Sequential infections by 32 isolates of Phoma medicaginis increase production of phytoestrogens in Medicago polymorpha var. brevispina

Keywords: 4′-O-methylcoumestrol, annual medic, annual Medicago spp, coumestrol, genetic variation, phoma black stem and leaf spot, Phoma medicaginis, phytoestrogen.

For decades there has been strong interest in annual Medicago species as high-quality forage legumes (Chatterton and Chatterton 1996). They are highly valued for animal production (Lüscher et al. 2014) and play a critical role in dryland farming systems (Barbetti et al. 2006). Phoma black stem and leaf spot (caused by Phoma medicaginis) is a damaging disease on both annual and perennial Medicago species in temperate and Mediterranean regions (Lamprecht and Knox-Davies 1984). It greatly reduces seed and herbage yield and, where severe, can cause incomplete defoliation and premature death in highly susceptible annual Medicago cultivars (Barbetti 1989b, 1995a; Barbetti and Fang 1991; Barbetti and Nichols 1991; Ellwood et al. 2006b). Phoma medicaginis has a wide host range (e.g. Lamprecht and Knox-Davies 1984), also infecting Pisum and Trifolium species (Barbetti and Khan 1987). Classification of Phoma species can be challenging (Fatehi et al. 2003), particularly when based on morphology, where the major distinction from Ascochyta species has long been conidial size and aseptate conidia (Boerema and Bollen 1975); however, this has been resolved with the use of molecular sequencing (e.g. Kamphuis et al. 2008, 2012).

First disease symptoms are brown to black spots on leaves, stems and petioles, and as disease progresses, these coalesce causing chlorosis and finally abscission of leaflets and stem collapse (Barbetti 1987, 1989b). Phoma medicaginis can also cause significant root disease on annual Medicago species (Barbetti 1989a), and a rot complex in both crowns (Rodriguez and Leath 1992) and roots (Rodriguez 2005) of lucerne (Medicago sativa). In Western Australia, surveys showed it to be widely associated with root disease in annual Medicago species (You et al. 2000). The disease is particularly severe during wet weather conditions in spring when conidia from pycnidia within lesions are readily spread by rain splash (Barbetti 1987, 1989b; Barbetti et al. 2006; Ellwood et al. 2006a). Rising spring temperatures also foster the disease (Barbetti 1987, 1991), and in annual Medicago species, disease is most severe at day/night temperatures of 21°C/16°C, followed by 18°C/13°C, and least severe at 15°C/10°C (Barbetti 1987).

Phoma medicaginis infection stimulates production of phytoestrogens in annual Medicago species (Barbetti 1991, 1995a, 2007; Barbetti and Nichols 1991; Barbetti et al. 2006, 2020). Fields et al. (2018) reported that lucerne inoculated with Stemphylium vesicarium contained 169 ± 25.1 mg kg⁻¹ of coumestrol compared with 3.4 ± 0.84 mg kg⁻¹ in control plants. Phytoestrogens can provide some advantages to plants, such as protection against some fungal and bacterial pathogens (Wasserman et al. 2013) and against attack by pests (He and Dixon 2000; Deavours and Dixon 2005). However, they can result in lower ovulation rates in grazing animals, with consequent reduced reproduction or even infertility, particularly with sheep (Croker et al. 1994a, 1994b, 1999, 2005; Smith et al. 1979) but also with cattle (Barbetti et al. 2020).

These phytoestrogens are similar to the endogenous estrogens of mammals in terms of structure and function (Pierson and Ferkin 2015). The majority of phytoestrogens are phenolic compounds including isoflavones and coumestans, for which extensive research has been conducted. (Patisaul and Jefferson 2010). Phytoestrogen content in annual Medicago species can also increase in absence of plant disease (Francis and Millington 1965), for instance, in response to soil type (Francis and Millington 1971), superphosphate availability (Marshall and Parkin 1970) and stage of plant growth (Francis and Millington 1965). However, it is P. medicaginis that particularly stimulates elevated levels of phytoestrogens such as coumestrol (Barbetti and Fang 1991; Barbetti and Nichols 1991; Barbetti 2007) in annual Medicago species.

