First release of a fungal classical biocontrol agent against an invasive alien weed in Europe: biology of the rust, Puccinia komarovii var. glanduliferae
Abstract
The rust fungus Puccinia komarovii var. glanduliferae was first identified infecting Impatiens glandulifera in its native range (western Himalayas) between 2006 and 2010. Subsequently, it was imported into quarantine in the UK for evaluation as a classical biocontrol agent. To assess the safety of the rust, plant species relevant to Europe were tested for susceptibility. To confirm the life cycle, all infective spore stages were inoculated on I. glandulifera to follow disease progression. Teliospores were primed using bleaching and low temperatures to break dormancy. Temperature and dew period experiments using urediniospores were conducted to assess the parameters required for infection. Of the 74 plant species tested, only I. balsamina, an ornamental species, was fully susceptible to urediniospore inoculum. The life cycle of the rust – an autoecious, full-cycled species with five spore stages – was confirmed. Urediniospores were infective between 5 and 25°C, with an optimum at 15°C. A minimum of 8 h dew period was required to achieve consistent infection. Based on a pest risk assessment, the rust poses no threat to native biodiversity within EU Member States; making P. komarovii var. glanduliferae a suitable candidate as the first fungal classical biocontrol agent against an exotic weed in the region.
Introduction
Rust fungi have been utilized for classical biological control (CBC) of invasive alien plants since the first release of Puccinia chondrillina against skeleton weed, Chondrilla juncea, in Australia in 1971 (Hasan & Wapshere, 1973). Their restricted host range, often to single plant species – coupled with the high levels of host damage – makes rust fungi ideal candidates for consideration in exotic weed CBC programmes (Barton, 2012). The high mobility of rusts – being carried within and between plant populations by wind and convection currents – facilitates long distance dispersal, which is imperative for the control of invasive alien weed species that, typically, occupy vast areas (Morin et al., 2012).
The relatively recent exploitation of fungi for CBC of exotic weeds, compared to the long history of employing arthropods, which dates back to the end of the nineteenth century, is somewhat surprising. It has been suggested that irrational concerns over the safety of moving plant pathogens between countries (‘pathophobia’), in contrast to the long history of transcontinental movement of arthropod ‘counterpests’ (Elton, 1958), is one reason why the uptake of CBC using fungal pathogens has been slow (Evans et al., 2001). However, with the benefit of hindsight, fungal CBC agents, in particular rusts, have proved to be highly host specific and, when applied to control invasive weed populations covering large territories, successes have been spectacular (Tomley & Evans, 2004). Moreover, there have been no unpredicted non-target impacts recorded from the 28 fungal CBC agents released thus far against weed targets, globally (Barton, 2012).
As part of a CBC programme against the invasive neophyte Impatiens glandulifera (Ericales: Balsaminaceae), commonly known as Himalayan balsam, surveys were conducted throughout the native range of the plant (the foothills of the western Himalayas) between 2006 and 2010 (Tanner et al., 2015, 2015). Symptoms indicative of the rust fungus Puccinia komarovii, characterized by aecia on swollen stems and uredinia/telia on the leaves, were recorded on I. glandulifera in the Kaghan Valley (Pakistan) in 2006 (Tanner, 2008), with additional observations from the Kullu Valley and Solang Valley (Himachal Pradesh, India) between 2009 and 2010 (Tanner et al., 2015). This identification was confirmed subsequently, following traditional morphological examination of herbarium specimens (Tanner et al., 2015). However, P. komarovii is known to be an autoecious, host-specific rust, infecting only Impatiens parviflora, in both its native (Central Asia) and exotic ranges (mainland Europe) (Piskorz & Klimko, 2006), and, prior to these surveys, I. glandulifera had not been recorded as a host of this or any other rust species.
Following a molecular comparison of rust accessions from I. glandulifera (from India) and I. parviflora (from Hungary and China), where the nrDNA ITS and 28S (LSU) region were compared, the analysis indicated that separation of the two rust strains was warranted (Tanner et al., 2015). In addition, cross-inoculation studies demonstrated the presence of species-specific rust pathotypes, where I. glandulifera demonstrated immunity to inoculations using P. komarovii ex I. parviflora and vice versa (Tanner et al., 2015). Based on these molecular and cross-inoculation studies, the rust on I. glandulifera was separated at the varietal level and assigned the name Puccinia komarovii var. glanduliferae R.A. Tanner, C.A. Ellison, L. Kiss & H.C. Evans (Tanner et al., 2015).
