Fire blight. Why do views on host invasion by Erwinia amylovora differ?




The fire blight pathogen, Erwinia amylovora, commonly infects flowers and shoots of certain rosaceous hosts and systemic (whole-tree) invasion sometimes follows. The bacterium may be found in the parenchyma of bark tissue and/or in mature xylem vessels of stem tissue. Views differ on initial sites of multiplication and the optimal route for systemic migration. This article presents the evidence on which the different views are based. There are limited observations on orchard pear and apple trees; in most experimental studies, young apple shoots on potted plants were used. Tissue maturity at the site of shoot inoculation is of prime importance. If xylem vessels are damaged, inoculum may be sucked into the vessels and the bacteria will multiply there. In younger tissue, there is less suction pressure. The critical stem entry site for the invasion of cortical parenchyma seems to be near or above the most recently unfolded leaf. No one has suggested that migration in bark tissue cannot be a major route. If the xylem route is followed, the pathogen needs a means of escape into bark tissue, where typical symptoms develop; means of escape from mature xylem vessels have not been demonstrated and remain a matter for speculation. Published evidence does not seem to support the idea that fire blight is a vascular wilt disease, nor that the extracellular polysaccharide produced by E. amylovora is a toxin and responsible for the wilting symptom seen in the early stages of the disease.


Prior to 1965, there was no dispute that, following infection of open flowers or shoots of susceptible rosaceous hosts, the bacterial fire blight pathogen, Erwinia amylovora, moved down the tree in bark tissue. It could also move downward in xylem vessels of stems but, in mature vessels, it might have no means of escape. In such vessels in the wood, it might remain latent. Symptom development (tissue water soaking, bacterial ooze production and emergence, and early necrosis and signs of wilt) depended on multiplication in bark tissue (Vanneste & Eden-Green, 2000). Invaded mature xylem vessels were often discoloured (red or brown). These conclusions were reached from visual observation of blighted trees, often in orchards, and visual tissue observations and light microscopy of inoculated blossoms and shoots. From 1965 to 1981, divergent views emerged in the USA from the Missouri group of workers led by R. N. Goodman, initially using light microscopy and later, electron microscopy. Suhayda & Goodman (1981b) claimed in the abstract of a study on migration in apple shoots that their electron micrographs indicated that ‘The site of early proliferation and systemic migration is in mature xylem vessels rather than in the cortex and pith as reported in the earlier literature.’

In recent years, some authors have accepted the xylem-vessel migration view without question. In his textbook, Agrios (2004) states that E. amylovora usually enters leaf tissues through wounds and initially colonizes and moves through vessels, although it colonizes other tissues later. Subsequently, he states, that in twigs the bacteria travel intercellularly or through the xylem, but invasion of large twigs and branches is restricted primarily to the cortex. Momol et al. (1998) mention previous findings (no reference) of rapid movement of the pathogen in woody tissue. In two recent genetic studies, Rezzonico & Duffy (2007) state that the pathogen moves through the vascular system and may advance into woody tissue leading to death of limbs, whilst Norelli et al. (2009) talk of rapid spread through the plant via vascular tissue and necrosis of woody tissue. Moltmann & Viehrig (2008) cite the suggestion of Vanneste & Eden-Green (2000) that the pathogen might escape from xylem vessels to cortical tissue at the rootstock or at shoot tips, and suggest escape into blossoms also. Goodman & White (1981) showed how escape can occur in young petiole tissue, but not when the pathogen was restricted in mature xylem vessels. Zamski et al. (2008) believed that, following flower colonization, the pathogen invades the intercellular spaces of the cortical parenchyma and then reaches the vascular system (implying that this was necessary for migration). In all these cases, entry to the vascular system is deemed an essential feature of migration and the possibility of migration in the cortical parenchyma or inner bark tissues is not always considered, nor is the means of escape from mature xylem vessels.

Vanneste & Eden-Green (2000), in a review of migration research, found no consensus on how E. amylovora migrates and invades tissues of a host plant. They considered that one problem is that a mature tree is difficult to study, so, for experimental purposes, young plants have to be used. Further problems arise concerning inoculation methods and methods for detecting the pathogen in plant tissues. The authors emphasized that the site of early multiplication might be determined by the port of entry. Other important issues described by these authors are discussed later in this report. It is evident from many studies that the fire blight pathogen can colonize both bark and woody tissue, notably in the intercellular spaces of the cortical parenchyma and in mature xylem vessels. The point of contention is that of the route taken when the pathogen migrates from high in the tree to main stems, the trunk and even the rootstock (systemic invasion). Some maintain that entry into xylem vessels is essential for systemic invasion, others that escape from mature xylem vessels into bark tissue (where typical symptoms are expressed) has not been demonstrated and that bark invasion is essential for migration. In spite of the warnings of Vanneste & Eden-Green (2000), views have remained polarized, so fire blight management rests on uncertainty. In this report, available evidence is re-examined so that readers can better judge the strengths or weaknesses of the arguments on which the conflicting views are based.

Tree anatomy

Information on the anatomy of fire blight host species and cultivars is sparse. Microscopic examination of healthy and diseased tissue of young shoots has presented no major problems, but good preparations of older woody stem tissues are hard to obtain and rarely attempted (Rosen, 1929). For the purpose of this review, the following broad assumptions have been made, based mainly on descriptions of plant anatomy by Esau (1977). Tree bark includes all tissues outside the vascular cambium. Beneath the external dead tissue, the outer bark may contain a layer of living cortical parenchyma cells, the cortex. The inner bark of older stems contains secondary functional phloem tissue, which includes sieve tubes, parenchyma cells and fibres. There is also parenchymatous ray tissue, which is concerned with radial transport of assimilates as opposed to the axial transport by other parenchymatous tissues. Intercellular spaces in parenchyma tissue may be large, allowing E. amylovora to multiply freely there and invade tissue where the cells are still expanding (Link & Wilcox, 1936). Parenchyma cells are living when mature and may resume meristematic activity under some conditions.