Although various management strategies such as close grazing (Barbetti 1989b; Barbetti et al. 2006, 2020) and application of fungicides (Barbetti 1983, 1989b) offer reduction of Phoma black stem and leaf spot, locating host resistance remains the most promising approach (Barbetti 1989b, 1990; Barbetti et al. 2020). In addition, our recent study (Omidvari et al. 2021) highlighted that the intrinsic ability of annual Medicago cultivars to produce phytoestrogens in the absence of the pathogen, and their comparative ability to produce phytoestrogens in the presence of P. medicaginis, were together critical in developing more resistant cultivars with improved reproductive outcomes for grazing animals. However, in that study, we utilised a ‘once-only’ pathogen inoculation, which contrasts with what happens naturally in the field. There is a need for studies that better epitomise the in-field, multiple infection cycles in relation to final disease incidence/severity and production cycles of coumestrol and 4′-O-methyl coumestrol.

We hypothesised that levels of both Phoma black stem and leaf spot and phytoestrogens would be higher after multiple infection cycles. Therefore, in the present study, we investigated differences across leaf and stem disease incidence and severity during and after multiple infection cycles of 32 isolates of P. medicaginis, and the relationship of these disease parameters with production of coumestrol and 4′-O-methyl coumestrol. We also examined phylogenetically informative gene regions to determine the diversity of local isolates of P. medicaginis and how these compared with isolates from elsewhere. The local isolates had previously been investigated for their virulence and effect on phytoestrogen production across three annual Medicago cultivars (M. truncatula cv. Cyprus, M. polymorpha var. brevispina cv. Serena, and M. murex cv. Zodiac) (Omidvari et al. 2021). The relative virulence of these isolates varied across different assessment parameters and their effects on phytoestrogen production were also different.

Thirty-two isolates of P. medicaginis collected from a range of annual Medicago species and cultivars in Western Australia were used. These were obtained from the Culture Collection at the Department of Primary Industries and Regional Development, Western Australia, as lyophilised cultures in glass ampules. Full isolate details are provided in Omidvari et al. (2021).

Inoculum preparation was as described by Omidvari et al. (2021). Briefly, preserved isolates in lyophilised ampoules were subcultured onto Sanderson and Srb medium (Dhingra and Sinclair 1995) for 21 days at 25°C (±2°C) under 16 h/8 h light/dark. After 3 weeks, 5 mL sterile water was added to each plate and then conidia were harvested by rubbing the colony surface with a bent glass rod. Conidial concentration of each isolate was adjusted to 5 × 10⁶ spores mL⁻¹ by using a haemocytometer counting chamber, and the inoculum produced was always applied immediately in virulence tests (Chihaoui et al. 2015).

The highly susceptible M. polymorpha var. brevispina (Barbetti 1990, 1995a, 1995b) was used to evaluate the virulence of different isolates across multiple inoculations. A single cultivar was chosen in order to avoid cultivar × isolate interactions as found in our earlier study (Omidvari et al. 2021). To remove any potential fungal seed contamination and to ensure high germination rates, seeds were surface-sterilised with 70% ethanol for 30 s, washed with sterile distilled water, and then scarified to break hardseed dormancy (when freshly ripened they also contain embryo dormancy) using sandpaper. Seeds were then transferred to Petri dishes lined with moist, sterile blotting paper. Imbibed seeds were transferred to water agar plates and incubated at 25°C in darkness until radicles emerged (Mhadhbi et al. 2005). Seeds were then transplanted into pots 6 cm by 6 cm by 14 cm, containing a portion of steam-pasteurised potting mix that consisted of 2.5 m³ fine composted pine bark, 1 m³ coco peat, 5 m³ brown river sand, 10 kg slow-release fertiliser Osmoform NXT (22 N:2.2 P₂O₅:9.1 K₂O:1.2 Mg + trace elements; Everris International, Geldermalsen, the Netherlands), 10 kg dolomite (CalMag), 5 kg gypsum clay-breaker, 5 kg extra fine limestone, 4 kg iron heptasulfate and 1 kg iron chelate. Potting mix was pasteurised at 63°C for 45 min. Each pot was sown with four seeds at 5 mm depth and pots were maintained in a naturally lit greenhouse at 18°C (±4°C), watered daily with deionised water and allowed to drain to field capacity. Plants were fertilised weekly, according to manufacturer’s recommendation, with Thrive (Yates Australia, Sydney, NSW, Australia), containing macronutrients (25 N:5 P:8.8 K) and the full range of micronutrients needed for plant growth.