Understanding the biology and, in particular, the life cycle of a CBC agent, prior to introduction and release, is an essential component in any weed biocontrol programme (Evans & Ellison, 2005), as also is demonstrating its specificity to the target species (Barton, 2012). The life cycle of the rust, P. komarovii var. komarovii, on I. parviflora has been described, in part, by Bacigálová et al. (1998), although the more cryptic spore stages (spermatia and basidiospores) have not been described in any detail. Furthermore, the ecology and biology of P. komarovii var. glanduliferae, especially the temperature and dew period requirements for spore germination and the host range, are essential considerations for assessing the persistence and safety of this rust to Europe. In this paper, results are presented of the research on the life cycle, ecology and host range of the rust P. komarovii var. glanduliferae, conducted both in the native range and under controlled quarantine conditions in the UK.
Materials and methods
Field surveys and collections
Between 2006 and 2010, surveys were undertaken to evaluate the natural enemies of I. glandulifera in its native range in the western Himalayas (Pakistan and India). Surveys were conducted throughout the growing season (June to September) – predominantly in subalpine meadows between 2500 and 3000 m a.s.l., where, in total, 30 geographically distinct populations of I. glandulifera were sampled. In Pakistan (2006–2009), only dried herbarium specimens were collected and, for logistical reasons, later surveys were concentrated in the Indian Himalayas from where the first introductions of the host plant into the UK originated in the 1830s (Herb K records). At each site, plants were selected at random along a W-shaped sampling transect through the whole population, for evidence of fungal pathogens. Rust-infected material was collected either on live seedlings, which were maintained in zip-lock bags with the roots wrapped in moist tissue, or as dried material preserved in a plant press. In 2010, the rust fungus was provisionally identified in the country of origin for export documentation, and imported into CABI's UK high-level quarantine facility under a Department of Environment, Food and Rural Affairs (DEFRA) plant health licence (no. PHL199C/6334) and the Indian Germplasm Exchange Scheme (Reference FS no. 11/2010).
Isolates of the rust were collected from three sites in different subalpine valleys in the Indian Himalayas and subsequently deposited in an internationally recognized fungal collection (Herb IMI): Rohtang, IMI 398718; Solang, IMI 398717; and Dhundi, IMI 502174.
In June 2010, the number of I. glandulifera plants infected with the rust was recorded at the Rohtang site following the same transect protocol as above. The transect length was 20 m and plants were selected at random along the sampling line. In total, 124 individual plants were examined for signs of infection. Plants infected with the aecial stem stage of the rust, readily identified by the stunted and swollen basal stem, were selected randomly from within the population and hypocotyl length was measured together with plant height. In total, 50 infected individuals were measured and compared to 50 randomly selected uninfected plants. Temperature data loggers (LogTag HAXO-8) were placed within infected populations at the Rohtang site for up to 7 days in June and August 2010, which recorded both temperature and humidity every 5 min.
Inoculation methodology
Inoculation experiments were carried out in controlled temperature (CT) rooms within a high-level quarantine facility at CABI's E-UK Centre. All of the inoculum used in the following experiments was derived from the rust accession IMI 398718 (ex Rohtang). This rust accession was used in the host range testing due to the higher level of damage observed on I. glandulifera plants at the associated site compared to Solang and Dhundi. Spores were harvested from stock plants using a sterile needle or by tapping infected leaves or stems over a Petri dish.
For the experiments evaluating the influence of temperature and dew period on infection level, a standardized concentration of spore/talc mix was applied to the plants. A 1:50 ratio of spores to talc were mixed in a 9 cm diameter Petri dish and placed in a cryovial (1·5 mL). By dipping a glass pipette into the mix, it was possible to obtain a replicable and precise volume of spore/talc mix (0·045 μg) that was applied to the abaxial leaf surface using a rubber teat (Ellison et al., 2006). In total, three aliquots were applied to each leaf, resulting in a concentration of c. 1·67 × 105 spores per leaf. The spore/talc mix was then evenly spread on to the lower leaf surface using a camel hair-brush (Humbro Senator, number 3). For each plant replicate, six to nine leaves of different ages, ranging from developing leaves in the meristem to mature fully expanded leaves, were inoculated.