The term ‘woody tissue’ is assumed to mean secondary xylem (wood) unless otherwise stated.

Invasion of parenchyma tissue

Field observations

Early reported observations were mainly on pear trees. Burrill (1880) stated that, in very young tissue, all parts except the epidermis were equally invaded. In older limbs, the bark parenchyma was the first and usually the chief seat of disease. Waite (1896) noted that the disease ran down the living bark to the older limbs and then to the trunk. O’Gara (1912) said that new shoots were attacked at the tip and that the disease ran down the bark of the shoots and sometimes down into main limbs. It had a tendency to develop in the green outer bark; sometimes the inner bark was untouched. The bacteria invaded the intercellular spaces, which O’Gara called ‘the vessels of the bark’. Medullary rays were also invaded and movement might be very rapid along the line of vascular tissue (not necessarily in it). However, in the rich sapwood of the Bartlett pear, the bacteria were occasionally seen to spread into the vessels. O’Gara also reported that blight might start in the axils of leaves. Day (1930) claimed that in pear trees, the blight first travels in the outer bark without affecting the cambium, when the canker could be cured with zinc chloride solution before the blight runs deeper. Smith (1920) said that the pathogen passed quickly downward by way of bark parenchyma, into the inner bark of larger branches and the trunk. In the shoot, it was largely a bark disease. Heald (1915) noticed that some infections on undamaged leaves started at lateral or terminal margins. He believed that the bacteria entered intercellular spaces through water pores or stomata. There was a tendency for the bacteria to advance rapidly down the lateral veins or the midrib (not necessarily in the xylem vessels). Hotson (1916) and Pierstorff (1931) and R. A. Lelliot (personal communication) also examined natural infections on undamaged leaves, but noted that they often failed to lead to typical shoot blight. Hotson (1916) sometimes saw reddish discoloration in the sapwood; these observations are described more fully later. Rosen (1929) also described cases where the pathogen was found in symptomless sapwood.

Widespread and severe outbreaks of shoot blight on orchard trees, hawthorns and other hosts have been associated with thunderstorms with heavy rain and hail or strong winds during the period of rapid shoot growth (Thomson, 2000; Billing, 2006). Storms may occur during daylight hours when trees are transpiring, or at night when the tissue water potential approaches zero and suction of inoculum into xylem vessels is unlikely; they are often accompanied by heavy rain. Shoots will be damaged in various ways and this damage is likely to greatly increase chances of infection, especially with low doses of inoculum. Highly susceptible folded leaves at the shoot tips may be bruised or torn and unfolded leaves may be torn off, exposing leaf traces and allowing direct entry into stem tissue. Some unfolded leaves may be torn so that xylem vessels are exposed in leaf veins. It could be valuable if, following a storm, the incidence of the different types of damage were assessed and the relative importance of each in relation to subsequent disease development determined. Late-season storms, prior to normal leaf fall, may tear off leaves and expose leaf traces to infection; few signs of disease may be seen before budbreak the following spring (Billing, 2008). Leaf-trace infections have been better studied in relation to other fruit tree diseases.

Experimental approaches

Ideally, migration studies should be made on field trees, but this is rarely possible. Vanneste & Eden-Green (2000) discussed the difficulties faced by the investigator. With most studies, experiments have to be made under greenhouse conditions. To locate the pathogen in host tissues, most workers have used light microscopy. Goodman and co-workers later used electron microscopy. Replication of microscopic observations is rarely mentioned and statistical analyses are not required. Electron microscopy is a specialist field, so repeat observations in several laboratories are unlikely. For publication, only a few fields can be illustrated using either method and selection bias may be tempting.

Bachmann (1913) inoculated pear flowers with a drop of culture without damage and observed migration in the intercellular spaces of the pedicel after blossom invasion. Rosen (1936) sprayed inoculum onto pear and apple flowers in the orchard and onto flowers on cut twigs steeped in water in the greenhouse. He too observed migration in the intercellular spaces of invaded tissue beneath the nectarial surface in both cases. Tullis (1929) studied factors favouring stomatal entry into undamaged leaves. Unlike Brooks (1926), who failed in his attempts, Tullis wetted leaves before and after inoculation. Following stomatal entry, the pathogen proceeded rapidly downward through intracellular spaces along the line of a vein. The underside of the petiole was then invaded through mesophyll-like tissue. Advance was always along the line of a vein. On the lower leaf surface, stomata are formed and, shortly afterwards, intercellular spaces appear in the mesophyll. Success of artificial inoculation depended on the age of the leaf or the portion of the leaf inoculated. Infection only occurred on young leaves with functional stomata; on very young leaves with undeveloped stomata, there was no infection. As soon as a leaf or any part of it was mature, that leaf or part of the leaf was not susceptible (Tullis, 1929). This would apply to the most recently unfolded leaf (leaf 0) and older ones, but not to folded leaves. Billing (2008) found that infection without damage only occurred on folded leaves, possibly via hydathodes and/or stomata. Hydathodes (modified stomata) are found on the upper surfaces at the tips of the serrations; they develop near the ends of vascular strands. In England, R. A. Lelliott (personal communication) sometimes saw dense populations of E. amylovora in the hydathodes of pear leaves in orchards.

Rosen (1933) was surprised that many able investigators failed to produce infections in the absence of wounds. He concluded that, in the light of work by Miller (1929) and Tullis (1929), many failures were the result of improper understanding of the relationship of the age of host tissue to infection, and failure to grasp the importance of sufficient water in the tissues and on the surface. Those workers (in chronological order) who failed to produce shoot infections without damage include: O’Gara (1912), Brooks (1926), Pierstorff (1931) and Crosse et al. (1972). It was suggested by Miller (1929) that Brooks had used shoots that had made too much growth; he also used shoots of two apple cultivars with hairy leaves, which might be more difficult to infect. Crosse et al. (1972) did not describe the methods that they used, so reasons for their lack of success remain uncertain. Both Brooks (1926) and Crosse et al. (1972) concluded that damaged veins with exposed xylem vessels were the most important sites for shoot infection. Other workers had varying degrees of success in producing infections without wounds (Hotson, 1916; Miller, 1929; Rosen, 1929; Tullis, 1929; Billing, 2008). The rates of infection were often low and usually too inconsistent for routine use.