Plants were spray-inoculated with the conidial suspension of each isolate until run-off, using a hand-held and operated aerosol sprayer. Tween 20 (0.001%) was added to conidial suspensions to ensure that inoculation droplets remained attached to the plant. Control plants were also treated with 0.001% Tween 20®, but in sterile distilled water. Inoculated plants were then placed into clear plastic containers with lids, measuring 770 mm long, 570 mm wide and 475 mm high, used as inoculation/incubation chambers to maintain high humidity (40 pots per container). Walls and lids of incubation boxes were internally misted with sterile distilled water and box lids closed for 72 h post-inoculation to maintain high humidity (Ellwood et al. 2007). The whole experiment was fully repeated once.

There were three inoculation timings. The first inoculation was when plants were 6 weeks of age, with assessments made in situ for the different disease parameters at 8 weeks of age (14 days post-inoculation (dpi)), and four plants from each isolate randomly picked for harvest a further 2 weeks later at 10 weeks of age (i.e. 28 dpi) for phytoestrogen assessments. The second and third inoculations were conducted when plants were 10 and 18 weeks old, respectively, with similar time intervals for disease assessments (14 dpi) and harvesting of plants for phytoestrogen analyses (28 dpi).

Four disease assessment parameters were used, as reported previously (Omidvari et al. 2021): (1) leaf disease incidence (%LDI), calculated as the percentage of leaves per pot showing disease symptoms; (2) leaf disease severity (%LDS), visually estimated as the leaf area diseased per pot as a percentage of the total leaf area; (3) petiole/stem disease incidence (%PDI), calculated as the percentage of petioles/stems per pot showing disease symptoms; (4) petiole disease severity (%PDS), visually estimated as the petiole/stem area diseased per pot as a percentage of the total petiole/stem area. All disease assessments were made by the same person throughout.

Details of solvent use for phytoestrogen assessments are as provided previously (Omidvari et al. 2021). Coumestrol and 4′-O-methylcoumestrol standards were purchased from Sigma-Aldrich (St. Louis, MO, USA) and ALB Technology (Henderson, NV, USA), respectively, and used to prepare a calibration curve from 1 to 100 mg L⁻¹ with nine concentrations for each standard. The relationship between peak area and concentration was linear, with correlation coefficient (r) >0.9997 for both coumestrol and 4′-O-methylcoumestrol. High-pressure liquid chromatography (HPLC)-grade methanol, HPLC-grade acetonitrile and HPLC-grade trifluoroacetic acid were purchased from Sigma-Aldrich. Water was obtained from a Milli-Q water purification unit (Millipore Australia). The stock standard solutions (100 μg mL⁻¹) of coumestrol and 4′-O-methyl coumestrol were prepared in aqueous methanol (2:8 v/v) and stored in the dark at 4°C. A reference standard (10 μg mL⁻¹) for HPLC retention times was prepared daily by dilution of the stock solutions with aqueous methanol (2:8 v/v).