For the life cycle and host range experiments, a non-standardized inoculum (i.e. to allow maximum coverage) was applied to the plants. Spores were mixed with talc at a ratio of c. 1:50 and applied evenly to each leaf using a camel hair-brush. In the host range experiments, both leaf surfaces were inoculated for non-target species.
Following inoculation, plants were placed in a dew chamber (Mercia Scientific) for 48 h. The dew chamber was set at 13°C (inner chamber) for aeciospore inoculations and 15°C for urediniospore inoculations. The difference in temperatures for each type reflected the different temperatures for June (aeciospores) and late July (urediniospores) recorded by the data loggers at the Rohtang site. Following the 48 h dew period, plants were removed and placed in a Perspex cage (1·0 × 0·5 × 0·7 m) in a controlled environment room at 21°C/15°C (12 h day/12 h night). Between each experiment, the dew chamber and cages were cleaned thoroughly using a disinfectant (Trigene; VWR) to avoid contamination from the previous test.
Rust development
Six I. glandulifera plants (12 weeks old) were evaluated microscopically following urediniospore inoculation to determine disease progression on and within the host. Plants were inoculated using the above method, and, following a 48 h dew period, inoculated leaves were removed from the plants every 24 h, stained, and observed microscopically for internal development. Leaves were stained using the clear-staining method of Bruzzese & Hasan (1983). External and internal development of the rust was first recorded photographically for I. glandulifera. This photographic reference was used as a standard to compare any symptoms observed on non-target plants.
Test plant selection, propagation and maintenance
The test plant list was compiled using the centrifugal phylogenetic method (Wapshere, 1974), and modified to include the work of Briese (2005) and the recent study on the phylogenetics of the genus Impatiens (Janssens et al., 2005). Thus, the initial selection involved closely related plant species from the genus Impatiens, which was then expanded to more distantly related species in other families within the same order (Ericales) as the target species. In addition, Wapshere (1974) advocated the inclusion of species with a similar morphology and biochemical composition to that of the target species. Therefore, species meeting these criteria were chosen, although often these species were already included due to their close relatedness to the target. Finally, a group of safeguard species, which by definition occur in similar ecological habitats to the target species, were included in the testing process.
Commercially available species were selected from the Plant Finder tool of the RHS (http://apps.rhs.org.uk/rhsplantfinder/). The nomenclature of species names followed that of Stace (2010), whilst Morgan (2007) was used specifically for Impatiens species. The full test plant list comprised 75 species, including the target (Table S1). Economically important Impatiens species that are widely cultivated in Europe, such as I. walleriana (five cultivars included) and I. hawkeri (four cultivars included), were represented by more than one cultivar. Impatiens noli-tangere, the only native Impatiens species in the UK and of high conservation importance (Hatcher, 2003), was represented by two distinct populations, one from Wales and the other from the English Lake District.
Seeds of the target species, I. glandulifera, were sourced from field collections in the UK in 2009 and subsequent years, and stored for a period of 6 months at 4°C before use (Mumford, 1990). Following storage, seeds were placed on sterile filter paper moistened with sterile distilled water (SDW) in a 9 cm diameter Petri dish, transferred to a 4°C incubator and observed every 5 days for signs of germination. Following germination, the seedlings were potted on in 10-cm diameter plant pots containing approximately 200 g John Innes compost. Plants were maintained in a growth room at 21°C, with natural lighting, for a period of between 6 and 12 weeks. After 6 weeks, the plants had attained a height of 20 cm and had formed four whorls of leaves and these plants were used for the aeciospore inoculations. After 12 weeks, the plants had attained a height of 40 cm and formed six whorls of leaves and these older plants were used for urediniospore inoculations and all the host range testing.
Non-target test species were either grown from seed sourced from various suppliers or bought in from plant suppliers. All plants were maintained in a growth room set at a minimum temperature of 21°C, with natural lighting.
Effect of temperature on urediniospore and aeciospore germination
Experiments were carried out to determine both the temperature range and the optimal temperature for germination of urediniospores and aeciospores. Germination was evaluated at 0·5, 1, 4, 12, 15, 19, 25, 29 and 35°C. Urediniospores or aeciospores were placed on three 5 cm diameter Petri dishes of tap water agar (TWA) and incubated (without light) at each of the above temperatures for 48 h. The temperature in each incubator was monitored every 10 min, using a temperature data logging system, to ensure consistency. After 48 h, the percentage of germinating spores per plate was calculated by microscopic examination of 300 spores. Spores were recorded as germinated when the germ tube was equal to or greater than the spore diameter (Bonde et al., 2007). The experiment was repeated for each temperature and spore type.
