Effect of temperature and duration of wetness during initial infection periods on disease development, fungal biomass and mycotoxin concentrations on wheat inoculated with single, or combinations of, Fusarium species

Authors


*E-mail: xiangming.xu@emr.ac.uk

Abstract

Experiments were conducted under controlled environment conditions to study the relationship between environmental conditions, development of fusarium head blight (FHB) and mycotoxin production. A single isolate from each of four Fusarium species (F. avenaceum, F. culmorum, F. graminearum and F. poae) was used to inoculate wheat ears separately. Combinations of two or three isolates were also used to inoculate ears simultaneously. Inoculated ears were subjected to various combinations of duration of wetness (6–48 h) and temperature (10–30°C). For all inoculations, both incidence of spikelets with FHB symptoms and concentration of mycotoxins generally increased with increasing length of wetness period and temperature. There were significant positive correlations among disease incidence, fungal biomass (quantified as total amount of fungal DNA) and mycotoxins. Mycotoxin production was also greatly enhanced by high temperatures (≥ 20°C) during initial infection periods. In single-isolate inoculations, F. poae was the least aggressive. There was no evidence to support synergetic interactions between fungal isolates in causing visual symptoms; rather the results suggest, in most cases, the presence of competitive interactions. Furthermore, the competition led to large reductions in fungal biomass compared to single-isolate inoculations, often > 90% reduction for the weaker isolate(s). In contrast, mycotoxin productivity increased dramatically in the co-inoculations, by as much as 1000 times, suggesting that competition resulted in greater production of trichothecene mycotoxins. The F. graminearum isolate was most competitive and isolates of the other three species were similar in their competitiveness.

Introduction

Fusarium head blight (FHB) has become a serious threat to wheat and barley worldwide (Parry et al., 1995; Dubin et al., 1997; McMullen et al., 1997; Xu & Berrie, 2005). In addition to causing significant yield losses, FHB is of greater significance under certain conditions because of accumulation of mycotoxins in infected grain. The toxigenic capabilities of the pathogens causing FHB differ. Fusarium avenaceum, F. culmorum, F. graminearum and F. poae can produce a range of mycotoxins (Bottalico & Perrone, 2002) whereas Microdochium nivale and M. majus (formerly M. nivale var. nivale and Microdochium nivale var. majus (Glynn et al., 2005)) apparently do not produce mycotoxins. The mycotoxins of most significance are the trichothecenes: deoxynivalenol (DON) and nivalenol (NIV) (Placinta et al., 1999). Contaminated grain is unsuitable for animal and human consumption because of adverse effects of such toxins on health (IARC, 1993; Li et al., 1999; Bennett & Klich, 2003).

Knowledge of the qualitative and quantitative effects of environmental factors on initial infection and subsequent FHB development is important for assessing the disease risks. It is generally agreed that plants are most susceptible to FHB at anthesis (Sutton, 1982; Lacey et al., 1999). Most studies investigating the relationship between environmental factors and FHB development have been based on field monitoring of FHB epidemics to infer the effects of environmental conditions on FHB development (Moschini & Fortugno, 1996; Lacey et al., 1999; De Wolf et al., 2003). These studies suggested that warm and moist conditions during the anthesis period are important for FHB development. In addition, field investigations also showed the crucial importance of environmental factors in influencing accumulation of the mycotoxins due to F. graminearum on cereals (Hooker et al., 2002). Controlled environment studies using detached ears were conducted to determine the effect of temperature and humidity on the infection of wheat ears by F. avenaceum, F. graminearum, F. culmorum and M. nivale (Rossi et al., 2001), although it is not known to what extent the information obtained using detached ears is directly applicable to whole plants. Mycotoxin research, however, has been mainly focused on the toxicity of individual mycotoxins and their detoxification (IARC, 1993; Li et al., 1999; Bennett & Klich, 2003).

In Europe, FHB can be caused by several pathogens: F. graminearum, F. culmorum, F. avenaceum, F. poae, M. nivale and M. majus. Pathogens may differ in their requirements for infection and subsequent development (Rossi et al., 2001) and hence models developed purely from field data may lead to significant errors. Equally, use of models derived from single-pathogen inoculations might also lead to erroneous conclusions if there are significant interactions between different FHB pathogens during disease development and mycotoxin production. The differential response of FHB pathogens to fungicides was one of the causes for the inconsistency of chemical control of FHB in field conditions, in terms of both the reduction in disease and mycotoxins (Simpson et al., 2001; Pirgozliev et al., 2003). Competitive interactions between two Microdochium species (nivale and majus) and F. culmorum varied greatly with temperature (Simpson et al., 2004). It is therefore essential to understand the effect of environmental factors on individual FHB pathogens and on the interactions between FHB pathogens, with respect to subsequent disease development and mycotoxin production.