For phytoestrogen assessment, plants sampled for each isolate or control treatment were harvested and bulked together into a single composite sample. Details of quantification of phytoestrogens used in phytoestrogen assessments are as provided by Omidvari et al. (2021). Three weeks after the final disease assessment, stem samples were harvested from all treatments for analysis of contents coumestrol and 4′-O-methyl coumestrol. Stem samples were used because diseased leaves and petioles had dehisced and because stems are a reliable indicator or phytoestrogen levels (Barbetti and Fang 1991). Samples were oven-dried at 70°C, and finely ground in two steps: first with a mechanical grinder that coarsely ground the stem material, and second by freezing with liquid nitrogen and finely grinding the material in a mortar and pestle. Ground samples (0.2 g) were extracted via methodology adapted from Fields et al. (2018) with 2 mL aqueous methanol (2:8 v/v) in glass sample vials (8 mL). All stem samples were hand-mixed for 30 s and put on an end-over-end mixer for 16 h at room temperature. Samples were filtered through glass fibre syringe filters, and supernatant (1 mL) was transferred into 2-mL glass auto-sampler vials. Extracted samples were analysed for coumestrol and 4′-O-methyl coumestrol content by HPLC. For comparative analysis, the extracts were spiked with 4-hydroxycoumarin (50 μg mL⁻¹) as an internal standard and relative quantifications determined based on HPLC–diode array detector (DAD) peak areas that were normalised based on the area of the recovered internal standard peak. Retention times for coumestrol and 4′-O-methylcoumestrol were at 28.5 and 43.8 min, respectively. These two compounds were identified in the samples based on their relative retention times and absorbance spectra. The HPLC analyses were performed with an Agilent 1260 series instrument (Agilent Technologies, Santa Clara, CA, USA) equipped with a quaternary pump, auto injector and DAD. Separation was performed using an Apollo reversed phase C18 column (250 mm × 4.6 mm ID, 5 μm particle size; Grace Discovery Sciences, Bannockburn, IL, USA) at 25°C. The mobile phase consisted of solvent A 0.1% (v/v) trifluoroacetic acid in water, and solvent B acetonitrile. The elution gradient was as follows: 0–40 min, 20–50% B; 40–41 min, 50–50% B; 41–50 min, 50–100% B; and held at 100% B for 10 min. The flow rate set at 1 mL min⁻¹ with the compounds detected at 254 nm and injection volume 20 μL. The column was re-equilibrated for 5 min between samples. Amounts of coumestrol and 4′-O-methylcoumestrol were quantified as mg kg⁻¹ dry weight of plant material extracted.

In addition to the above, three separate subsamples were taken from each of the dried harvest materials from each isolate × timing of inoculation treatment (and similarly for control treatment), and separately assessed for amounts of coumestrol and 4′-O-methylcoumestrol. The mean of the values for three subsamples is presented as the coumestrol or 4′-O-methylcoumestrol value for each isolate, or for the control. Preliminary studies (data not presented) highlighted that three subsamples ensured a reliable and repeatable measure of the level of coumestrol or 4′-O-methylcoumestrol present.

Sample DNA was extracted from 2-week-old fungal cultures of 28 isolates previously identified as P. medicaginis growing on Sanderson and Srb medium. The remaining four isolates were referenced according to their already available molecular identification details in GenBank. A modified procedure of Cenis (1992) was used for DNA extraction. Briefly, fresh mycelial mat, ∼2 mm by 2 mm, was put into a 2-mL tube containing two ceramic beads (2 mm) and 300 μL extraction buffer (200 mM Tris–HCl pH 8.5, 250 mM NaCl, 25 mM EDTA, 0.5% w/v SDS). The mycelial mat was homogenised in a Precellys Evolution homogeniser (Bertin Technologies, Montigny-le-Bretonneux, France) at 2490g in two cycles of 60 s. Then, 3 M sodium acetate (150 μL, pH 5.2) was added and maintained at room temperature for 10 min. Tubes were then centrifuged at 16 816g for 10 min and the supernatant was transferred to a fresh 2-mL Eppendorf tube. An equal volume of isopropanol (∼450 μL) was added to the tube and maintained for at least 15 min at 22°C to precipitate the DNA. Precipitated DNA was pelleted by centrifuging at 16 816g for 15 min, then washed with 70% w/v ethanol, vacuum-dried for 10 min and resuspended in 50 μL Tris-EDTA buffer. DNA samples were held at 4°C until needed.

The internal transcribed spacer (ITS) ribosomal DNA (rDNA) region can be of limited value for resolution of closely related species or strains (Ellwood et al. 2006a); therefore, three additional phylogenetically informative regions including β-tubulin, calmodulin, and P. medicaginis-specific elongation factor nested intron (EFNI)-1α were examined for distinguishing isolates. A PCR-based assay was conducted using each primer set to amplify a DNA product of predicted size from P. medicaginis isolates (Table 1) (Ellwood et al. 2006a). Amplification was performed in a volume of 25 μL with 1 μL each primer (final concentration of each primer 10 pmol μL⁻¹), 1 μL 10 ng DNA template, 12.5 μL DreamTaq Green PCR Master Mix (2×) (Thermo Fisher Scientific, Waltham, MA, USA), and 9.5 μL nuclease-free water. DNA amplifications were carried out in a 9600 Thermocycler (PerkinElmer, Waltham, MA, USA) programmed at 95°C for 8 min, followed by 35 cycles at 95°C for 15 s, then 55°C (58°C for β-tubulin; Table 1) for 20 s, then 72°C for 60 s, then a final elongation step of 72°C for 5 min. The DNA products of predicted size from P. medicaginis were visualised by electrophoresis on a 1% (w/v) agarose gel containing GelRed stain (diluted to 1/10 000) and viewing under UV light. PCR products were then sequenced by Macrogen, Seoul, South Korea.