Dew period and temperature requirements of urediniospore infection
The effect of six dew periods (3, 5, 8, 12, 16 and 24 h) was investigated to determine the minimum requirement of free water on the leaf surface needed for infection. Twelve plants were inoculated with a known concentration of urediniospores and placed in a dew chamber at 15°C. For each plant, four leaves were inoculated and two plants were removed from the dew chamber at each of the dew periods. The experiment was repeated to give a total of four replicates per dew period. The density of pustules was calculated for each replicate after a 3-week period by placing a 1 cm2 lattice over the leaf and counting the number of pustules in five randomly selected squares.
Six temperatures (5, 10, 15, 20, 25 and 30°C) were evaluated for their effect on infection levels on I. glandulifera plants. Urediniospores were inoculated onto three 8-week-old plants using a known concentration of spore/talc mix and placed in the dew chamber for 24 h. The experiment was repeated to give a total of six replicates per temperature. Plants were maintained and pustule density was calculated using the method detailed above.
Life cycle elucidation
Aeciospores were inoculated onto six 6-week-old I. glandulifera plants following the standard protocol given in the inoculation methodology. Urediniospores, subsequently developing on these plants 14 days post-inoculation, were harvested and transferred to 12-week-old I. glandulifera plants. Leaf material containing teliospores – that developed on the inoculated plants usually 7–10 days after urediniospore expression – was dried in a plant press for 5 days and transferred to a freezer (−18°C) for long-term storage. When required, pustules containing teliospores were cut from infected leaves and surface sterilized using two bleach concentrations (1·4 or 14% sodium hypochlorite solution) for a period of 5 min (Evans, 1987). After three washes in SDW, the leaf pieces were placed on TWA with chloramphenicol and penicillin G (75 μg mL−1 each; Sigma) with the teliospores on the upper facing surface. The plates were sealed and placed in incubators at five different temperatures (4, 8, 10, 12 and 15°C). All plates were examined weekly for signs of germination using a compound microscope.
Cut leaf pieces with germinating teliospores were placed on 10-day-old seedlings where the leaf piece was positioned to be touching the developing hypocotyl. The seedlings were misted with sterile water sealed in a zip-lock bag and placed in a dark incubator at 4°C for a period of 48 h after which they were transferred to a Perspex cage within the CT room. Developing seedlings were monitored weekly for evidence of infection.
Host specificity testing
Urediniospores were inoculated onto three replicates of I. glandulifera, as the positive control, and three replicates of a non-target species using the standard protocol given in the inoculation methodology. Two or three non-target species were included in each test run, with the number of species for each run depending on plant size and available space in the quarantine dew chamber.
Target plants were maintained for a period of 21 days (until sporulation and harvesting of the urediniospores for use in the subsequent screening tests). Non-target species were maintained for a period of 42 days, three times longer than the expression of uredinia on I. glandulifera. For each non-target species, the experiment was repeated to increase the number of replicates per species, to six where available. Infection of I. glandulifera in each experiment served to confirm viability of the spores and that conditions for infection were suitable (Seier et al., 2009).
The level of susceptibility of the test plants was assessed both macroscopically and microscopically by assigning a score using a qualitative scoring system, based on Ellison et al. (2008) and Seier et al. (2009):
- 0 Immune: no symptoms
- 1 Resistant: chlorosis and/or necrosis (no further symptom development)
- 2 Weakly susceptible: delayed symptom expression, macroscopic symptoms first visible 20 days post inoculation, very sparse uredinia formation up to 48 days post-inoculation, (species not considered as a natural host)
- 3 Fully susceptible: macroscopic symptoms first visible 10 days post-inoculation, dense uredinia formation by day 15
Non-target plants that showed macroscopic symptoms on the leaves, but no spore formation, following urediniospore inoculation, were examined microscopically to evaluate the resistance response (Bruzzese & Hasan, 1983; Ellison et al., 2008). Non-target species that developed viable pustules during the host range testing procedure were tested further using both basidiospore and aeciospore inoculum.
Statistical analysis
Statistical analyses were conducted using R v. 2.12.2. (R Development Core Team, 2011). To determine the effect of temperature on urediniospore and aeciospore germination, an analysis of variance (anova), following arcsine transformation of the percentage data, was conducted. A t-test was conducted to compare the difference between the hypocotyl length and total plant height of infected and uninfected I. balsamina (quarantine experiment) and infected and uninfected I. glandulifera plants (field study).