The apple shoot model

For convenience and reliability, most experiments concerned with tissue susceptibility and migration routes have used young apple shoots as the model and introduced the pathogen using damage, either by needle inoculation into stem tissue or by cutting leaves or petioles so that the pathogen enters exposed xylem vessels. At that time, some bacteria may also enter parenchyma tissue in small numbers.

Various factors affect shoot susceptibility. For consistent findings within and between laboratories, shoots should be as uniform as possible, and it is important to understand how this can best be achieved. Link & Wilcox (1936) described factors that might affect shoot susceptibility, first, the maturity of the inoculated part. The most susceptible tissue was that where there was cell division and enlargement, and cell walls were thin. They claimed that as blight advanced down the stem, it progressively reactivated meristematic tissue and cambium again gave rise to thin-walled cells in cortical, vascular and ray tissue. Overall, host susceptibility depended on the initial amount of susceptible tissue, its retention as the tissue matured and any resumption of meristematic activity. High nitrogen levels would aid these processes and so too would shading.

For maximum host susceptibility and consistency, Shaw (1934) considered that techniques should be standardized, including conditions such as soil moisture, nutrients, soil and air temperature, and lighting; shading should be controlled. Most such conditions were considered by Eden-Green (1972), but rarely described by other workers. There are indications that, in some Missouri experiments, plants may have been exposed to sunlight at the time of inoculation. Lewis & Goodman (1965) shaded plants after inoculation; Goodman & White (1981) inoculated plants on a sunlit day. For uniform shoot susceptibility, Eden-Green (1972) and Aldwinckle & Preczewski (1976) trained each potted plant to a single shoot. Shaw (1934) recommended one to two shoots. In other reports the number of shoots per tree is not specified and statements are confined to 1-year-old trees and/or young succulent shoots or rapidly growing shoots. Where there are several shoots on a potted tree, they might differ slightly in maturity. Shaw (1934, 1935) stressed the importance of high intercellular humidity, for which the watering regime would need to ensure a high level of soil moisture.

It is convenient when considering leaf maturity to label the youngest unfolded (unrolled) leaf as leaf 0 (L0) and older leaves down the stem as L1 to L6. Leaf 0 will not be fully expanded and will, like folded leaves, still be importing nutrients. The same may apply to a lesser extent to leaves L1 and L2. Folded leaves can be numbered, if necessary, as L–1 to L–6. From the above account, the susceptibility of L0, in particular, might vary between replicate shoots and separate experiments, and between laboratories.

Methods of inoculation

Needle puncture is the usual method for stem inoculations and may introduce a measured dose. The needle will pass through both the cortical parenchyma and xylem vessels. Establishment of infection in either or both will depend on their susceptibility at the site of inoculation.

Cut-leaf and cut-petiole inoculations were the methods used by the Missouri group (Crosse et al., 1972; Huang et al., 1975; Huang & Goodman, 1976; Goodman & White, 1981; Suhayda & Goodman, 1981a,b). Cut surfaces were either immersed in inoculum or drops of inoculum placed on the cut surfaces of the midvein or the petiole. The droplets were rapidly sucked into the exposed xylem vessels by negative pressure during transpiration, except in the case of younger cut leaves. In the xylem vessels, a few bacteria were transported some distance downwards, but most remained near the site of inoculation. Because of the variation in suction rate, invasion rates would be difficult to judge.

Relation of the method and site of inoculation to the tissue invaded

When Suhayda & Goodman (1981b) questioned the conclusions reached by earlier workers, they only cited Bachmann (1913) and Nixon (1927) and did not consider the fact that their own method and sites of inoculation differed from those of these early workers. In fact more than 10 other workers (listed later) observed migration in the cortical parenchyma. Enough of these workers described their inoculation methods in sufficient detail for comparison with those of the Missouri Group. Using needle inoculation, Nixon (1927), Wahl (1932) and Link & Wilcox (1936) inoculated shoot tips, Miller (1929) near the shoot tips, and Eden-Green (1972) at the first node from the tip. When comparing apple cultivar shoot susceptibility, Aldwinckle & Preczewski (1976) inoculated just above L0, aiming at typical shoot blight symptoms. Their observations are described later in relation to xylem vessel staining. In early migration studies, Gowda & Goodman (1970) inoculated stems just above the first node from the tip. In their ringing experiments, they found that migration downward occurred in the bark tissue, not in the xylem.

By contrast, other observers in the Missouri group, who favoured xylem vessels as the primary site of early multiplication and migration, inoculated cut L0 or older (Crosse et al., 1972) or cut petioles (Huang et al., 1975; Huang & Goodman, 1976; Goodman & White, 1981; Suhayda & Goodman, 1981a,b). In the experiments described in 1975 and 1976, some invasion of cortical parenchyma was seen. Suhayda & Goodman (1981b) also inoculated stems 5 cm below the shoot tip (likely to be below L0).

Thus, it would seem that stem inoculation above L0 is likely to favour invasion of cortical parenchyma, whilst inoculation at L0 or older with xylem vessels exposed, or lower down the stem, is likely to favour invasion of xylem vessel. However, near the L0 level, sometimes both xylem vessels and cortical parenchyma may be invaded (Huang et al., 1975; Aldwinckle & Preczewski, 1976; Huang & Goodman, 1976).