Table 1.  Primer sets for PCR assay of target loci within Phoma medicaginis isolates used in this study.

The P. medicaginis isolates in this study were initially compared with closely related species using ITS sequence data alone because those data were available in GenBank. Then additional phylogenetic trees were constructed including β-tubulin, calmodulin, P. medicaginis-specific EFNI-1α fragments. The 32 isolates used in the phylogenetic tree included 28 sequenced isolates from this study with the addition of another four isolates (WAC4736, WAC4738, WAC4741 and WAC7980) previously sequenced by Ellwood et al. (2006a) (see accession numbers in Table 2).

Table 2.  Phoma medicaginis isolates and GenBank accession numbers for gene regions used in this study.

For ITS, we selected the most closely related GenBank ITS sequences from five P. medicaginis reference isolates in GenBank (AY504634, CBS316.90, DQ092494, KF181254 and MK765001) and, with our 32 ITS isolates’ sequence data, assembled a tree rooted to Phaeosphaeria nodorum (U77362). For the second tree, P. medicaginis isolate CBS316.90 from GenBank was included, with accession nos AY831518, AY831541, AY831548 and AY831563 (Table 2) relating to β-tubulin, calmodulin, P. medicaginis-specific EFNI-1α and ITS sequences, respectively. Sequence alignment and phylogenetic tree building were performed using Geneious v. 9.1.4 (Biomatters, San Diego, CA, USA). Sequence consensuses were used for BLAST in NCBI, and multiple alignment (alignment type, global alignment with free end gaps; cost matrix, 65% similarity (5.0/−4.0); gap open penalty, 12; gap extension penalty, 3) was used for phylogenetic tree building (genetic distance model, Tamura–Nei; tree building method, neighbour-joining tree; resampling method, bootstrap; random seed, 534 442; no. of replications, 100; support threshold, >50%).

The whole greenhouse experiment was fully repeated once. There was one annual Medicago species (M. polymorpha var. brevispina), 32 P. medicaginis isolates and one control (mock inoculation), and three inoculation times and corresponding harvest times. The total number of pots in each experiment was 396, comprising cultivar (1) × isolates (33) × inoculation times (3) × replications (4), where 132 pots were inoculated once (at 6 weeks, harvested at 10 weeks), 132 pots were inoculated twice (at 6 and 10 weeks, and harvested at 14 weeks) and 132 pots were inoculated three times (at 6, 10 and 18 weeks and harvested at 22 weeks). Each experiment was arranged as a randomised block design using the ‘Generate a Standard Design’ function in GenStat 18.1 (18th Edition; VSNi, Hemel Hempstead, UK). Normality of data and homogeneity of the original and repeat experimental datasets were tested before conducting analyses. Data from the original and the repeat experimental datasets were not significantly different (P > 0.05) by t-test. Therefore, data from the original and repeat experiments were combined, and analysed as a single dataset. %LDI, %PDI, %LDS and %PDS were assessed. Fisher’s least significant differences (l.s.d.s) were used to indicate significant differences between treatments. Correlations among all of the disease parameters and between all of the disease parameters and phytoestrogens were computed by using the linear regression function of GenStat.

There were significant effects of P. medicaginis isolate on %LDI, %PDI, %LDS and %PDS (all P ≤ 0.001) in every round of inoculation, confirming variation for virulence across isolates of P. medicaginis as assessed across the different disease parameters. Control plants in all inoculation rounds remained disease-free (Table 3).

Table 3.  Phoma black stem and leaf spot disease parameters, namely leaf disease incidence (%LDI), petiole/stem disease incidence (%PDI), leaf disease severity (%LDS) and petiole/stem disease severity (%PDS) at 14 days post-inoculation, and production of coumestrol and 4′-O-methylcoumestrol in plants at 28 days post-inoculation, with 32 P. medicaginis isolates from Western Australia onto M. polymorpha var. brevispina.