Results
Rust phenology
Puccinia komarovii var. glanduliferae was observed at 10 sites throughout the native range during 2006 to 2010 (Table 1). Aecial stem infection, the first stage to express on young plants, was patchy at Rohtang, with only 12·9% of the population infected. However, within the same population the densities of infection varied and in suitably damp microclimates more than 50% of the plants were infected. Hypertrophy of the hypocotyls, induced by the rust, was evident. Infection often induced rupturing of the epidermis leaving the plants susceptible to secondary infection. Plants infected with the aecial stem rust were significantly taller (t = 2·509, df = 83, P < 0·05), with longer hypocotyls (t = 4·86, df = 62, P < 0·001) than uninfected plants (Fig. 1). However, as the season progressed, plants infected with the aecial stage soon lost their height advantage and healthy plants attained an aerial monopoly. In June the infection was observed at an early stage with fresh aecia still being produced on the swollen stem below the cotyledons. In early July chlorotic spotting on the abaxial leaf surface was visible with some mature uredinia present. By late July, and throughout August, dark chestnut brown coloured uredinia covered the abaxial surface, followed by dark brown telia in August and September.
| Country | Landmark | County/District | Habitat | Geographical coordinates | Altitude (m a.s.l.) |
|---|---|---|---|---|---|
| Pakistan | North Naran | Khagan Valley | Meadow | 34° 56′ 022″ N, 73° 48′ 439″ E | 2651 |
| Pakistan | North Naran | Khagan Valley | Meadow | 34° 54′ 436″ N, 73° 40′ 684″ E | 2829 |
| Pakistan | North Naran | Khagan Valley | Meadow | 34° 54′ 919″ N, 73° 47′ 885″ E | 2815 |
| Pakistan | North Naran | Khagan Valley | Meadow | 34° 56′ 022″ N, 73° 45′ 439″ E | 2651 |
| Pakistan | Besal | Khagan Valley | Meadow | 34° 56′ 598″ N, 73° 52′ 352″ E | 2935 |
| Pakistan | Besal | Khagan Valley | Meadow | 34° 56′ 380″ N, 73° 51′ 562″ E | 2938 |
| India | Solang | Solang Valley | Riparian | 32° 19′ 129″ N, 77° 09′ 359″ E | 2450 |
| India | Solang | Solang Valley | Riparian | 32° 19′ 132″ N, 77° 09′ 371″ E | 2459 |
| India | Rohtang | Kullu Valley | Meadow | 32° 19′ 778″ N, 77° 12′ 707″ E | 3067 |
| India | Dhundi | Kullu Valley | Woodland | 32° 21′ 215″ N, 77° 07′ 526″ E | 2837 |
Microscopic evaluation of urediniospore inoculation showed that the germination tube grew towards, and formed an appressorium over, the stoma 48 h post-inoculation (hpi; Fig. 2a). Intracellular development of the mycelium was seen 72 hpi (Fig. 2b), followed by the development of the uredinia on the abaxial leaf surface 240 hpi (Fig. 2c). The mature uredinia, releasing the urediniospores, were observed on the abaxial leaf surface 336 hpi (Fig. 2d).
Effect of temperature on urediniospore and aeciospore germination
Germination experiments on agar using aeciospores and urediniospores showed that both spore types have a wide temperature range at which they will germinate (1–25°C) with an optimal temperature of 15°C for both spore types (aeciospores: F6,35 = 62·49, P < 0·001, Fig. 3a; uredinio-spores: F8,45 = 432·3, P < 0·001, Fig. 3b). Uredinio-spores had a higher germination percentage (95·6%) than aeciospores (44·75%) at 15°C. However, percentage germination of aeciospores declined more quickly at temperatures above 15°C than for urediniospores. At 25°C urediniospore germination only decreased by 5% whereas aeciospore germination had decreased by 30%. There was no germination at 29 or 35°C for either spore type. It is interesting to note that even at the low temperatures of 1 and 4°C, the germination of urediniospores was relatively high, 33·9 and 73·1% respectively.