Evidence for migration in the intercellular spaces of the cortical parenchyma

Apart from Bachmann (1913) and Nixon (1927), cited by Suhayda & Goodman (1981b), the following observers reported migration in the intercellular spaces of the cortical parenchyma; most worked in different laboratories in the USA or in Europe. In chronological order, they were: Heald (1915), Haber (1928), Miller (1929), Tullis (1929), Rosen (1929, 1936), Reid (1931), Wahl (1932), Link & Wilcox (1936), Hildebrand (1937), Eden-Green (1972), Hockenhull (1974), Huang et al. (1975), Huang & Goodman (1976). This makes a total of 15 groups, which is difficult to ignore. To these names might be added: Shaw (1934, 1935), who was concerned about relative humidity in the intercellular spaces, and evidently considered this to be a key migration route, and Gowda & Goodman (1970), who did bark ringing experiments which demonstrated that the pathogen moved downwards in bark tissue (phloem or cortical parenchyma). This experimental evidence is also in line with the field evidence described earlier that migration in older stems is seen to occur in the bark, rarely in woody tissues.

The complete genome of E. amylovora was recently sequenced (Smits et al., 2010). Observations made in the first 72 h after inoculation in some of the above accounts might be of interest in genomic studies.

Late-season leaf-trace infections and frost effects

If leaves are torn off, stem vascular tissue is exposed at the leaf trace. Such infections may occur, prior to normal leaf fall, during damaging storms late in the season, usually in September, and be responsible for the development of mature shoot (twig) blight (Billing & Berrie, 2000; Billing, 2008). Leaf-trace infections have not been studied microscopically in the case of fire blight, but they have been well studied in the case of other fruit tree diseases which may provide useful models (see below). Cases of late twig blight will not always be detectable during pruning and may be important sources of inoculum high in the tree during flowering, or they may extend and cause blight in developing leaves and blossoms below a canker.

Cherry canker caused by Pseudomonas syringae pv. morsprunorum

Crosse (1955, 1956) studied autumn infections of cherry trees via leaf traces. Most infections occurred in September or October, before normal leaf fall. Wind-blown rain favoured both spread of inoculum and exposure of leaf traces when leaves were torn off. Disease incidence was related to the rate of suction into xylem vessels, which could be high after a dry summer. Disease spread from leaf-trace infections to branches during the winter. In the early stages, there was limited invasion of the cortex and medulla, but later there was extensive necrosis of these tissues and spurs and buds were invaded. Branch invasion progressed until the end of May. Where bacteria were confined to the vascular strands, there was very little necrosis and buds developed normally. Initially, considerable bacterial growth occurred in the intercellular spaces of the cortex without apparent damage to host cells. P. syringae pathovars grow more quickly at low temperatures than E. amylovora, so tissue invasion would be slower in fire blight infections of a similar nature.

Peach canker caused by Xanthomonas campestris pv. pruni

Feliciano & Daines (1970) studied factors influencing the ingress of the pathogen through peach leaf traces. Again, wind-blown rain was implicated in the spread of inoculum to fresh leaf traces when wind tore leaves off. The tops of twigs had the highest incidence of cankers. Spring cankers developed with inoculations made between 17 September and 1 November. The intercellular spaces of cortex, phloem and xylem parenchyma, were invaded and also phloem fibres, within a few days. There was no evidence of escape of bacteria sucked into xylem vessels. In spring, the bacteria were in the innermost layer of the cortex in the zone where the parenchyma was rich in starch cells and abutted fibres. The cortex and phloem tissues were then invaded via parenchyma cells along fibres. Invasion of xylem cells appeared to be of little pathological importance. There was no evidence that the pathogen moved from xylem cells to adjacent healthy vessels. There was no evidence of assistance from pectic or cellulytic enzymes.

Olive knot disease caused by Pseudomonas syringae pv. savastanoi

In a study by Hewitt (1938), a large number of knots developed at leaf traces, following infection, and many bacteria exuded in a slimy mass from fissures of knots and were spread by rain. At the time of infection, the pathogen penetrated deeply in xylem vessels. This was the only route of entry and the vessels became filled with bacteria. The bacteria were freed from the vessels when the forming periderm pulled the vessels apart. The bacteria were then released into the newly formed periderm, a region of actively dividing cells. Bacteria confined deeply in the vessels did not break out. The presence of the bacteria in the periderm seemed to stimulate cell division there [as Link & Wilcox (1936) observed in the case of fire blight] and successive rows of parenchyma cells were formed. Pockets formed which increased in size and grew with surrounding tissue.

Pear canker caused by Pseudomonas syringae pv. syringae

Severe autumn storms can lead to canker development on pear trees caused by this organism (Wilson, 1934). There may also be death of buds due to leaf trace infections. These bacteria grow at lower temperatures than E. amylovora and cankers develop slowly from winter to spring. At the start, brown streaks can be seen in the cortex and the outer phloem; in some seasons, they progress to the inner bark. The cankers differ from those of fire blight in several ways, but misdiagnosis is a possibility. The pseudomonad cankers develop in cool weather, the discoloration of the bark is brown, not reddish brown, and the bark becomes spongy, not water-soaked; brown periderm may loosen and slough off (Wilson, 1934). In England, unusually extensive P. syringae pv. syringae cankers were seen on pear trees in 1959 following a severe storm on 5 September 1958, which was also important in early outbreaks of fire blight in southeast England (Billing, 2006).

Peach and apricot die back caused by Pseudomonas syringae pv. persicae

Vigouroux (1989) noted that plants adapt to low temperatures (−1 to −10°C) by transferring cellular water to intercellular spaces where it freezes (Burke et al., 1976). When the temperature rises, the ice thaws and continuous films of water form. It has long been known that a water-soaked condition promotes entry and spread of bacteria in plant tissue. Vigouroux warned against pruning just before a frost period, suggesting that the resurgence of latent internal populations might be favoured by frost and promote expansion of existing cankers. In England, copious ooze was seen on a pear stem in January, following several days of hard frost (H. H.Glasscock, personal communication). Blisters or ooze have been seen at times on hawthorn twigs following frosts of < 2°C (personal observations). Frost records are, therefore, important in fire blight warning systems. Frost injury is a predisposing factor for pear blossom blight caused by P. syringae pv. syringae.