For the first inoculation, the three most virulent isolates in terms of %LDI were WAC3704 (41%), WAC4744 (39%) and WAC4260 (38%) and three least virulent isolates were WAC3653, WAC4254 and WAC4741 (all 0%). For the second inoculation, isolates WAC4736 (38%), WAC4251 (37%) and WAC4254 (36%) were most virulent and WAC4741 (12%) was least virulent. For the third inoculation, most virulent isolates were WAC4255 (76%) and WAC4734 (75%) and least virulent was WAC4741 (43%) (Table 3).

The most virulent isolates in the first inoculation as assessed by %PDI were WAC4744 (27%) and WAC4732 (26%) and the least were WAC3653, WAC3656, WAC4254, WAC4736, WAC4738 and WAC4741 (all 0%) and WAC3706 (2%). For the second inoculation, the most virulent isolates were WAC4736 (22.5%), WAC4259 (21.1%), WAC4260 (20%) and WAC4739 (19.9%), whereas least virulent were WAC4255 (2%), WAC4741 (2%) and WAC3653 (3%). In relation to the third inoculation, most virulent were WAC4732 (61%) and WAC4260 (58%) and least was WAC7980 (13%) (Table 3).

For the first inoculation, the most virulent isolates with regard to %LDS were WAC4734 (41%), WAC6729 (39%) and WAC4745 (38%), whereas WAC3653, WAC4254 and WAC4741 rated nil LDS. For the second inoculation, the most virulent isolates were WAC4252 (45%) and WAC4254 (42.5%) and the least virulent were WAC3653 (5%), WAC4741 (5%), WAC3658 (7%) and WAC4742 (7%). For the third inoculation, most virulent were WAC4732 (52%) and WAC4736 (51%) and least virulent WAC6729 (12%) and WAC7980 (13%) (Table 3).

For the first inoculation, most virulent isolates in terms of %PDS were WAC7980 (30%), WAC6729 (29%) and WAC4732 (28%), whereas WAC3653, WAC3656, WAC3706, WAC4254, WAC4736, WAC4738 and WAC4741 all rated no nil PDS. In the second inoculation, most virulent was WAC4736 (34%) and least virulent were WAC4255 (0%), WAC4741 (1%), WAC3653 (2%) and WAC3654 (2%), but these last four did not differ from the control plants. For the third inoculation, most virulent isolates were WAC4736 (53%), WAC4732 (52%) and WAC4258 (52%) and least virulent WAC4741 (8%) and WAC6729 (10%) (Table 3).

There were significant main effects of inoculation rounds on all disease parameters %LDI, %PDI, %LDS and %PDS (all P ≤ 0.001) (Table 3). The highest %LDI was recorded for the third inoculation (57.6%) and the lowest for the first inoculation (23.5%), with the second inoculation intermediate (26.6%). The highest, intermediate and lowest %PDI occurred for the third (41.6%), second (13.8%) and first (11.6%) inoculations, respectively. With regard to %LDS, the third inoculation had the highest rating (30.1%), followed by the second (22.2%) and first (17.5%) inoculations. Likewise, for %PDS, the third inoculation had the highest rating (27.3%), the second inoculation was intermediate (12.3%), and the first inoculation had the lowest (11.3%) (Table 3).

There was a significant P. medicaginis isolate × inoculation round interaction effect on all disease parameters %LDI, %PDI, %LDS and %PDS (all P ≤ 0.001) (Table 3).

Leaf disease incidence (%LDI) was highest with isolates WAC4255 and WAC4734 in the third round of inoculation, whereas isolates WAC3653, WAC4254 and WAC4741 showed the lowest %LDI in the first round of inoculation (Table 3).

Petiole disease incidence (%PDI), was highest for isolates WAC4732 and WAC4260 in the third round of inoculation, whereas it was lowest for isolates WAC4736, WAC3653, WAC3656, WAC4254, WAC4738, WAC4741 and WAC3706 in the first round of inoculation and for isolates WAC4741, WAC4255 and WAC3653 in the second round of inoculation (Table 3).

