Screening Corymbia populations for resistance to Puccinia psidii



The exotic rust pathogen Puccinia psidii is now widespread along the east coast of Australia from temperate Victoria to tropical far north Queensland, with a current host range exceeding 200 species from 37 myrtaceous genera. To determine the threat P. psidii poses to plantation and native eucalypts, artificial inoculation was used to screen germplasm of spotted gum (Corymbia spp.) for resistance to the biotype of P. psidii that has become established in Australia. The objective was to characterize resistance to P. psidii within the Corymbia species complex so that management strategies for the deployment of germplasm from existing breeding programmes of these spotted gum species could be developed. Symptom development initiated 7 days after inoculation, with resistant and susceptible seedlings identified within all species, provenances and families. Inter- and intraspecific variability in rust resistance was observed among spotted gum species. There was no apparent relationship between climatic conditions at the provenance origin and disease resistance. The heritability estimates for all assessments are moderate to high and indicate a significant level of additive genetic variance for rust resistance within the populations. The results of this study clearly identify potential to select for resistance at the family level within the tested populations. While the potential for P. psidii to detrimentally impact upon Corymbia in the nursery and in young plantations was demonstrated, estimations of the heritability of resistance suggest that efforts to enhance this trait through breeding have reasonable prospects for success.


Puccinia psidii was first described from Psidium pomiferum in Brazil by Winter in 1884 (Coutinho et al., 1998). It has since been reported from a range of hosts in South and Central America and in Florida and California in the United States (Coutinho et al., 1998; Zambino & Nolan, 2011). More recently, P. psidii has been reported for the first time outside of the Americas, with detections in Hawaii in 2005 (Uchida et al., 2006) and Japan in 2009 (Kawanishi et al., 2009). Zhuang & Wei (2011) published a new report of P. psidii infecting Syzygium jambos on Hainan Island in southern China.

Outside of Australia, P. psidii has been reported on more than 100 species from 29 genera in the Myrtaceae (Coutinho et al., 1998; Simpson et al., 2006; Carnegie & Lidbetter, 2011). Puccinia psidii has had a significant impact on industries relying on Myrtaceae, including the all spice (Pimenta dioica) industry in Jamaica (MacLachlan, 1938) and the eucalypt plantation industry in Brazil (Ferreira, 1983; Furtado & Marino, 2003; Glen et al., 2007).

Eucalyptus rust, originally named guava rust, has long been recognized as a significant threat to Australia's Myrtaceae-rich natural forests and industries that rely on species in this family (Glen et al., 2007). In Brazil, the impact of P. psidii on introduced eucalypts was so significant, including reduced growth, destruction of growing shoots leading to stem malformation and tree mortality, that the term ‘eucalyptus rust’ was used to describe the disease (Ferreira, 1983). Infection can affect growth rates and subsequent profitability (Booth et al., 2000).

Puccinia psidii was first identified in Australia in April 2010 when it was detected in a cut flower nursery in New South Wales (NSW; Carnegie et al., 2010). The fungus was given the common name ‘myrtle rust’, while taxonomic efforts attempted to distinguish it from the well-known guava and eucalyptus rusts endemic to South America but hitherto absent from Australia (Department of Agriculture, Fisheries & Forestry, 2010). By December 2010, P. psidii had spread throughout NSW and was detected in Queensland for the first time (Carnegie et al., 2010). By September 2012, P. psidii had spread along the east coast of Australia from temperate areas in Victoria to tropical regions of far north Queensland. Based on observations of the natural infection within the host range, the rust has been found on more than 200 species from 35 genera (Carnegie & Lidbetter, 2011).

Several Eucalyptus and Corymbia (spotted gum) spp. have been naturally infected by P. psidii, with the potential impacts on native forests of great concern. In addition, commercially important species such as E.agglomerata, E. pilularis, E. cloeziana, E. grandis, C. citriodora subsp. citriodora, C. citriodora subsp. variegata, C. henryi and C. torelliana have been infected and indicate the pathogen could impact upon the Australian plantation hardwood industry as well as ecologically focused reforestation efforts (Carnegie & Lidbetter, 2011).

Elite Eucalyptus clones are extensively used to achieve improvements in pulpwood quality, rapid growth and high yield (Junghans et al., 2003a). However, one of the most commonly used Eucalyptus species, E. grandis, is highly susceptible to P. psidii (Junghans et al., 2003a), a disease still considered one of the most damaging to eucalypt plantations in Brazil. While management of rust in nurseries may be achieved with chemical treatment, management of rust in plantations across South America relies on developing resistant cultivars (Furtado & Marino, 2003). Junghans et al. (2003a) identified segregation ratios in families suggesting major gene control of rust resistance. This work identified the Ppr-1 genetic marker as being useful for marker-assisted introgression of resistance and has provided an initial lead for positional cloning efforts to identify the causal resistance allele. However, Graça et al. (2011) recently found clones carrying the resistance gene Ppr-1 to be susceptible to a newly discovered rust biotype and suggested multiple genes may govern resistance.

Wide variation in resistance to eucalyptus rust has been observed between and within Eucalyptus and Corymbia spp. tested with a single rust biotype in Brazil as part of a study to understand the risk the disease posed to Australia (Zauza et al., 2010). Inter- and intraspecific variability was also observed among and within a range of Myrtaceae species tested. Of the species tested, accessions of E. grandis and E. cloeziana were ranked as most susceptible, while spotted gum species C. citriodora subsp. citriodora and C. maculata had >72% of seedlings rated as resistant (Zauza et al., 2010). This confirmed earlier glasshouse and field-based assessments (Dianese et al., 1984) that identified E. grandis and E. cloeziana as being more susceptible to rust infection.