A series of experiments were conducted to study the disease development, fungal growth and mycotoxin accumulation following inoculation of wheat ears with one or more toxigenic Fusarium pathogens. Wheat ears of potted plants were first inoculated with four toxigenic FHB pathogens separately (F. avenaceum, F. culmorum, F. graminearum and F. poae) and then subjected to various combinations of temperature and duration of wetness in controlled environment (CE) cabinets. In addition, wheat ears were co-inoculated with two or three of the pathogens and then subjected to different conditions in CE cabinets. As well as visual disease assessment and quantification of mycotoxins in the grain, FHB pathogens were also detected and quantified using molecular techniques (Nicholson et al., 2003).

Materials and methods

A single isolate from F. avenaceum, F. culmorum, F. graminearum and F. poae was used to inoculate wheat ears separately or in combinations in Sanyo CE cabinets; isolates were from UK field populations. In co-inoculations, the focus was on inoculation of F. poae with the other three toxigenic species as this was most frequently detected toxigenic species during field sampling in 2001 and 2002 (Xu et al., 2005). For each inoculation experiment, there were four temperatures (in the range of 10–30°C, depending on the FHB pathogen to be inoculated) and five wetness periods of 6, 12, 24, 30 and either 36 or 48 h (Figs 1 and 2) at each temperature, giving 20 combinations of wetness period and temperature for modelling (Table 1). Because of the limited number of available CE cabinets at a given time, it was not possible to replicate treatments. Instead, each inoculation experiment was repeated over time (Table 1). In some instances one of the cabinets failed during the experimental period and hence only three temperatures were available; thus for three combinations of FHB pathogens the exact temperature differed between replicate inoculations (Table 1).

Figure 1.

Incidence of spikelets with fusarium head blight symptoms on wheat ears inoculated with a single isolate of Fusarium avenaceum (a, b), F. culmorum (c, d), F. graminearum (e, f) and F. poae (g, h) in controlled environment cabinets plotted against (a, c, e, g) the temperature and duration of wetness and (b, d, f, h) the amount of fungal DNA (filled symbols) and/or mycotoxin (unfilled symbols) in the grain. Different symbols represent different replicate inoculation experiments.

Figure 2.

Incidence of spikelets with fusarium head blight symptoms on wheat ears inoculated with a mixture of two isolates of: Fusarium culmorum and F. poae (a, b), F. graminearum and F. poae (c, d) and F. avenaceum and F. poae (e, f) in controlled environment cabinets plotted against (a, c, e) the temperature and duration of wetness and (b, d, f) mycotoxins or amount of fungal DNA. Different symbols represent different replication inoculation experiments.

Table 1.  Summary of the 19 inoculation experiments on wheat (cv. Clare) with one or more of the four toxigenic fusarium head blight pathogens in controlled environment cabinets: number of treatment (temperature x wetness) with detectable and quantifiable amount of fungal DNA (dPCR & qPCR), and average incidence of diseased spikelets and mycotoxins per inoculation over all treatments. In co-inoculation experiments, each component species (isolate) has the same inoculum concentration
SpeciesaRepInoculumTreatmentbNo. of treatmentFusarium poaeF. graminearumF. avenaceumF. culmorumIncidence (%)NIV (ppb)DON (ppb)
dPCRqPCRdPCRqPCRdPCRqPCRdPCRqPCR
  • a

    Fa –Fusarium avenaceum, Fc –F. culmorum, Fg –F. graminearum, Fp –F. poae.

  • b

    For each temperature, there were five wet periods: 6, 12, 24, 30 and 36 h, unless stated otherwise; each treatment was a single combination of temperature and duration of wetness.

  • c

    48 h instead of 36 h wetness at each temperature.

  • d

    Not available because the flour samples were destroyed by fire in the analytical laboratory in the University of Naples.