Leaf disease severity (%LDS) was highest for isolates WAC4732, WAC4736 in the third round of inoculation, whereas it was lowest for isolates WAC3653, WAC4741 and WAC4254 in the first round of inoculation (Table 3).

Petiole/stem disease severity (%PDS) was highest for isolates WAC4736, WAC4732 and WAC4258 in the third round of inoculation, whereas WAC4736 was one of six isolates with the lowest %PDS in the first round of inoculation (Table 3).

There were highly significant overall effects of inoculation round in terms of phytoestrogen expression (P ≤ 0.001) (Table 3). The highest level of coumestrol was for the third inoculation (658.3 mg kg⁻¹ stem dry weight), and the lowest level for the first inoculation (262.5 mg kg⁻¹), and the level for the second inoculation was intermediate (483.4 mg kg⁻¹) (Fig. 1). Likewise, the highest level of 4′-O-methylcoumestrol was for third inoculation (183.2 mg kg⁻¹), the lowest for the first inoculation (24.2 mg kg⁻¹), and the level for the second inoculation was intermediate (65 mg kg⁻¹) (Fig. 1).

Fig. 1.  Production of coumestrol and 4′-O-methylcoumestrol at 28 days post-inoculation for three inoculation timings: first at 6 weeks only; second repeated at 10 weeks; third repeated at 18 weeks after germination.

Overall, Phoma black stem and leaf spot resulted in very high levels of both coumestrol and 4′-O-methyl coumestrol compared with levels in uninoculated control plants (Table 3).

For the first inoculation, levels of coumestrol produced ranged from 45 to 447 mg kg⁻¹ and levels of 4′-O-methyl coumestrol ranged from 0 to 82 mg kg⁻¹. The isolate that resulted in the highest level of coumestrol was WAC4258 (447 mg kg⁻¹), and the lowest levels of coumestrol were caused by isolates WAC4732, WAC4741, WAC3706 and WAC4261 (96, 99, 104 and 116 mg kg⁻¹, respectively). For the first inoculation, isolate WAC3657 resulted in the greatest production of 4′-O-methyl coumestrol (82 mg kg⁻¹).

For the second inoculation, the greatest amount of coumestrol was produced in plants inoculated with isolates WAC7980 and WAC4744 (909 and 874 mg kg⁻¹, respectively), and the least in plants inoculated with WAC4261 (166 mg kg⁻¹). In relation to 4′-O-methylcoumestrol, the highest production resulted from isolates WAC4739, WAC4256, WAC4258, WAC3658 (135, 120, 119 and 112 mg kg⁻¹, respectively).

For the third inoculation, the greatest amount of coumestrol resulted from inoculation with isolates WAC4739 and WAC4741 (1247 and 1166 mg kg⁻¹, respectively). In terms of 4′-O-methylcoumestrol, isolates WAC3705, WAC3658 and WAC3704 (344, 284 and 268 mg kg⁻¹, respectively) caused greatest production.

There was a significant P. medicaginis isolate × round of inoculation interaction effect on both coumestrol and 4′-O-methylcoumestrol production (P ≤ 0.001) (Table 3). Production of coumestrol was greatest for isolates WAC4739 and WAC4741 in the third round of inoculation, whereas nearly one-third of the isolates produced the lowest level of coumestrol in the first round of inoculation (Table 3). Production of 4′-O-methyl coumestrol was highest for isolates WAC3705, WAC3658 and WAC3704 in the third round of inoculation, whereas isolates WAC3706, WAC4261, WAC4732 and WAC4741 did not produce any 4′-O-methyl coumestrol in the first round of inoculation (Table 3).

Overall, meaned across the three inoculations, there was a significant positive correlation of leaf disease incidence with coumestrol (P < 0.05), and of both leaf and petiole disease incidence with 4′-O-methyl coumestrol contents (P < 0.01 and P < 0.05, respectively). This was most evident in the first inoculation, where %LDI was significantly and positively correlated with production of coumestrol (P < 0.05), and third inoculation, where %LDI was significantly and positively correlated with coumestrol and 4′-O-methylcoumestrol (P < 0.05 and P < 0.01, respectively) and %PDI was correlated with 4′-O-methylcoumestrol production (P < 0.05).