Invasion of xylem vessels

Field observations

Hotson (1916) noticed that, in fire blight infections in pear trees, some shoots and branches showed a reddish colour in the sapwood and sometimes in the pith. Such branches were not more seriously affected than hundreds of others that did not show this peculiarity. The red colour was evident 12–18 inches beyond where the bark showed discoloration. For disease control, the grower needed to cut beyond the red streak. The streaks were never found on normal branches or shoots, only where there was evidence of blight. The pathogen was isolated from red streaks, but never in the abundance found in diseased bark. No bacteria were found towards the lower limits of the red colour. This was in contrast to affected bark tissue, where the pathogen was found regularly beyond the discoloured tissue. The red streak was found 6 or 8 ft from the tip of a branch, but no bacteria were found there. Nearer the tip of the branch, however, often 2 ft from it, the bacteria were plentiful. Sometimes, in branches of the preceding year’s growth that were over an inch in diameter, the streak could be detected readily. In such cases, the infection could be traced to a small lateral branch, bacteria gaining access to the xylem tubes near the tip and then working their way into the larger branch. The colouration extended beyond the region of the organisms. When tips of healthy branches were cut under water and immediately placed in a bottle of eosin, a 4-oz bottle was emptied within a few hours. To a certain distance, the colouring matter was evenly distributed through the sapwood, wood and pith, but it had travelled further in the sapwood than in the pith. There was practically no trace of it in the phloem tissues of the bark. Thus, liquids might pass readily in the sapwood from the tip of a branch backward, at least when weather was extremely warm and dry. Rosen (1929) cited several early observers who occasionally found the pathogen in the symptomless sapwood. He believed that vascular invasion was far more common than was generally believed, but he did not investigate the problem in detail.

Experimental observations

Shaw (1934) inoculated apple shoots 1·0 cm from the tip by pricking growth from a slope culture with a needle. Shoots were examined visually for external signs of disease and, in some cases, microscopically and by culturing the pathogen. Shaw found that the pathogen was commonly present in the xylem and pith at considerable distances towards the basal ends of shoots beyond cork layers. It was not found in the cortex, phloem or cambium at the basal ends. Movement out of the xylem there appeared to occur in only seven out of about 800 shoots on potted plants in the greenhouse. Cortical blight occurred at the base of a leaf petiole towards the basal end of the shoots beyond the cork layer. R. A. Lelliott (personal communication) studied the persistence of E. amylovora in young pear trees growing in soil in a polythene tunnel. Blight that developed, after spraying trees with inoculum, was cut out and the trees then remained symptomless. Three years after inoculation, microscopic examination showed that many bacteria were present deep in the wood. When cultured, they proved to be typical, pathogenic E. amylovora.

Crosse et al. (1972) examined various aspects of the entry of E. amylovora into xylem vessels when leaves on growing apple shoots on young trees in greenhouse experiments were cut to expose veins. During daylight hours, shoots will be transpiring and there will be negative pressure in the xylem vessels, so drops of inoculum applied to the cut ends are likely to be sucked into the vessels. The tension will vary according to leaf age, shoot maturity and the conditions of cultivation of the plant. Tension will tend to be low when soil moisture levels are high and when transpiration rates are low, for example, at night. There was no direct evidence of the depth of penetration of the bacteria but, using stained bacterial suspensions, bacteria were observed in stained xylem vessels up to 3·0 mm below the cut edge of the leaf. It was certain that some penetrated much deeper, possibly as much as 30 mm. Following inoculation, typical blight symptoms often appeared at the shoot tip within 6–12 days. Whether or not the bacteria responsible for these symptoms migrated to the shoot tip in the xylem vessels was not determined. When drops of inoculum were applied to the cut ends of leaves, the drops disappeared within 5 min, except in the case of the youngest unfolded leaves where they persisted longer and some of the inoculum could be lost through spillage. However, in later experiments using cut petioles of L0 leaves and inoculation at 09·00–10·00 h on a sunlit day, droplets were absorbed within 2–4 min (Goodman & White, 1981). Inoculation of the youngest leaves was most likely to give rise to shoot-tip infection. This occurred with 61% of shoots inoculated at L0, 47% at L1 and 11% at L2. Inoculation at L3–L5 did not lead to blight at the shoot tips. To give rise to typical symptoms there, the pathogen must have invaded cortical parenchyma at some stage. The authors claimed in the abstract of their paper that there was evidence that the bacteria migrated in the xylem vessels to the shoot axis (the stem) but their microscopic evidence did not go beyond a petiole whose leaf age was not stated.

In field and greenhouse tests on the fire blight susceptibility of 92 apple cultivars, Aldwinckle & Preczewski (1976) often observed discoloured streaks in the xylem. Grafted trees were trained to a single shoot to ensure uniformity. The shoot tips were inoculated using a needle just above the youngest unfolded leaf, thus favouring cortical parenchyma invasion. Fire blight lesions were measured 6 weeks after inoculation when visible development of blight had ceased. Many individual shoots had longitudinal dark brown streaks in the stele extending further than the visible cortical lesions to a varying extent. Histological studies indicated that the streaks were confined to the primary xylem. Bacteria were seen there and pathogenic E. amylovora were isolated from such tissues. Discoloured streaks were more common in resistant than in susceptible cultivars. The bacteria were presumed to have gained access to the primary xylem vessels of the stem at the time of needle inoculation. This type of damage would not be normal under field conditions, but it was suggested that entry to xylem vessels near the shoot tip might sometimes occur as a result of other types of damage, e.g. during storms, or, in nurseries during budding, grafting or pruning.