Effect of dew period and temperature on urediniospore infection
A minimum dew period of approximately 8 h was required to achieve urediniospore infection of I. glandulifera leaves. An average of 5·14 ± 1·515 pustules per cm2 was recorded after 21 days with an 8 h dew period, increasing to 30·08 ± 2·024 pustules per cm2 after 21 days with a dew period of 24 h (Fig. 4a). There was no infection following the 3 and 5 h dew periods. The minimum temperature required to achieve urediniospore infection of I. glandulifera leaves was 5°C, where an average of 4·85 ± 0·2 pustules per cm2 was recorded. Maximum infection was achieved at 15°C, with an average of 17·12 ± 1·631 pustules per cm2. Infection was found to sharply decline above 20°C; by 25°C infection had fallen to 0·2 ± 0·125 pustules per cm2 (Fig. 4b). There was no infection at 30°C.
It is interesting to note that urediniospores are capable of germination (85%) at 25°C on agar, but are not able to infect the plant at this temperature.
Life cycle
Inoculation experiments confirmed that P. komarovii var. glanduliferae is an autoecious, full-cycled (macrocyclic) rust with five distinct spore stages (aeciospores, urediniospores (Fig. 5a), teliospores (Fig. 5b), basidiospores and spermatia). Aeciospore inoculations on 6-week-old I. glandulifera plants resulted in chlorotic spotting on the underside of leaves 12 days post-inoculation (dpi). The development of uredinia within these lesions could be seen a further 4 days later, which was indicative of that seen in the native range (Fig. 5c). Subsequent inoculations using urediniospores on 12-week-old plants also produced chlorotic spotting on the underside of the leaves at 12 dpi and uredinia were expressed some 2–3 days later. As the plants matured, telia developed on the aging leaves, within the uredinia.
Germination of teliospores was observed only at 4°C for both bleach concentrations, with a minimum of 52 days and an average 72·6 ± 2·99 days required to break dormancy. There was no difference in teliospore germination between the two bleaching concentrations (t = 0·61, df = 5·837, P = 0·564) and no germination was observed at any other temperature after 100 days. Germination was recorded when an external metabasidium was observed to have formed with basidiospores developing on the tips of sterigmata (Fig. 5d).
Inoculations with basidiospores resulted firstly in a slight deformation of the stem followed by the formation of a white lesion on the hypocotyl 17 dpi. Spermogonia were observed 14 dpi (Fig. 5e). The lesion developed along the hypocotyl causing hypertrophy of the stem and distortion as the plant grew. Aecial cups became visible, erupting through the epidermis 4–7 days later. Figure 6 depicts the life cycle of P. komarovii var. glanduliferae, based on the field observations in the native range and the experiments conducted under quarantine conditions.
Host range
Of the 74 non-target plant species tested, only one species, I. balsamina, was consistently fully susceptible (score 3) to urediniospore inoculations. All cultivars of this species produced viable urediniospores at a comparable time to that of the positive control (I. glandulifera replicates). One replicate of I. scabrida was weakly susceptible (score 2) to the rust, whilst the remaining replicates showed only necrotic spotting (score 1). Five additional Impatiens species were resistant (score 1) where, following spore germination and penetration, internal hyphal development was halted due to internal cell necrosis (Table S1). The remaining 68 plant species were immune (score 0) to the rust (Table S1).
Impatiens balsamina and I. scabrida were tested further using both aeciospores and basidiospores. Both cultivars of I. balsamina were fully susceptible (score 3) to aeciospore inoculations producing viable urediniospores 14 days post-inoculation, similar in time and intensity to the positive controls. Impatiens balsamina cultivars were weakly susceptible (score 2) to basidiospore inoculations, although for I. balsamina ‘Topknot’ only one replicate was susceptible. For both cultivars, initial symptoms were expressed on the stem 17 dpi, comparable to that on I. glandulifera. However, aecial cups did not always form within these lesions and when they did they were observed 29–32 dpi, 8 days later and in far lower numbers than observed on I. glandulifera. When infection did occur, hypertrophy of the hypocotyl was not as pronounced on I. balsamina as that seen on I. glandulifera. The length of the hypocotyl (t = 0·1146, df = 13·83, P = 0·9104) and total height (t = −0·178, df = 13·90, P = 0·8613) of I. balsamina ‘Tom Thumb’ was similar when compared to uninoculated controls. Impatiens scabrida was immune to both aeciospore and basidiospore inoculations.