Preliminary host testing of the rust in Australia has revealed that a range of species, some of which are relied upon by the plantation forest industry (E. agglomerata, E. cloeziana, E. dunnii, E. globulus, E. grandis, E. pellita, E. pilularis, E. saligna and C. citriodora), are susceptible to P. psidii (Carnegie & Lidbetter, 2011; Morin et al., 2012). So far the impact of P. psidii on Eucalyptus and Corymbia spp. in Australia, based on field observations alone, is restricted to seedlings (e.g. C. citriodora subsp. variegata, C. torelliana, E. cloeziana, E. tereticornis). However, on E. carnea and E. curtisii the disease has been observed, causing dieback of mature trees and young coppice regrowth (authors' unpublished data). To date, P. psidii has not been detected within forest tree plantations in Queensland. This may be because the disease has not yet established in areas where the majority of Queensland plantations exist.

This study aimed to understand the threat P. psidii poses to plantation and native forests, through screening a diverse range of Eucalyptus and Corymbia germplasm for resistance to the biotype of P. psidii (Carnegie et al., 2010) now widespread in Australia. A second aim was to characterize resistance to P. psidii within the Corymbia species complex, providing guidance for developing management strategies for the deployment of germplasm from existing industrial breeding programmes. Three different disease scoring methods were also compared. Spotted gum taxa tested in this study included three closely related, valuable native forest and plantation species, C. citriodora subsp. variegata (CCV), C. citriodora subsp. citriodora (CCC) and C. henryi (CH) (Ochieng et al., 2010). In addition, resistance within seed lots of C. torelliana (CT) and associated Corymbia hybrids (Lee et al., 2009) were studied. In total, nine spotted gum provenances were evaluated, with pedigreed families representing six of these provenances. This allowed the investigation of patterns of resistance within and among species of the widely distributed and environmentally significant Corymbia genus.

Material and methods

Species, provenance and family resistance to Puccinia psidii

To examine the influence of origin on rust resistance patterns, 15 seedlings from between nine and 11 open-pollinated families from a range of provenances were examined for disease levels following inoculation with P. psidii (Table 1). Two provenances from CCC were compared to five CCV and two CH provenances. Of the CCC provenances, one was from an inland location in far north Queensland (Mt Garnet) and a second was from a coastal location at the southern end of the range of CCC (Yeppoon) in Queensland. Presho, a CCV provenance frequently severely damaged by a native fungal foliage pathogen (Quambalaria pitereka), originates in the westernmost range of the species. The resistance of Presho to P. psidii was compared to the more coastal provenances of Woondum, Brooyar, Mt McEuan and another inland provenance, Ballon.

Table 1. Corymbia species, provenances, seed lots, hybrids and clones used in studies to characterize resistance levels to myrtle rust, Puccinia psidii
TaxonProvenance or sourceSeed lots or clonesLatitude (S)Longitude (E)Altitude (m a.s.l.)Mean annual rainfall (mm)
  1. a

    Details of the provenances with family structure are provided in Table 4.

Corymbia citriodora subsp. citriodora (CCC)Mt Garnet10101 (10 families)a18°00′145°11′700832
Yeppoon11246 (9 families)23°06′150°44′301325
Corymbia citriodora subsp. variegata (CCV)Woondum11185 (11 families)26°25′152°81′4001536
Brooyar10248 (10 families)26°10′152°30′901143
Mt McEuan12805 (10 families)26°14′151°39′300726
Ballon5584 (bulk of 11 trees)26°30′150°55′350656
Presho4928 (bulk of 50 trees)25°11′149°10′470675
Seed orchard seedX7370–X7379 (10 families)
Corymbia henryi (CH)Lockyer10250 (10 families)27°28′152°17′150840
Nerang10257 (10 families)27°59′153°19′1001439
Seed orchard seedX6088–X6097 (10 families)
Corymbia torelliana (CT)LandraceX15, X34, X164, X168, X169 (5 families)
Corymbia hybrids: (CT×CCC, CT×CCV, CT×CH)Control crossesX135, X137, X159, X161, X194, X196, X136 (7 families)
Corymbia hybrid clones: (CT×CCV)1, 6, 8, 11, 13, 18, 22, 39 (8 clones)

For CH, two provenances were selected from different rainfall zones: Lockyer and Nerang. Resistance levels were also examined at a family level for seven provenances (CCC – Mt Garnet, Yeppoon; CCV – Brooyar, Mt McEuan, Woondum; CH – Lockyer, Nerang) with a family structure. Ballon and Presho provenances were excluded from family level comparison as the seedlings originated from a bulked seed lot. Also included were families from CCV, CH and CT seed orchards. To examine repeatability of results at a family level, a second inoculation was done on seedlings from three CCV provenances, Brooyar, Mt McEuan and Woondum and results compared to the first inoculation.

In addition, seedlings from seven full-sib controlled cross Corymbia hybrids (details provided in Table 1) and eight commercial Corymbia clones selected from within control pollinated hybrid families of CT mother trees pollinated with CCV pollen, were included in the study. The clones had had a restricted genetic base, originating from three full-sib families with two to three clones being from each sib family. These clones had been commercialized based on their superior early growth, form and amenability to vegetative propagation as rooted cuttings relative to CCV. The clones were similar in size to the seedlings but their age was unknown, as they were supplied by a commercial nursery and grown with the seedlings for a 4 week period prior to screening the trial.

Experimental design

The trial with 112 treatments (seed lots) was established as a randomized incomplete block design with 15 replicates, of a single tree from each family treatment in each replicate. Bulk seed lots were represented by 10 allocations in the design (e.g. 10 × 15 replicates = 150 seedlings) to ensure that realistic provenance means for the assessments were obtained. In this sort of design, the number of treatments per block (a subset of the replicate, used to account for variation within the replicate) is less than the number of treatments being evaluated. This type of trial design increases precision of the analysis and reduces experimental error. Incomplete block design trials can be thought of as two-factor experiments (block and treatment) with unequal numbers and no interaction between the block and treatment factors.