Fa1c7·5 × 10510, 15, 20, 25°C20    2020  26Insufficient flour for quantification
 26·3 × 10510, 15 ,20, 25°C20    2020  16
Fc12·0 × 10515, 20, 25°C15      151520 4 160     0
22·0 × 10515, 20, 25°C15      151525 5 586    91
Fg15·6 × 10415, 22, 30°C15  1515    46    81 5 233
21·4 × 10415, 22, 30°C15  1515    17    85 4 341
Fp12·4 × 10510, 18, 25°C151515      26 1 022     0
21·0 × 10510, 18, 25°C1515 7      14    41    67
Fc, Fp11·0 × 10510, 15, 20, 25°C2012 0    19 523 3 597   116
 21·5 × 10510, 18, 25°C1510 5    12 327 3 731    55
Fg, Fp1c5·0 × 10510, 18, 25°C15 0 01515    21N.A.d
 23·0 × 10515, 22, 30°C15 1 01514    16  
Fa, Fp13·4 × 10515, 23, 30°C1510 0   9 3  2510 459   269
 22·8 × 10515, 23, 30°C1515 0  14 6  3311 770   291
Fa, Fg, Fp12·5 × 10510, 15, 20, 25°C20 4 0151515 8  74N.A.
 22·5 × 10510, 15, 20, 25°C2019 0202020 6  29  
Fa, Fp, Fc12·5 × 10510, 18, 25°C1515 0   0 015 75714 117   296
 22·5 × 10510, 15, 20, 25°C20 2 0  1813201334 9 650   240
Fc, Fp, Fg11·0 × 10510, 15, 20, 25°C2014 02012  15 047 1 82311 439

Spore production

Conidia were produced on sucrose nutrient agar incubated at 20°C for 8–10 days and transferred to near UV light for 10–14 days (photo-period 12 h daily). Conidia were washed from plates with distilled water, and conidial concentration was adjusted as necessary (Table 1) so as to be the same for all component species, and equal numbers of conidia of each species were mixed before inoculation. Conidia were assessed for germination for each inoculation experiment; germination was high, ranging from 70% to 99%.

Plant materials

Seeds of wheat cultivar Clare were treated with fungicides (Beret Gold – a.i. fludioxonil 24·3 g L−1, Syngenta – applied at 2 L tonne−1) and sown in trays prior to vernalising at 4°C for 3 months. Seedlings were then transferred to 2 L pots, five seedlings per pot, and placed in a glasshouse compartment at 20°C. The plants were monitored and, where necessary, fungicides were used to control powdery mildew (no fungicides were applied after the ear emergence).

Inoculation and disease assessment

Plants were inoculated at mid-anthesis, having been misted with distilled water prior to inoculation. Each ear was inoculated using a fine hand-held aerosol sprayer, with approximately 1·5 mL conidial suspension. Inoculated plants were placed in CE cabinets set to an appropriate temperature with two misting nozzles switched on to maintain wetness (high humidity). There were four or five pots for each treatment (a single temperature and wetness combination), usually five plants in each pot and 1–3 ears for each plant, giving a total of 25–50 ears (at least 300 spikelets) for each treatment.

Appropriate numbers of plants were taken out of each cabinet after the designated duration of wetness and moved to another glasshouse compartment, at 20°C, where they remained until harvest. During this time plants were watered from below using automatic watering benches. The number of spikelets with FHB symptoms on all inoculated ears was recorded each week until the natural process of ear maturity prevented further recording (usually two assessments were done). At maturity, all inoculated ears were harvested, hand threshed, weighed, freeze-dried, weighed again for estimating moisture content and then stored in heat-sealed plastic bags to prevent re-absorption of water. The dried samples were stored at 4°C until milling. After milling, 0·5 g of flour was used to extract fungal DNA and the remaining flour was used to quantify relevant mycotoxins.

Detecting and quantifying fungal DNA

DNA was extracted from flour as described previously (Nicholson et al., 1996), quantified using the method described by Simpson et al. (2000), and adjusted to 20 ng µL−1 for use in PCR. Diagnostic PCR was carried out following the protocols described previously by Nicholson et al. (2003) and Xu et al. (2005). The species-specific PCR primer pairs used to detect M. nivale and Fusarium species were those described in Nicholson et al. (1996) and Nicholson et al. (2003), respectively. Reaction components and amplification conditions for competitive PCR were the same as those for diagnostic PCR except for the addition of the relevant competitor template. Following amplification, PCR products were separated by electrophoresis, and fungal and competitor PCR products were quantified using Molecular Analyst software (Bio-Rad). The ratio of target and competitor PCR product was estimated and the amount of fungal DNA derived by reference to a standard curve generated using purified DNA of the relevant target species. Standard negative, positive and competitor control reactions were included in each PCR run. All PCRs were replicated at least twice.

