, 1931; Wilson, 1939; Shanmuganathan, 1962; Crosse, 1966; Serizawa et al., 1994; Scortichini & Lazzari, 1996; Scortichini, 2002). Similarly, Bultreys et al. (2008) found that Psa caused larger lesions on twigs of European horse chestnut when inoculated in winter compared with summer. Previous authors have explained these observations in part by the fact that dormant trees are unable to initiate an active structural defence response, which requires meristematic activity (Wilson, 1939; Shanmuganathan, 1962; Crosse, 1966). This allows bacterial pathogens to colonize and multiply in the infection court, whereas actively growing trees are able to seal off infections by the production of a necrophylactic periderm.

Psa evidently survived and remained active for several years within many discrete branch lesions as shown by the positive identification of Psa from lesions which killed the cambium as far back as the 2004/05 dormant season and the continued tangential spread of lesions in the phloem and cortex after subsequent secondary growth in the host. Previously, it has been found that P. syringae pv. morsprunorum can remain localized within infection sites in the cortex and phloem of the woody parts of cherry if bacterial numbers are not high enough to overwhelm the host’s natural defence responses (Crosse, 1966). In this case, P. syringae pv. morsprunorum survives in infection sites during the summer before becoming active again in the following autumn. The fact that Psa survived within discrete branch lesions over several years, as a result of limited expansion of established necroses, is significant since it indicates that the pathogen may retain the potential for systemic host infection for several years following initial colonization.

Dissection of the two whole trees from Glasgow revealed an extensive network of cankers in the phloem and cambium which had evidently formed very quickly within each tree. The original infection sites in the Glasgow trees were not determined since the cankers were continuous and extended from the small-diameter branches to the lower main stem. However, it is feasible that the cankers may have originated from numerous infection sites on the woody parts of the tree where there were bark discontinuities, including lenticels, leaf scars, nodes and side shoots in the crown, and branch insertion points or wounds on the main stem. The morphology of the cankers, which spiralled up the stems of the trees following the line of the vascular tissues, suggests that Psa had spread within the phloem or xylem. Xylem degradation or disruption was not observed in any of the lesions or cankers examined in this study, either in the whole trees or excised branches. However, colonization of the xylem by Psa and systemic transport in these tissues remains a possibility since this could occur without visible disruption of the xylem vessels. Other pathogenic pseudomonads of woody hosts, P. savastanoi pv. savastanoi on olive and P. syringae pv. syringae on plum, have been found to colonize xylem tissues (Roos & Hattingh, 1987; Rodríguez-Moreno et al., 2009), although extensive systemic movement through the xylem has yet to be demonstrated for these pathogens.

The observation that the cankers on the Glasgow trees extended during the host’s growing season contrasts with the observations of discrete lesions on branches in which the cambium was always killed during host dormancy. This suggests that Psa may have a disease cycle similar to other pathogenic pseudomonads of trees which initially invade the woody host parts during host dormancy and may remain localized, overwintering in the bark, before spreading systemically within the host when active growth resumes in the spring (Crosse, 1966; Scortichini, 2002). The processes which allow Psa to spread in the stem and branches of European horse chestnut from the original infection foci remain uncertain. The mechanism of quorum sensing has been recently described in a wide range of plant pathogenic bacteria, including pathovars of P. syringae (Cha et al., 1998). By this mechanism the bacteria are able to sense the density of their population and will only activate the transcription of target genes, including those involved in motility and pathogenicity, once a threshold population has been reached (Cha et al., 1998; Quiňones et al., 2005; Dulla & Lindow, 2008). Thus, it could be speculated that such a mechanism operates during infection of European horse chestnut by Psa, with canker extension occurring once Psa has multiplied to a sufficient cell density to allow it to overcome the host defences.

Psa was positively identified from <50% of host tissue samples associated with lesions, which represents a fairly low recovery rate. Boa & Preece (1978) reported difficulty in isolating P. savastanoi pv. fraxini, the causal agent of ash canker disease, from cankers on infected trees and suggested that this was due to the bacterium being present only in low numbers. The authors of this paper have found that recovery of Psa from diseased European horse chestnut is most successful from fresh, actively bleeding cankers on stems and branches. This is because the necrotic tissues of older cankers can become infected with numerous other species of opportunistic bacteria which tend to predominate on culture plates (Green et al., 2009). In this study, symptomless horse chestnut tissues were not sampled for the positive identification of Psa. Therefore, the possibility that Psa may spread systemically within healthy horse chestnut tissues from an unknown infection point cannot be ruled out. However, this scenario is unlikely given the nature of the observations made in this study.

The oldest branch lesions recorded in this study killed the cambium during the 2004/05 dormant season and the number of new lesions increased in each subsequent year. This seems likely to reflect the arrival of the pathogen in the area concerned and a subsequent increase in local inoculum levels. The data also correspond with the observed progression of the horse chestnut bleeding canker epidemic in Scotland; symptoms of bleeding canker were initially reported in southern Scotland in 2004, although Psa was not isolated from diseased trees until 2006 (H. Steele & G.A. MacAskill, unpublished data). No evidence of cankering was found in the 2003 annual rings within the xylem of these trees, which were evidently susceptible to the disease, lending support to the theory that Psa is a recent introduction (Green et al., 2010), and most probably arrived in the south Scotland study area in late 2003 or 2004.

Tree diseases caused by Pseudomonas spp. are most prevalent in regions with cool, wet climates and the causal pathogens are thought to be spread mainly in wind-blown rain (Crosse, 1966; Hattingh et al., 1989; Scortichini, 2002; Kennelly et al., 2007). The Psa epidemic on European horse chestnut appears to be geographically limited to northwestern Europe where cool, wet climatic conditions prevail, and Psa may well be disseminated in wind or rain since lesions on branches were almost certainly initiated by aerial inoculum. It has been suggested that plant pathogenic strains of P. syringae can be disseminated across very long distances at high altitudes, with deposition in precipitation (Morris et al., 2006), and a similar mode of dispersal by Psa might explain its rapid spread across northern Europe.

This study is the first to demonstrate that Psa invades the woody parts of European horse chestnut directly. The study also indicates that this pathogen has a unique capability for infection of its host via lenticels. Currently, very little is known about the genetic basis for the association of bacterial tree pathogens with their hosts or the complement of genes required for infection and pathogenesis within the woody parts of trees (Kennelly et al., 2007). Given that Psa is highly aggressive and evidently capable of attacking woody branches directly, it may prove to be a useful model organism for further studies to address these questions.

Acknowledgements

We wish to thank Mr Neil MacIntyre, Veterinary Pathology Unit, The Royal (Dick) School of Veterinary Studies, The University of Edinburgh, for preparation of microscope slides, and Mr Glenn Brearley for photographic assistance. We also thank Mr Graeme Golding of Glasgow City Council for arranging the felling and cutting of trees in Glasgow.

Ancillary