Seeds were sown in 40-celled hyco trays (70 mL) in the glasshouse using a potting medium consisting of 50% pine bark fines (0–10 mm), 25% pine bark peat, 25% coarse perlite, a mix of 12–14 month slow release Osmocote (N 17·9: P 0·8: K 7·3) fertilizer at a rate of 4 kg m−3, gypsum (1 kg m−3), Micromax (1 kg m−3) and a granular wetting agent Hydroflo2 (1 kg m−3). Plants were irrigated twice a day for 10 min at each irrigation time using an overhead spray. Cuttings of the Corymbia hybrid clones tested were obtained from Clonal Solutions Australia Pty Ltd. These clones were acclimated with seedlings of the other Corymbia spp. for 4 weeks prior to inoculation.


A single pustule isolate of P. psidii (BRIP# 57793) was collected from Rhodamnia sessiliflora growing in the Chapel Hill suburb of Brisbane, Queensland, Australia. Urediniospores from a single pustule were collected using a fine-bristled paintbrush and washed into 5 mL sterile distilled water (SDW) to which one drop of the surfactant Tween 20 was added. This suspension was then applied onto Syzygium jambos and Rhodamnia rubescens seedlings to further propagate spores for screening.

These plants were then covered with black plastic bags, which were sealed to maintain a high humidity level, and placed into an incubator at 20°C for 24 h. After 24 h plants were grown in a controlled environment room (CER) using a 12 h/26°C light and 12 h/20°C dark photoperiod. Urediniospores were collected using a vacuum line 10–12 days after inoculation, continuing until spores were no longer produced. These urediniospores were then inoculated back onto S. jambos and R. rubescens seedlings, a process repeated until sufficient numbers of spores had been collected for inoculation. Urediniospores were then placed into a desiccator for 48 h before being placed into Nunc tubes and stored at −80°C until required.


Urediniospores were removed from −80°C storage and allowed to warm to room temperature prior to being added to SDW. The surfactant Tween 20 was added at a rate of two drops per 100 mL SDW and the spore suspension stirred to reduce clumping. Spore counts were then conducted using a haemocytometer and the suspension adjusted to a concentration of 1 × 105 spores mL−1 for use in subsequent inoculations.

Corymbia seedlings were inoculated using a fine mist spray (2·9 kPa pressure), generated by a compressor driven spray gun (Iwata Studio series 1/6 hp; Gravity spray gun RG3), applied to the upper and lower leaf surfaces of the seedlings, ensuring all leaves were coated with a fine mist but run-off of the spore suspension was avoided. Ten R. rubescens seedlings were used as susceptible controls and placed haphazardly within replicates of the seedlings tested.

Seedling trays were placed into solid plastic tubs and then onto a metal bench layered with a plastic sheet. Immediately after inoculation, seedlings were covered with a plastic sheet for 24 h to maintain high humidity levels and leaf wetness in a controlled environment room set between 18 and 20°C in the dark. Hot tap water (60°C) was applied to the lower plastic sheet immediately after inoculation to ensure high humidity levels were achieved rapidly. After 24 h, plastic bags were removed and plants grown in a shade-house and hand watered as required. Disease symptom progression was monitored daily.

Disease assessments

Seedlings were assessed 12 days after inoculation for incidence of disease (% of seedlings with symptoms) and severity of infection on new shoots and expanding leaves using a disease rating scale: 1 = no symptoms evident or presence of yellow flecking; 2 = presence of a hypersensitive reaction (HR) with fleck or necrosis; 3 = small pustules, <0·8 mm diameter, with one or two uredinia; 4 =  medium-sized pustules, 0·8–1·6 mm diameter with about 12 uredinia; 5 = large pustules, >1·6 mm diameter, with 20 or more uredinia on leaves, petioles and/or shoots (Junghans et al., 2003b; Fig. 1). Ratings 1–3 are considered as indicating resistance.

Figure 1.

Rating scale for assessment of resistance to Puccinia psidii on the spotted gums (Corymbia taxa) with five classes of severity based on ratings developed by Junghans et al. (2003b). From left to right: 1 = hypersensitive reaction with yellow flecking; 2 = hypersensitive reaction with necrotic lesions and no evidence of pustules with uredinia; 3 = small pustules <0·8 mm in diameter with 1–2 uredinia; 4 = medium pustule size 0·8–1·6 mm diameter with up to 12 uredinia; 5 = pustule size >1·6 mm in diameter with 20 or more uredinia on leaves, petioles and/or shoots.

Disease incidence (I) was also assessed as a percentage of the four youngest inoculated leaves on seedlings showing evidence of pustule development and uredinia. Disease severity (S) was scored as a subjective assessment of the percentage of the total area of infected foliage on diseased leaves only. In total, four assessments were available for analysis as response variables: disease incidence (I) and disease severity (S), disease rating scale (1–5) and the percentage of resistant seedlings based on the disease rating scale.

Statistical analyses

Two sets of analyses were used to interrogate the data collected from the assessment of inoculated seedlings. Firstly, a broad examination of all species, taxa and type of plant (clone versus family) assessed was undertaken including CCC, CCV, CH, CT and Corymbia hybrids following the approach of Lee et al. (2011). Comparisons were then made among CCC, CCV and CH populations within the species and families within provenances of the species. These analyses excluded data from CT and its hybrids and were primarily undertaken to indicate the significance of difference between means so that comparisons could be made among experimental units, and relationships examined between resistance ratings and environmental variables at the provenance origin (e.g. mean annual rainfall), and population performance could be investigated. All proportion data were arcsine square root transformed prior to analysis using anova and compared using Fisher's PLD post hoc test (statview). Reverse-transformed data are presented. Chi-square analysis was used to compare non-parametric disease rating data and seedling resistance frequencies between provenances within species and families within provenances.