Momol et al. (1998), using the same inoculation method as Aldwinckle & Preczewski (1976), investigated the possibility that the pathogen might migrate directly from an infected scion into the rootstock. In one series of experiments where scion shoots were inoculated in July, the pathogen was later detected in the rootstocks of five out of six plants, but no symptoms developed. The migration route was not investigated, but the pathogen had moved through stems up to 3 years old in either the bark or in the sapwood. Their statement that their results ‘supported previous findings of rapid internal movement in woody tissue’ (no reference given) does not seem justified by the evidence presented.

Suhayda & Goodman (1981a) considered fire blight to be a vascular wilt disease and that bacterial multiplication in the xylem vessels was the primary event required for the development of wilting symptoms and subsequent spread of the pathogen in host tissue. One objective of their work was to determine the distribution and rate of movement of virulent and avirulent strains of E. amylovora in apple shoot tissue using 32P-labelled bacteria. Only the results obtained with the virulent strain are considered here. Migration in the petioles was studied by placing a bacterial suspension on the cut petiole surface immediately after excision. Movement of the pathogen in 5-cm-long apple shoots was also examined by immersing the cut stem bases in labelled inoculum. In petioles (as might be expected) some bacteria were rapidly drawn into xylem vessels and moved 5 mm in 20 min. Most of the inoculum remained at the site of inoculation. In the cut shoots, the pathogen was judged to have moved up the stem at a rate of 34 mm h−1. It was concluded that leaf damage that exposes xylem vessels could be an important factor in the infection process under natural conditions. These findings are in line with those of Crosse et al. (1972).

The objectives of another study by Suhayda & Goodman (1981b) were: to determine the tissue in which bacterial proliferation occurred initially, to relate the proliferation to extracellular polysaccharide (EPS) production by virulent strains and to determine whether or not there was a link between EPS production and vascular occlusion. Apple shoot stems were inoculated 5 cm below the shoot apex on the assumption that this would give an equal opportunity for infection to occur in the intercellular spaces of the cortical parenchyma as well as in xylem vessels. That distance suggests a region well below L0 where the cortical parenchyma cells might no longer be expanding rapidly and not be highly susceptible to infection. It is not clear how many shoots were examined nor their degree of maturity; they were described only as young shoots on trees. Using transmission electron microscopy, the pathogen was seen to colonize xylem vessels and, 48 h after inoculation, the vessels were blocked. This coincided with signs of wilt in the inoculated shoots. The authors concluded that the blocking material was largely EPS, that this probably restricted water flow in the vessels and that this was responsible for the wilting. There was no evidence of multiplication of the pathogen in the cortical parenchyma beyond the wound site. The authors concluded that: ‘The site of early proliferation and systemic migration is in mature xylem vessels rather than in the cortex and pith as reported in the earlier literature.’ The weakness of this argument is that entry to xylem vessels was likely at the site of inoculation and the cortical tissue may not have been highly susceptible there. To test their hypothesis, they did not make parallel observations at less mature inoculation sites. Several questions arise. Were their microscopic studies sensitive enough to detect migration in the cortical parenchyma to the shoot tip by very small numbers of bacteria? Does this study justify the dismissal of earlier studies where younger tissue would have been inoculated? Wilting is not the only symptom associated with shoot blight; early signs include ooze production and necrosis as well as wilt, which are associated with colonization of cortical parenchyma, where invasion of xylem vessels may not have occurred.

Goodman & White (1981) wanted to find an explanation for early observations by Huang & Goodman (1976) and others concerning the parenchyma cells surrounding the xylem vessels. Inoculum drops were placed on cut leaf petioles (L0) at 09·00–10·00 h on a sunlit day so that drops were absorbed in 2–4 min. After 24 h, plasmolysis of xylem parenchyma cells was seen using electron microscopy. This was most common in the cells that bordered the xylem vessels containing bacteria. At this time there were no external signs of disease on the petioles. At 48 h, protoplasts of the xylem parenchyma cells were further aggregated and distorted. Diminished turgor in several files of parenchyma cells allowed adjacent xylem vessels to twist and rupture. This released bacteria into what had been intercellular space that had developed from the collapse of the xylem parenchyma and lysigenous cavities developed. There was then rapid growth and movement of the bacteria. The authors suggested that the loss of turgor in the parenchyma cells was the result of plasmolysis induced by EPS production by the pathogen. Visual signs of petiole reddening and necrosis usually appeared 72 h after inoculation. Plasmolysis of xylem parenchyma was also reported by Bachmann (1913). This report showed how the pathogen may escape from xylem vessels in young petiole tissue, but not from vessels near the base of a shoot or in older invaded stems.

In Germany, Bogs et al. (1998) used small apple seedlings in a growth chamber. Such delicate tissue enabled them to detect the pathogen marked with genes for bioluminescence and fluorescence. Leaves were inoculated by cutting off the tips with contaminated scissors, puncturing the petiole with a needle or by applying filter paper discs soaked in inoculum to avoid damage. Bacteria were observed in xylem vessels; they later broke out into intercellular spaces. The pathogen moved from inoculation areas as far as the roots. Entry to leaf tissue occurred at the base of leaf hairs when inoculum was applied without damage. How the results relate to orchard tree infections is uncertain. Unfortunately, this approach can only be used with very young tissue.

Vascular wilt disease

The fact that E. amylovora can colonize xylem vessels and that cut shoots wilt when placed in solutions of its EPS is not disputed. These observations, however, seem to be the main reason for the simple statement by Suhayda & Goodman (1981a) that fire blight is a vascular wilt disease. In a later publication, Goodman (1983) does not use the term ‘vascular wilt disease’, but maintains the view that all the symptoms are caused by the fact that the EPS is a toxin. Leigh & Coplin (1992), describe E. amylovora alongside a typical vascular wilt pathogen, Pantoea stewartii (the cause of Stewart’s disease of maize); both, they suggest, are able to block xylem vessels to cause wilting in their hosts. Agrios (2004) includes fire blight in the vascular wilt disease group, but he notes that, in contrast to other bacterial wilts, the bacteria move rapidly to other tissues.