Discussion
Field observations and host range testing, backed up by molecular studies, have supported the identification of a new variety of rust fungus on I. glandulifera in the western Himalayas, P. komarovii var. glanduliferae (Tanner et al., 2015). As a classical biocontrol agent, P. komarovii var. glanduliferae has promise because it attacks its host twice during the growing season. In the spring, the rust infects young seedlings, often causing early mortality and hence no seed set. Later, in the summer, uredinia and telia densely cover the abaxial leaf surfaces reducing the area available for photosynthesis (Murray & Walters, 1992).
The potential of the aecial stage to impact significantly on its host is highlighted by the observations of Bacigálová et al. (1998) who recorded up to 100% mortality on seedlings of I. parviflora infected with P. komarovii var. komarovii in its exotic range in mainland Europe. What is highly relevant to, and extremely encouraging for, the present CBC programme is that P. komarovii var. komarovii is an example of ‘accidental’ CBC, whereby the rust has caught up with its host in an exotic situation, without deliberate human intervention, with dramatic consequences (Evans, 2000). The present field studies showed that infected I. glandulifera plants had an early season height advantage over uninfected plants, which may be an adaptation for releasing aeciospores above the vegetation. However, infected plants soon lost their early season height advantage, appearing significantly smaller and less healthy than the surrounding uninfected plants. The aecia often induced rupturing of the epidermis and secondary infection was clearly visible, leading to plant collapse.
Alien plants become invasive, at least in part, due to the absence of coevolved natural enemies– the enemy release hypothesis that underpins CBC. More recently, the absence of coevolved fungal endophytes has been considered a second factor in the success of alien plants; some of these endophytes act as bodyguards to protect the plant from natural enemies, but in return they use the host's resources. Thus, plants arriving in exotic situations without their coevolved natural enemies and fungal endophytes have increased levels of fitness compared to endemic species, but remain vulnerable to their coevolved natural enemies because of the absence of the endophytic bodyguards – the endophyte–enemy release hypothesis (Evans, 2008). A further variable can also be factored into the invasive-CBC equation: natural enemies of the plant's natural enemies. In the case of the rust on I. glandulifera, an unusual mycoparasite is associated with the aecial stage in the Himalayas. This is a species of Tuberculina, a close relative of the rusts, that is systemic in the plant and specifically parasitizes the haploid rust host, completely replacing the aecial cups (Fig. 5f) (Evans & Ellison, 2005). Studies show that these can be highly specific to a rust species or genus and exert control of their host (Lutz et al., 2004). Thus, in the Himalayas, P. komarovii var. glanduliferae is probably kept in check by Tuberculina and, therefore, fitness of the rust will increase in the absence of the parasite. Hence, selection of mycoparasite-free CBC agents is important. This would again suggest why the ‘accidental’ CBC aecial-rust, P. komarovii var. komarovii, arriving without its coevolved mycoparasites, has been so successful.
Urediniospores are spread through the population by wind and convection currents giving them high mobility (Morin et al., 2012). This spore stage may cycle through two or more generations until the autumn when the infection is expressed as teliospores (the overwintering stage), presumably linked with leaf age and lower temperatures that trigger the replacement of uredinia by telia. Often there is an overlap in spore stages: potentially an evolved adaptation to protect against stochastic events and thereby maximize the persistence of the pathogen. In the Himalayas, when the plants begin to senesce, or are killed by the first frosts, the telia, attached to the leaves, become incorporated into the soil. A rise in spring temperature induces the germination of both I. glandulifera seeds and dormant teliospores, which form basidiospores that infect the emerging hypocotyl of the germinating plants.
Piskorz & Klimko (2006) recorded a high level of infection as a result of P. komarovii var. komarovii on I. parviflora in Wielkopolski National Park in Poland, where over 90% of the population was infected. In the UK, I. glandulifera forms dense linear monocultures along river catchments, out-competing native plant (Hulme & Bremner, 2006) and invertebrate species (Tanner et al., 2013). Releasing the rust into similar habitats would be considered best practice from an establishment perspective, giving maximum opportunity for spread throughout the population.
In Europe, P. komarovii var. komarovii is found infecting I. parviflora predominantly in wooded habitats, although often these border riparian catchments (R. A. Tanner, personal observations). Thus, it is anticipated that the stable wooded habitats and riverbanks above the flood plain, with populations of I. glandulifera, would form the overwintering environment for the teliospores. Aeciospores and urediniospores would then spread to populations developing in less stable habitats on riverbanks close to the water.