Genetic parameters of rust resistance

Genetic analyses were undertaken to describe patterns of variation in disease resistance in the Corymbia populations with variance components estimated for provenances and families within provenances, to allow estimation of genetic parameters. Genetic parameters were estimated across and within species to provide an indication of the heritability of rust resistance as well as an estimate of the relative importance of within or among population variation for rust resistance (Brawner et al., 2011). Comparisons between the three types of assessment data (incidence, severity, disease rating) were made for each of these analyses.

Genetic parameters were approximated using estimates of the causal components of variance from a mixed model fit with ASreml (Gilmour et al., 2009), assuming the provenances sampled for this experiment were randomly selected races within each Corymbia taxon. Two separate analyses were undertaken; the first used a multivariate analysis including CCV, CCC and CH while the second set of univariate analyses considered each taxon separately. The multivariate analysis included all taxa, as CCV and CCC are taxonomically similar subspecies and CH could not be distinguished from CCV using molecular genetic approaches (Ochieng et al., 2008). Therefore, data from these populations were analysed jointly to estimate genetic parameters for the greater Corymbia complex. The statistical model used was equivalent to equation 3 detailed in Brawner et al. (2011). Narrow-sense heritability estimates were approximated as ratio of additive genetic variance to the within provenance phenotypic variance. A similar statistic, proportion of provenance variance, was estimated to evaluate the relative importance of among provenance variance. In addition, genetic correlations were estimated at the family and provenance level to understand the relationships between assessment traits.


Symptom development

Symptoms were first observed 5 days after inoculation, presenting as blistering on leaves and shoots of R. rubescens and red-brown coloured lesions, necrotic and chlorotic spots on the Corymbia species (Fig. 2). Pustules with uredinia were first observed 6 and 7 days after inoculation on R. rubescens and Corymbia spp. seedlings, respectively (Fig. 2).

Figure 2.

Symptoms first observed 5 days after inoculation appearing as blisters on Rhodamnia rubescens (a), and red-brown lesions (b) or yellow-coloured flecking (c) on spotted gum (Corymbia taxa). Pustules with yellow-coloured uredinia were first observed 6 days after inoculation on R. rubescens (d) and 7 days on spotted gum (e,f).

Disease assessment

With one exception, all Corymbia species evaluated in this screening experiment exhibited disease symptoms and incidence and severity scores greater than zero; the Corymbia hybrid clones were resistant to infection by rust with only chlorotic or necrotic spots observed on new shoots and expanding leaves. Resistant and susceptible seedlings were identified within all species, provenances and families.

Resistance of Corymbia species and subspecies

CCC was the most susceptible of the taxa but not significantly different from CH, CT and CCV based on disease rating scores (Table 2). Corymbia hybrids were more resistant to P. psidii infection than CCC, CH, CCV and C. torelliana when comparing disease incidence, severity and disease rating scores.

Table 2. Estimates of performance with ± standard errors by Corymbia taxa for the three traits assessed in the rust resistance screening trial, listed from least to most susceptible
TaxaIncidenceaSeverityaDisease ratingb
  1. a

    Estimates for incidence and severity data followed by different letters significantly (< 0·05) differ from one another according to Tukey's multiple comparison tests.

  2. b

    Chi-square analysis was used to compare disease ratings between species (pairwise comparison < 0·05).

C. hybrid16·9 ± 5·0 a3·0 ± 1·4 a2·1 ± 0·2 a
C. torelliana 34·4 ± 7·9 b7·4 ± 2·3 ab2·8 ± 0·3 b
C. henryi 44·4 ± 3·5 bc10·1 ± 1·0 b2·9 ± 0·1 b
CCV50·2 ± 3·9 cd9·6 ± 1·1 b3·0 ± 0·1 b
CCC57·1 ± 4·1 d12·9 ± 1·2 c3·2 ± 0·1 b

Resistance of provenances of CCC, CCV and CH

All provenances studied, based on means of families and bulk seed lots, showed symptoms of infection by P. psidii (Table 3). Of the nine provenances tested, Mt McEuan (CCV) had the highest percentage of resistant seedlings (Table 3) followed by Lockyer (CH) and Presho (CCV). Mt Garnet (CCC) and Ballon (CCV) had the lowest percentage of resistant seedlings.

Table 3. Spotted gum (Corymbia spp.) provenance susceptibility in order of least to most susceptible, based on percentage of resistant seedlings calculated from the disease severity ratings
SpeciesProvenanceDisease incidence (% leaves infected)aNumber of plants/severity class (1:2:3:4:5)Disease ratingResistant seedlings (%)a
  1. a

    Percentage followed by same letters shows means that do not differ significantly (disease incidence Fisher's PLSD test < 0·05; resistant seedling pairwise comparison < 0·05).

CCVMt McEuan39·2 ± 3·7 a20:59:36:14:202·7 ± 0·177 a
CHLockyer40·7 ± 3·6 a35:40:27:21:262·7 ± 0·1269 b
CCVPresho51·0 ± 3·6 ab4:52:44:33:163·03 ± 0·167 b
CCVBrooyar53·4 ± 3·6 bc14:37:47:27:233·05 ± 0·166 b
CCVWoondum52·6 ± 3·4 bc11:46:47:36:233·1 ± 0·164 b
CCCYeppoon62·4 ± 3·6 cd7:37:46:38:203·2 ± 0·161 b
CHNerang47·9 ± 3·6 ab40:22:26:31:302·9 ± 0·1259 bc
CCCMt Garnet51·6 ± 3·6 b15:40:31:35:273·1 ± 0·158 bc
CCVBallon65·8 ± 3·4 d3:32:38:39:363·5 ± 0·149 c

Resistance of families within CCC, CCV and CH provenances

Resistance to P. psidii varied among and within families in all provenances (Table 4). No family showed total resistance or total susceptibility to P. psidii. However, in each provenance tested there were significant differences between the most resistant family and the most susceptible family when comparing the percentage of resistant seedlings (Table 4).