There are various reasons why other pathologists do not consider fire blight to be a vascular wilt disease. With Stewart’s disease, when stems are cut, yellow slime oozes from the vascular bundles. The foliage shrivels gradually, the lower leaves usually first (Smith, 1920). In the case of E. tracheiphila, which causes cucumber wilt, leaf wilt is sudden and white bacterial ooze emerges from xylem vessels when stems are cut (Smith, 1920). In brown rot of potatoes and other hosts caused by Ralstonia solanacearum, the foliage may wilt completely. On potato tubers, the ring of brown stain and the bacterial ooze are limited to the outer vascular part of the tuber (Smith, 1920). In such diseases, root infection may be a feature and the pathogen migrates upwards from there; it may not, therefore, always be isolated from the wilted top of the plant (Vanneste & Eden-Green, 2000). By contrast, in fire blight, bacterial ooze emerges from bark tissue and not from xylem vessels exposed when cut. On apple and pear fruit slices, ooze appears over the whole surface of the slice. Signs of wilt are first seen at growing shoot tips where the pathogen is present and may extend downwards to some lower leaves, but not necessarily to all. In potted young trees where roots have been inoculated, upward disease progression in stems is slow and limited (Gowda & Goodman, 1970). Sjulin & Beer (1978) studied the effect of the EPS on cotoneaster shoots. They confirmed that, when cut shoots were immersed in a solution of EPS, wilting was induced by the restriction of water uptake. Dextran solutions and blocking the shoot base produced similar results. The water relationships of such shoots were not comparable to those infected with E. amylovora. The authors concluded that the wilt of naturally infected shoots resulted from disruption of membrane integrity rather than by restriction of water flow. Vanneste & Eden-Green (2000) concluded that, in natural infections, wilting associated with pathogen invasion of young shoots (sometimes described as ‘shepherd’s crook’) is not the result of disruption of water flow following plugging of water flow by the pathogen or its EPS. Rather, it is caused by collapse of parenchyma. Drops of exudate may sometimes be seen before signs of wilt and necrosis.

The term ‘vascular wilt disease’ was never used by early fire blight workers, nor is it used for fire blight by Lelliott (1988) or Vanneste (2000). Goodman himself ceased to use it whilst still maintaining that the EPS was a toxin (Goodman et al., 1974) and responsible for tissue discoloration, plasmolysis of xylem parenchyma cells and necrosis, as well as wilt (Goodman, 1983). Eden-Green & Knee (1974) found no evidence that the EPS had a toxic effect on apple fruits or potato discs. There is, however, general agreement that the EPS is an essential pathogenicity factor for E. amylovora.

Phloem migration

Lewis & Goodman (1965) and Gowda & Goodman (1970) claimed that they had demonstrated migration in phloem tissue of growing shoots. In their experiments, they removed a segment of bark from shoots which included outer bark down to the xylem (bark ringing). This would have prevented migration of E. amylovora both in the phloem and in the cortical parenchyma. They made no mention of the latter tissue and concluded that migration occurred in the phloem sieve tubes. In a case study by Goodman (1983), phloem migration was no longer suggested and the work of Gowda & Goodman was not cited. No other workers have reported direct or indirect evidence of phloem migration in growing shoots. Ge & van der Zwet (1996), Momol et al. (1998), Hickey et al. (1999) and Agrios (2004) considered that the phloem migration evidence was valid. However, Momol et al. (1998) were sometimes working with 3-year-old stems, so migration in secondary phloem tissue was a possibility there, especially late in the season. There remains an untested possibility that the fire blight pathogen might move in association with phloem fibres, as reported in the case of peach canker (Feliciano & Daines, 1970).

Latent and inapparent infections

Bud infections

Baldwin & Goodman (1963) claimed that they had isolated avirulent E. amylovora from healthy apple buds including terminal, lateral and fruiting spur buds. Unfortunately, they used non-specific phages to identify the pathogen (Billing et al., 1960; Thomson, 2000) and their isolates are likely to have been common epiphytic bacteria. Cultures sent by Goodman to the UK National Collection of Plant Pathogenic Bacteria were not agglutinated by the antiserum specific for E. amylovora, nor was their colony form typical of that pathogen (R. A. Lelliott, personal communication). A few other workers have claimed that they isolated the pathogen from buds; Thomson (2000) wrongly cites Eden-Green & Knee (1974). If bud samples contain fragments of leaf traces, false positive results are possible. Schroth et al. (1974) failed to find the pathogen in thousands of buds examined.

Canker blight

If latent infections are taken as infections that might become active under some circumstances, they could include overwintering cankers and canker blight (a term coined by Steiner, 1990). It is used to describe movement of the pathogen from cankers into nearby shoots or limbs. According to Steiner, the earliest evidence on apple trees is the appearance of necrotic areas along the petiole or midvein of leaves on shoots near the canker site. Later, the growing shoots develop a distinctive orange discoloration at the tips and slight wilt. The intervening stem and leaves may appear symptomless; the whole shoot then dies. Unless examined carefully for the tip discoloration, these shoots can easily be mistaken for ordinary shoot blight where direct infection occurs at or near the tip and moves down the stem. In canker blight, disease moves upward from the canker. Hickey et al. (1999) saw such cases in orchard trees. On 18 June 1997, in a Virginia orchard, at least 10 shoots showed typical symptoms of canker blight. Some shoots without orange tips showed early necrosis along the midribs of leaves near the middle of the shoots. They cultured parts of 3·0-mm-thick stem segments from symptomless shoots at different locations on the trees for the presence of E. amylovora with varying degrees of success. In West Virginia, four orchard trees showed numerous shoots with typical canker blight symptoms. Again, isolations of the pathogen were made with varying degrees of success. Many of the blighted shoots were 1·0–3·0 m from active cankers. The authors concluded that the shoot-tip symptoms resulted from endophytic bacteria present in the vascular system of the shoots, but their samples would have included bark tissue. E. Billing (unpublished data) in 1973 saw sudden shoot-tip wilt and necrosis on a greenhouse-grown apple tree where a canker had formed following a blossom infection. The blighted shoot, which had emerged from older stem tissue above the canker, externally appeared symptomless, but the bark tissue was discoloured (it was not cultured). On another shoot, there were signs of ooze and early necrosis on petioles at the base of the shoot and below the wilted shoot tip but, at first sight, the intervening stem of the shoot appeared normal. On closer examination of the stem, there were signs of necrosis at some points. Examination of a second shoot showed that the bark tissue was water-soaked throughout between the wilted tip and early symptoms at the base. It was impossible to ensure complete separation of bark, wood and pith tissue, but culturing showed that the pathogen was present in highest numbers in the bark tissue, less in the wood and least in the pith. It was concluded that, as the shoot grew, the pathogen was carried upward in the bark tissue, but not in high enough numbers to produce symptoms until a late stage. However, it could be argued that the pathogen was carried upward in the newly formed xylem vessels instead or as well. More research is needed on the upward movement of the pathogen in shoots as they grow.