Urediniospores had an optimum temperature for infection near to 15°C, which is in line with those temperatures experienced at night during the summer months in the UK. Significantly, the rust had a broad temperature tolerance, being able to infect, albeit at a low rate, at 5 and 10°C. This will facilitate infection over a prolonged growing season. Likewise, only an 8 h dew period was required to achieve urediniospore infection, which is compatible with expected dew periods in the invasive range.
The results from the dew period experiment, where infection levels are shown to increase from 8 to 24 h, suggest that spores were continuing to germinate and infect over the 24 h of the experiment, because infection had not yet plateaued. This indicates that spore germination is staggered over time, rather than all viable spores germinating simultaneously once the conditions are conducive. Indeed, in vitro germination experiments provide evidence that, within a sample of fresh urediniospores, some will be dormant. This is probably a survival strategy, so that if conditions do not remain favourable for sufficient time during the first dew period encountered by the spores, some may survive until the next dew cycle.
The establishment, persistence and rigid host specificity of P. komarovii var. komarovii in mainland Europe (Bacigálová et al., 1998; Piskorz & Klimko, 2006) provides evidence that P. komarovii var. glanduliferae should establish successfully and be equally damaging and safe in the UK. Tanner et al. (2015) confirmed that both P. komarovii var. komarovii and P. komarovii var. glanduliferae could infect cultivars of I. balsamina. However, the authors are unaware of any records in mainland Europe where I. balsamina is recorded as a host of P. komarovii var. glanduliferae, despite their geographic overlap. Indeed, I. balsamina has now become an invasive alien weed in northern India, where there are at least 50 native Impatiens species (Khuroo et al., 2012), suggesting that it is highly competitive and not attacked by the rust on I. glandulifera (Evans, 2013).
Although non-target infection of I. balsamina is undesirable, the results must be taken in context of the value of the species to the UK. Whereas some Impatiens species, namely I. walleriana and I. hawkeri, have a high economic value to the horticultural industry, that of I. balsamina is considerably lower, with sales figures in 2012 from one of the UK's largest seed suppliers amounting to £67·20 and £64·60 annually for I. balsamina ‘Topknot’ and I. balsamina ‘Tom Thumb’, respectively (Chiltern Seeds). Indeed, this species is considered to be the ‘prima donna’ of Impatiens species, being susceptible to many pests and diseases, and hence unlikely to become a popular horticultural species in the future (M. A. Spencer, Natural History Museum, London, UK, personal communication).
The results from the host range testing confirm, with the exception of I. balsamina, that P. komarovii var. glanduliferae is highly host specific to I. glandulifera. The variable non-target symptoms on the non-native species, I. scabrida, where one replicate was weakly susceptible to rust inoculation whereas the remaining replicates were immune, indicates that these symptoms may be due to an artificially high inoculum load in the experiments, a common occurrence in this type of experiment (Barton, 2012). For this species, subsequent basidiospore inoculations demonstrated that I. scabrida is immune to infection and, therefore, not a natural host of P. komarovii var. glanduliferae.
In 2014, a pest risk assessment (PRA) detailing the research conducted over the course of the CBC programme was accepted by UK regulators (FERA) and the EU Standing Committee on Plant Health. After a 6-week public consultation, and following approval by DEFRA Ministers, FERA approved the release of the rust from quarantine restrictions and it has since been released in experimental trials in England in August 2014. In conclusion, as a management tool to control I. glandulifera on both a catchment and a national scale, CBC is the only feasible long-term solution. Over the 8 years of research on the life cycle and host range of P. komarovii var. glanduliferae, the rust has been shown to be specific to and, potentially, highly damaging on its host I. glandulifera.
Acknowledgements
The authors thank the Department for Environment, Food and Rural Affairs (Defra), the Environment Agency (EA), UK, the Scottish Executive, West Country Rivers Trust and Network Rail for funding this research. Mool Chand Singh, Usha Dev and Jyoti Bhardwaj (National Bureau of Plant Genetic Resources, New Delhi, India) are also thanked for assisting in the field collections in the Indian Himalayas. The authors acknowledge the expert review of the test plant list conducted by Dr Mark Spencer (Natural History Museum, London, UK) and by Drs Melanie Tuffen and Helen Anderson (Food and Environment Research Agency, FERA, York, UK). The work of CABI colleagues in the Indian and Pakistan offices for assisting with surveys in their countries and for facilitating export of biological material is also acknowledged along with Suzy Wood for drawing and designing the life cycle diagram. Finally, the authors wish to thank the landowners and the Governments of India and Pakistan for allowing access to the areas surveyed.