Table 4. Susceptibility of spotted gum (Corymbia spp.) families to Puccinia psidii ranked least to most susceptible
SpeciesProvenanceFamilyNumber of plants/severity class (1:2:3:4:5)Disease ratingResistant seedlings (%)a
  1. a

    Resistant seedling percentage followed by same letters shows means that do not differ significantly (pairwise comparison < 0·05).

CCCMt Garnet10101-115:6:3:0:12·1 ± 0·393 a
10101-12:7:2:3:12·6 ± 0·373 ab
10101-25:2:3:2:32·7 ± 0·467 abc
10101-101:3:5:5:13·1 ± 0·360 bc
10101-30:5:4:6:03·1 ± 0·260 bc
10101-61:4:4:3:33·2 ± 0·360 bc
10101-80:7:1:2:53·3 ± 0·453 bcd
10101-40:3:5:4:33·5 ± 0·353 bcd
10101-51:1:3:5:43·7 ± 0·336 cd
10101-90:2:1:5:64·1 ± 0·321 d
CCCYeppoon11246-81:7:5:2:02·5 ± 0·287 a
11246-30:5:7:3:02·9 ± 0·280 a
11246-61:5:4:5:02·9 ± 0·367 ab
11246-20:5:3:5:13·1 ± 0·357 ab
11246-51:1:6:5:13·3 ± 0·357 ab
11246-70:3:5:4:33·5 ± 0·353 ab
11246-11:2:3:7:23·5 ± 0·340 b
11246-100:2:4:5:43·7 ± 0·340 b
11246-90:0:5:1:94·3 ± 0·233 b
CCVBrooyar10248-81:6:7:0:12·6 ± 0·293 a
10248-44:5:4:1:12·3 ± 0·387 ab
10248-53:7:1:2:22·5 ± 0·373 ab
10248-62:4:5:3:12·8 ± 0·373 ab
10248-71:2:8:4:03·0 ± 0·273 ab
10248-21:5:3:2:33·1 ± 0·371 abc
10248-90:4:6:3:23·2 ± 0·367 abcd
10248-101:1:7:4:23·3 ± 0·360 bcd
10248-31:1:3:3:63·9 ± 0·336 cd
10248-10:2:3:5:53·9 ± 0·333 d
CCVMt McEuan12805-236:7:1:1:01·8 ± 0·293 a
12805-273:9:1:2:02·1 ± 0·287 ab
12805-183:5:5:1:12·5 ± 0·387 ab
12805-262:6:5:1:12·5 ± 0·387 ab
12805-220:6:6:2:12·9 ± 0·280 ab
12805-190:6:4:2:23·0 ± 0·379 ab
12805-241:7:3:2:22·8 ± 0·373 ab
12805-252:3:5:2:33·1 ± 0·373 ab
12805-212:6:2:0:53·0 ± 0·467 ab
12805-201:4:4:1:53·3 ± 0·460 b
CCVWoondum11185-1123:6:5:0:12·3 ± 0·393 a
11185-1111:10:1:2:12·5 ± 0·380 ab
11185-1152:4:5:2:22·9 ± 0·380 ab
11185-1140:4:7:4:03·0 ± 0·273 ab
11185-1201:3:7:3:13·0 ± 0·373 ab
11185-1191:2:6:1:43·6 ± 0·364 b
11185-1172:4:3:5:12·9 ± 0·360 b
11185-1101:3:5:5:13·1 ± 0·360 b
11185-1160:5:4:2:43·3 ± 0·360 b
11185-1130:4:3:6:13·3 ± 0·350 b
11185-1180:1:1:6:74·3 ± 0·213 c
CCVSeed orchardX73751:6:6:2:02·6 ± 0·287 a
X73765:3:3:2:22·5 ± 0·473 ab
X73713:4:4:1:32·8 ± 0·473 ab
X73770:3:8:2:23·2 ± 0·273 ab
X73782:6:1:4:22·9 ± 0·360 abc
X73740:4:5:3:33·3 ± 0·360 abc
X73701:1:6:1:53·6 ± 0·357 abc
X73731:2:4:4:23·3 ± 0·354 bc
X73790:3:3:5:43·7 ± 0·340 bc
X73720:1:4:6:43·9 ± 0·233 c
CHLockyer10250-35:7:2:1:01·9 ± 0·293 a
10250-25:5:4:1:02·1 ± 0·293 a
10250-66:4:0:3:22·0 ± 0·387 ab
10250-74:6:3:0:22·3 ± 0·387 ab
10250-104:6:1:2:12·3 ± 0·379 ab
10250-92:2:7:2:23·0 ± 0·373 ab
10250-46:4:0:3:22·4 ± 0·467 ab
10250-10:5:4:0:63·5 ± 0·360 bc
10250-50:3:1:4:74·0 ± 0·327 cd
10250-81:1:1:6:64·0 ± 0·320 d
CHNerang10257-910:2:1:1:11·7 ± 0·387 a
10257-35:2:4:3:12·5 ± 0·373 ab
10257-101:9:1:0:42·8 ± 0·473 ab
10257-55:0:5:3:22·8 ± 0·467 ab
10257-84:3:2:5:12·7 ± 0·460 abc
10257-46:0:3:3:32·8 ± 0·460 abc
10257-15:0:3:3:43·1 ± 0·453 bc
10257-72:2:3:2:63·5 ± 0·447 bc
10257-61:2:3:6:23·4 ± 0·343 bc
10257-21:2:1:5:63·9 ± 0·326 c
CHSeed orchardX60928:1:2:3:12·2 ± 0·473 a
X60968:2:1:2:22·2 ± 0·473 a
X60931:1:0:1:02·3 ± 0·967 ab
X60947:2:1:1:42·5 ± 0·567 ab
X60903:3:2:3:22·8 ± 0·461 ab
X60912:4:3:4:23·0 ± 0·360 abc
X60892:0:1:2:23·3 ± 0·642 abc
X60951:4:1:3:63·6 ± 0·440 abc
X60880:0:3:5:23·9 ± 0·230 bc
X60970:0:0:1:24·7 ± 0·30 c