Leaf disease incidence was significantly and positively correlated with %PDI, %LDS and %PDS in first, second and third inoculation rounds (all P < 0.001) and for all inoculation rounds grouped and meaned together (P < 0.001) (Table 4). %PDI was significantly and positively correlated with leaf disease severity %LDS and petiole disease severity (%PDS) in the first (P < 0.05 and P < 0.01), second (both P < 0.001) and third (P < 0.01 and P < 0.001 respectively) inoculation rounds. %LDS was significantly and positively correlated with %PDS in the first (P < 0.05), second (P < 0.001) and third (P < 0.05) inoculation rounds and for all inoculation rounds grouped and meaned together (P < 0.001) (Table 4).

Table 4.  Correlation coefficients (r) among Phoma black stem and leaf spot disease parameters, namely leaf disease incidence (%LDI), petiole/stem disease incidence (%PDI), leaf disease severity (%LDS) and petiole/stem disease severity (%PDS), and for disease parameters with phytoestrogen contents in Medicago polymorpha var. brevispina in first, second and third inoculation rounds by Phoma medicaginis separately and meaned across all inoculation rounds as a group.

The P. medicaginis isolates were initially compared by using ITS sequence data alone because it provided the most readily available comparative data in GenBank. The phylogenetic tree constructed with 32 isolates of P. medicaginis as well as five accessions from GenBank (AY504634, KF181254, CBS316.90, MK765001 and DQ092494) separated into two clades (Fig. 2). The P. medicaginis isolates from GenBank clustered together to form Clade I. All isolates from the present study (including four P. medicaginis isolates previously sequenced by Ellwood et al. 2006a; shown in bold in Fig. 2) clustered together into Clade II. Generally, there was a very high degree of similarity for all isolates from Western Australia (Fig. 2). All sequences have been deposited in the GenBank database (see details in Table 2).

Fig. 2.  Neighbour-joining tree of ITS generated by Geneious with the Tamura–Nei genetic distance model demonstrating the phylogenetic relationship of isolates from GenBank (in bold and italic) (Clade I) and 32 Phoma medicaginis isolates including four reference isolates from Western Australia previously sequenced by Ellwood et al. (2006a) (in bold). GenBank accession numbers for isolates of P. medicaginis submitted in the present study and as based on their ITS sequences are provided in Table 2. Branch support (bootstrap values) presented above branches based on 1000 bootstrap replicates. Tree is rooted to Phaeosphaeria nodorum (U77362).

Sequencing of EFNI-1α highlighted significant differences in length and sequence among P. medicaginis isolates and compared with other species. This dissimilarity inhibited sequence alignments with other species such that the phylogenetic tree using combined gene regions was solely constructed among P. medicaginis isolates. The tree of concatenated sequences of β-tubulin, calmodulin and EFNI-1α constructed for the 32 isolates of P. medicaginis and CBS316.90 retrieved from GenBank separated into two clades (Fig. 3). WAC3704 and WAC3706 clustered together in Clade I. All remaining isolates (including four isolates previously sequenced by Ellwood et al. 2006a) clustered together with CBS316.90 to form Clade II.

Fig. 3.  Neighbour-joining tree of the concatenated sequences of β-tubulin, calmodulin and EFNI-1α sequence data (accession numbers are provided in Table 2) generated by Geneious with the Tamura–Nei genetic distance model. It demonstrates phylogenetic relationship of WAC3704 and WAC3706 (Clade I) and 30 other Phoma medicaginis isolates including four reference isolates from Western Australia previously sequenced by Ellwood et al. (2006a) (in bold) and isolate CBS316.90 from GenBank (in bold and italic). Branch support (bootstrap values) presented above branches based on 1000 bootstrap replicates.

Fig. 3.  Neighbour-joining tree of the concatenated sequences of β-tubulin, calmodulin and EFNI-1α sequence data (accession numbers are provided in Table 2) generated by Geneious with the Tamura–Nei genetic distance model. It demonstrates phylogenetic relationship of WAC3704 and WAC3706 (Clade I) and 30 other Phoma medicaginis isolates including four reference isolates from Western Australia previously sequenced by Ellwood et al. (2006a) (in bold) and isolate CBS316.90 from GenBank (in bold and italic). Branch support (bootstrap values) presented above branches based on 1000 bootstrap replicates.