Symptomless shoots

Keil & van der Zwet (1972) described studies where the pathogen was isolated from symptomless shoots on diseased trees and Ge & van der Zwet (1996) confirmed such observations. Eden-Green (1972) examined symptomless shoots following leaf inoculations where the pathogen had moved down the leaf-trace xylem vessels. Large numbers of bacteria were recovered from the wood but not from the bark regions. All attempts to reactivate such infections failed. This type of infection was associated with less susceptible cultivars. R. A. Lelliott (personal communication) and Billing (2008) recovered E. amylovora from symptomless tissue more than a year after inoculation. In the latter case, red streaks were present below the inoculation point (a leaf trace on an apple stem), suggesting xylem vessel invasion. The problem of symptomless infections, and their possible importance, is discussed by Vanneste & Eden-Green (2000). The evidence suggests that the pathogen survives well in xylem vessels but has difficulty in escaping from there into the cortical parenchyma. It is uncertain how long the pathogen may persist in mature symptomless bark tissue. If meristematic activity is resumed in time (Link & Wilcox, 1936), disease progression might resume.


The aim of this review was to discover why divergent views on bacterial invasion persist in fire blight studies and to present the evidence on which these views were based. The divergence is illustrated in a statement made in the abstract of a paper by Suhayda & Goodman (1981b): “The site of early proliferation and systemic migration is in mature xylem vessels rather than in the cortex and pith as reported in earlier literature.” The term ‘systemic’ applied only to inoculated shoots, not to whole-tree invasion. The view was based on a single experiment on shoots on young greenhouse-grown apple trees. The inoculation point was 5 cm below the shoot apex, which they assumed, but did not demonstrate, presented equal opportunity for infection to occur in the intercellular spaces of the cortical parenchyma and the xylem vessels. If this level was below L0, this assumption seems unlikely. Against the claim of systemic migration in xylem vessels must be set the field and experimental evidence of the many earlier observers. No one since 1981 has retested the xylem-vessel hypothesis, but evidence presented in this review shows that there is no doubt that E. amylovora can and does enter xylem vessels and multiply there, and may remain viable in mature vessels for long periods, possibly several growing seasons. Storm damage or pruning tools may allow escape of the pathogen to external surfaces, but other means of escape from mature xylem vessels into bark tissue, where typical symptoms are expressed in the growing tree, have so far eluded discovery. The site of early multiplication will be determined mainly by the method and site of inoculation. As shown in this review, in apple shoots, the site of inoculation in relation to the most recently unfolded leaf, L0, may be critical in determining whether or not xylem vessels or the cortical parenchyma or both are invaded.

Convincing evidence for migration in phloem sieve tubes is lacking. The observations in ringing experiments were equally consistent with cortical parenchyma migration. No one has produced further evidence for this possibility. However, migration associated with phloem fibres reported for peach canker (Feliciano & Daines, 1970) might be worth investigation. The weight of evidence, judged by orchard and experimental evidence, seems at present to be in favour of migration in bark tissue, not in xylem vessels. Because of strongly held views in favour of xylem vessel migration, more critical studies may be needed to clarify the problem. Failing this, it is important to make the best possible use of past research.

Some of the reservations described in this review concerning migration theories and the role of EPS have been expressed by others in earlier reviews. Schouten (1993) doubted that bacteria sucked into damaged xylem vessels necessarily played a role in subsequent pathogenesis, or that plugging of xylem vessels was responsible for early signs of wilt in host plants. Vanneste & Eden-Green (2000) considered that wilting of invaded young shoots was caused by collapse of parenchyma, not vessel plugging. When discussing the way in which the pathogen could migrate from the top of the tree to the rootstock, they stressed the problems associated with suitable experimental plant material and with inoculation techniques. Studies on mature trees are rarely feasible, so young plants need to be used. For inoculation, wounding is usually necessary and there is likely to be a link between the site of early multiplication and the port of entry of the pathogen. There is then the problem of detecting the pathogen in the invaded tissues, especially where numbers are low. Migration in xylem vessels seemed an attractive possibility and they speculated on possible escape routes. They admitted, however, that how, and under what conditions this might occur was still unknown.

Evidence presented in this report strongly suggests (but does not prove) that the main route for systemic migration of E. amylovora is in the intercellular spaces of the parenchymal bark tissue. The pathogen can also invade and multiply in mature xylem vessels for long distances down the tree. How far suction pressure is involved in that movement is not known. If this is to be an important migration route, the bacteria need a ready means of escape from the vessels into the bark tissue. How this might be achieved also remains unknown.

Note added in proof

A forthcoming book, Fire Blight. History, Biology and Control Management, T. van der Zwet, N. Orolaza-Halbrendt & W. Zeller, APS Press, will provide further opportunity for critical assessment of, and debate on, issues raised in this paper.


I thank those who helped me to gain access to early published reports. Also those who encouraged me to write this review and provided valuable critical comment.