Within CCC provenances, the percentage of resistant seedlings ranged from 93 to 21% for Mt Garnet and 87 to 33% for Yeppoon provenance. For each CCV provenance (Mt McEuan, Brooyar and Woondum), a single family was identified with 93% of seedlings rated as resistant to P. psidii with only one seedling identified as susceptible (Table 4). With Mt McEuan the most susceptible family had a higher average disease rating of 3·3 with 60% resistant seedlings in comparison to Woondum (4·3; 13%) and Brooyar (3·9; 33%) (Table 4).

A large range in resistance was also identified between Lockyer and Nerang (CH) families. Only 20% of seedlings were rated as resistant from the most susceptible Lockyer family in comparison to 93% of seedlings in the most resistant family. Similarly in Nerang provenance, the percentage of seedlings rated as resistant ranged from 87% in the most resistant family to 26% in the most susceptible family (Table 4).

Similar ranges of rust resistance were observed in families derived from seed orchards (Table 4) with CCV ranging from an average disease rating of 2·6 with 87% resistant seedlings to 3·9 and 33% resistance. Disease resistance ranged from 73% of seedlings to 0% within CH seed orchard families. All but one CT and Corymbia hybrid family (Table 5) had average disease ratings levels of less than 3 with the percentage of resistant seedlings ranging from 93 to 53% for CT and 100 to 33% for Corymbia hybrid.

Table 5. Susceptibility of Corymbia torelliana and Corymbia hybrid families to Puccinia psidii ranked least to most susceptible
SpeciesSeed lotNumber of plants/severity class (1:2:3:4:5)Disease ratingResistant seedlings (%)a
  1. a

    Resistant seedlings percentage followed by same letters shows means that do not differ significantly (pairwise comparison < 0·05).

C. torelliana X1683:5:6:1:02·3 ± 0·293 a
X1640:10:4:1:02·4 ± 0·293 a
X1690:7:6:1:12·7 ± 0·287 a
X151:6:3:3:22·9 ± 0·367 ab
X341:2:5:4:33·4 ± 0·353 b
Corymbia hybrid clonesCTVA-02211:3:0:0:01·2 ± 0·1100 a
CTVA-01110:5:0:0:01·3 ± 0·1100 a
CTVA-01811:4:0:0:01·3 ± 0·1100 a
CTVA-0019:6:0:0:01·4 ± 0·1100 a
CTVA-0083:3:0:0:01·5 ± 0·2100 a
CTVA-0062:4:0:0:01·7 ± 0·2100 a
CTVA-0132:4:0:0:01·7 ± 0·2100 a
CTVA-0395:10:0:0:01·7 ± 0·1100 a
Corymbia hybridX1610:3:0:0:02·0 ± 0100 a
X1591:10:4:0:02·2 ± 0·1100 a
X1961:8:6:0:02·3 ± 0·2100 a
X1351:6:6:2:02·6 ± 0·287 ab
X1360:6:2:2:02·6 ± 0·380 ab
X1372:5:3:3:22·9 ± 0·367 bc
X1941:1:3:6:43·7 ± 0·333 c

When examining the repeatability of resistance screening results at the family level, variability in resistance levels was identified among families of provenances tested (Table 6). However, the only significant difference (= 0·03) identified was for Woondum family 11185-115 where 80% of seedlings were rated as resistant in the first assessment and only 43% in the second assessment. Nevertheless, when comparing the most susceptible families within the three provenances across the two experiments (Tables 4 & 6), the same family was recorded for each provenance (Brooyar 10248-1; Mt McEuan 12805-20; Woondum 11185-118). For both Brooyar (10248-1) and Woondum (11185-118), the same percentage of resistant seedlings was recorded in both experiments.

Table 6. Comparison for differences between experiments when retesting selected CCV families for susceptibility to myrtle rust (Corymbia spp.) using chi-square analysis
ProvenanceFamilyResistant seedlings (%) Experiment 1Resistant seedlings (%) Experiment 2χ2P value
Mt McEuan12805-2393673·330·06

To further investigate the repeatability of the screening process, data from both trials were simultaneously analysed using a reduced form of the model used to estimate the genetic parameters, where insignificant Tray effects were dropped and Screening was used to index the analysis rather than assessment traits. This was done in order to estimate a genetic correlation between the two screenings. This type-B correlation ranges from −1 to 1, with the latter estimate indicating that the relative performance of families screened in the two screenings was the same. The estimate of the type-B correlation at the family level (followed by the standard error) was 0·97 ± 0·20. The type-B correlation at the provenance level was very poorly estimated due to the limited sample size, but traditional analyses of variance confirmed the provenance and provenance by screening interaction effects were insignificant across the two trials.

Genetic parameters for rust resistance in CCC, CCV & CH

The genetic parameter estimates (Table 7) indicate the proportion of phenotypic variation that is attributable to differences among provenances (P2) and differences among families within provenances (narrow sense heritability, h2). When data were analysed either one species at a time or across all species, there was clear evidence from all three assessment methods that there was little variation among provenances compared to among families for all three assessment variables.

Table 7. Genetic parameter estimates for the proportion of provenance variation (P2) and narrow sense heritability (h2) from univariate analyses of each taxon (single taxa) presented above parameter estimates for all taxa analysed jointly, where P2 and h2 estimates are presented along the diagonal (italics) with type-A correlations at the provenance and family level below the diagonal
   IncidenceSeverityDisease rating
P 2 Single taxaCCC0·02 ± 0·050 ± 00 ± 0
CCV0·02 ± 0·030·03 ± 0·040·02 ± 0·03
C. henryi 0 ± 00 ± 00 ± 0
All taxaIncidence 0·003 ± 0·003   
Severity0·5 ± 0·77 0·003 ± 0·004  
Rating0·93 ± 0·540·78 ± 0·48 0 ± 0
h 2 Single taxaCCC0·38 ± 0·170·45 ± 0·180·54 ± 0·19
CCV0·18 ± 0·090·24 ± 0·110·26 ± 0·11
C. henryi 0·43 ± 0·150·46 ± 0·150·47 ± 0·15
All taxaIncidence 0·32 ± 0·08   
Severity0·93 ± 0·04 0·42 ± 0·09  
Rating0·98 ± 0·020·98 ± 0·02 0·43 ± 0·09

It should be noted again that analyses to generate the genetic parameter estimates used data that excluded all hybrids and CT in an attempt to obtain parameter estimates that were applicable to the closely related species within the Corymbia complex. The small sample of provenances included for each taxon adds uncertainty to the variance component estimates as evidenced by the large standard errors relative to the estimate. Nevertheless, there is a consistent pattern across the three species and across the three traits with differences among provenances being minimal relative to differences among families. Comparisons among the three taxa at the provenance level indicate variation among CCV provenances is numerically, but not significantly, greater than the other two taxa for most assessment methods (Table 7). Correlations between assessment methods at the provenance level are high but again poorly estimated (large standard errors), indicating all three scoring methods are evaluating similar resistance mechanisms.

The heritability estimates for all three assessment variables are moderate to high and indicate a significant level of additive genetic variance for rust resistance within the populations. Of the three assessments, heritability estimates for severity and rating were highest, with the assessment using disease rating consistently greater than the other two assessment methods. Given that the greatest differentiation among genetic entries was found using the subjective rating, a tentative recommendation of using this scale for further screening may be made. Comparisons among taxa indicate there is greater additive variation within the CCC and CH populations than there is within CCV, and selection of rust resistant populations would require a greater sampling intensity within CCV to achieve similar levels of improvement to what may be achievable in the other taxa.


This is the first study to characterize rust resistance within CCC, CCV, CH, CT and Corymbia hybrids to P. psidii. Inter- and intraspecific variability was observed among Corymbia species and the results clearly identify the potential to select for resistance at the family level within species and provenances tested. Relative to selection among families within provenances, selection at the provenance level alone would provide less genetic improvement for rust resistance. However, the results also suggest the potential for P. psidii to impact on young plantations developed using unimproved seed from all provenances studied, including some commonly used in plantation development in Queensland. For example, >30% of the seedlings from Woondum, the most commonly used provenance in hardwood plantations in Queensland, were given a rating of 4 or 5 and deemed highly susceptible to rust infection.

When comparisons were made between all species, including the Corymbia hybrids (seedling and clones), CCC was the most susceptible to P. psidii and the vegetatively propagated hybrid clones were the least susceptible. The difference between infection in hybrid clones (zero) and the seven full-sib hybrid families is non-intuitive, as neither had been selected for rust resistance and both sources of the hybrids had similar selection strategies for growth and form traits. A hypothesis that different foliage characteristics between the vegetatively propagated clonal hybrids and the seedling family hybrids may lead to differences in infection could be plausible in this case, even though the clones and the seedlings were acclimated together for 4 weeks prior to inoculation. Various inoculation studies using the same system as that employed in this study have demonstrated an influence of leaf age on infection after artificial inoculation (authors' unpublished data). Given the differences in the propagation history and clear preference of P. psidii to infect juvenile foliage, further inoculation studies of vegetatively propagated and seedling hybrids must be undertaken to verify the resistance of the hybrid clones.

The Corymbia hybrid families ranged from 33 to 100% resistant to P. psidii, with three of the seven hybrid families tested rated as 100% resistant. None of 95 families of the parental taxa had families rated as 100% resistant to the disease. This may indicate that the hybrids have benefited from polygenic or additive gene action controlling resistance to the disease. This needs further investigation, but supports findings by Alves et al. (2012) that additive genetic variation from the parental species is an important contributor to rust resistance of eucalypt hybrids.

There was no apparent relationship between climatic conditions at the provenance origin and disease resistance. While the three provenances ranked as most resistant, Mt McEuan, Lockyer and Presho, are from drier inland areas (703, 892 and 675 mm mean annual rainfall (MAR), respectively), the three most susceptible provenances included one higher rainfall provenance, Nerang (1439 mm MAR) and also the two drier inland provenances of Ballon and Mt Garnet (656 and 832 mm MAR), respectively. Woondum (MAR 1536 mm), a provenance used commonly in plantation development in Queensland, showed moderate levels of resistance to rust with only six out of the 11 families tested having a percentage of resistant seedlings >60%.

Woondum shows higher disease tolerance levels to another basidiomycete fungus endemic to Australia, Quambalaria pitereka (Lee, 2007). However, similar to the results obtained in this study with P. psidii, resistance to Q. pitereka can be identified within all provenances (Pegg et al., 2010; Brawner et al., 2011). While it is not clear if resistance to one pathogen implies resistance to the other, there is some evidence that different resistance mechanisms may be at work. Correlations between breeding value predictions for rust resistance from this study and Quambalaria breeding values from two other field trials (451D and 451H; Brawner et al., 2011) were not significantly different than zero, implying little to no relationship between resistance mechanisms. Given the biology of these fungi, this is not surprising. Quambalaria pitereka only enters via stomata but never grows intercellularly in comparison to P. psidii, which penetrates directly through the cuticle, entering the cell and producing haustoria (Xavier et al., 2001; Pegg et al., 2009). Interestingly, Presho provenance, identified as being highly susceptible to Q. pitereka (Lee, 2007), shows similar levels of resistance to P. psidii as Woondum.

The fact that seed orchard material (CCV and CH), which has been field selected for resistance to Q. pitereka, showed similar resistance levels to ‘unimproved’ families further supports differences in resistance mechanisms for the two pathogens. However, it could also reflect the fact that selection for resistance was based solely on performance in field trials, the result of which may be influenced by variable climatic conditions and non-uniform exposure to inoculum (Pegg et al., 2010).

Previous studies have identified that Q. pitereka, a co-evolved pathogen, is quite variable based on DNA sequencing (Pegg et al., 2009) and isolate aggressiveness (Pegg et al., 2010). It therefore must also be considered that the level of natural inoculum and variation in virulence may have influenced field selection, and further screening under controlled conditions is required to compare resistance to both Q. pitereka and P. psidii. Given this, the level of variation in rust resistance of families from the seed orchard should allow selection of good growth form and rust resistance for future breeding and deployment. The high rust resistance of the CT and Corymbia hybrid families may indicate that CT has higher levels of resistance to the strain of rust in Australia. This needs further study.

There was good correspondence between the two screenings, with significantly different results (< 0·05) being obtained for only one of 31 families (11185-115). This finding supports the idea that the artificial inoculation process used to screen seedlings is robust between screenings. While the very high genetic correlation between the two artificial screening procedures demonstrates repeatability in this study, greater certainty that this process will be useful in identifying germplasm that will express resistance requires further optimization, examining influence of seedling age and nutrient status in addition to testing under field conditions.

Comparisons of genetic parameters estimated for each species separately indicate a lower level of genetic control in CCV, which may require a greater screening effort to achieve similar levels of genetic improvement relative to CCC and CH. However, similarities in the ratio of the amount of variation explained by provenance to variation explained by families within provenances indicate resistance is present in most populations of the species examined. This provides guidance to tree improvement programmes so that they may focus on selecting individuals from a wide range of populations rather than focusing on a single population to provide resistance. Studies by Ochieng et al. (2010) and Shepherd et al. (2008) indicate that there is weak genetic structuring between populations within the northern spotted gums (CCC, CCV and CH), with an isolation-by-distance model occurring where proximal populations were more similar than more distant ones. This weak genetic structuring may be caused by long distance gene flow associated with two important pollinator groups of the spotted gums, fruit bats (e.g. grey headed flying foxes) and nectar feeding birds (Southerton et al., 2004; Bacles et al., 2009). This weak genetic structure and large interpopulation gene flow is probably the reason that resistance genes are found in all populations and may explain why it is better to select for resistance at the family rather than population level.

The heritability of traits indicates that giving individual seedlings a rating provides a better means of discriminating between resistant and susceptible material than using incidence or average severity assessments. Nevertheless, the genetic correlations near to one suggest all three assessment methods are likely to provide similar rankings of families for rust resistance, and the method applied for future resistance screening work could therefore be based upon the ease of use and the speed of assessment. Given the limited range of provenances sampled in this study, it will be essential to screen additional material to verify the repeatability of these genetic parameter estimates.

Based solely upon clear visual differences in the pattern and extent of damage within families and populations, there appear to be multiple resistance mechanisms acting within the spotted gums. For example, obvious differences were noted in the location and type of pustule formation with individuals demonstrating either numerous, small pustules or few very large pustules. Differences within the hypersensitive rating class were also noted (data not presented). These findings may indicate that rust resistance shown here has a complex mode of inheritance regulated by multiple genes rather than a single resistance gene. Alves et al. (2012) found that additive (major gene) and non-additive effects were equally as important in rust resistance. The present data may fit this model. While this complicates selection for disease resistance, it means that some individuals may use major resistance alleles or a range of mechanisms to achieve resistance to rust. Further work on the molecular genetics of rust resistance in the genetically diverse plant populations that the rust is currently spreading through is clearly required.

This study demonstrates the feasibility of identifying spotted gum germplasm that is resistant to P. psidii using an artificial screening method. However, there is a need to determine if the resistance to the single rust biotype identified in Australia confers resistance to different biotypes present in other countries. In addition to this, including a range of isolates of P. psidii collected from different hosts and locations over time in Australia will be important, accompanied by development of a host differential allowing for rapid testing for differences in pathogenicity between different isolates. Genetic parameter estimates indicate that the variation that is available within provenances and families is sufficient to develop rust resistant breeds of spotted gum. Infected individuals were identified within most families, indicating either that a single resistance gene is not typically dominant in the maternal population or that many genes are interacting to provide disease resistance within this genus. Larger scale studies with a greater number of provenances and sources from Corymbia seed orchards, as well as families within provenances, should be effective at identifying P. psidii resistant selections for breeding populations.


The authors would like to thank John Oostenbrink and Tony Burridge for preparation of seedlings and David Bush, Helen Nahrung and John Huth for their review and comments for this manuscript.