Detection and quantification of mycotoxins

DON, NIV, T2 (T-2 toxin), DAS (diacetoxyscirpenol), ZON (zearalenone) and HT2 (HT-2 toxin) standards were purchased from Sigma Chemical Company. Water for the high-performance liquid chromatography (HPLC) mobile phase was purified in a Milli-Q system (Millipore). All other chemicals and solvents of HPLC-grade were purchased from Merck. Clean-up columns used to prepare samples were purchased from Rhone Diagnostic Technologies Ltd. Trichothecene standard stock solutions (10 µg mL−1) were prepared by dissolving 1 mg of each pure standard into 100 mL of methanol HPLC grade. Then 200 µL of each trichothecene standard stock solution were transferred into a 100 mL flask, completely dried under nitrogen stream at room temperature, and dissolved in methanol to 20 µg L−1. The vials were closed and stored in the dark at 4°C; fresh stocks were prepared every week.

Trichothecene extraction

A representative homogenized flour sample (10 g) was shaken for 1 h at room temperature in acetonitrile-methanol solvent (84:16, v/v; 50 mL). The suspension was centrifuged to 4000 g for 10 min at 4°C and 5 mL of supernatant was vacuum dried and dissolved in methanol-water (70:30, v/v). The solution was passed through a 0·22 µL cellulose syringe filter (Millipore) prior to injection (20 µL) onto the HPLC column, which was linked to a mass spectrometer.

HPLC conditions

The HPLC column was a Prodigy 5 (m ODS-3V (Phenomenex, 250 × 4·6 µm)), set at a constant temperature (50°C). Mobile phases used were: A – water, B – acetonitrile/methanol (50:50 v/v). Flow rate was fixed at 0·8 mL min−1 and the eluent conditions were those in Table 2. Mycotoxins were quantified by comparing peak areas to those of authentic trichothecene standards using a calibration curve.

Table 2.  Eluent conditions used in the HPLC analysis of mycotoxins
Time (min)% water% CH3CN/CH3OH (70:30)
090 10
765 35
920 80
1320 80
15 0100
19 0100
2190 10

Spectrometric conditions

Spectrometric conditions required an API 3000 Triple Quadrupole Mass Spectrometer equipped with Turbo Ion-Spray as interface and MRM experiment (multiple reaction monitoring), negative ions for NIV, DON and ZON, positive ions for T2, HT2 and DAS. Turbo-gas (nitrogen) temperature was fixed to 350°C and IonSpray Voltage (IS) was +5500 V (positive ions) and –4500 V (negative ions). Other parameters such as DP (declustering potential) or collision energy were optimized for each mycotoxin by the infusion technique using standard solution at 10 µg mL−1 and flow of 8 µL min−1. The recovery rate for NIV, DON and ZON was 81%, 95% and 73%, respectively; limit of detection and limit of quantification for the three toxins were all < 5 ppb.

Statistical analysis

The main objective of this study was to determine the quantitative relationships between FHB development, fungal growth and mycotoxin accumulation with the duration of wetness and temperature during the initial infection. Logistic regression analysis (Cox & Snell, 1989) was used to relate the incidence (p) of spikelets with FHB symptoms to duration of wetness (w) and temperature (t). In this form of analysis the number of spikelets with FHB symptoms per treatment (a single combination of temperature and wetness duration in each inoculation experiment) is assumed to be binomially distributed. Then the logit transformation of the proportion, i.e. ln(p/(1 − p)) or ln((D + 0·5)/(N + 0·5 − D)) (where D and N are the number of diseased spikelets and total number of spikelets inoculated, respectively) is expressed as a multiple linear regression on independent variables. Similarly, logistic regression analysis was used to investigate the relationship of fungal biomass with disease incidence, temperature and duration of wetness, since fungal biomass DNA was quantified as percentage of total DNA in the sample. In the analysis of fungal biomass data, the error distribution was assumed to be binomial with the binomial ‘total’ set to 100 because the percentage data were used. For analysis of fungal DNA, a replicate inoculation experiment was included only if most treatments had quantifiable DNA. For a few samples with F. poae in which the pathogen was detected by diagnostic PCR but amounts were too low to quantify, the amount of DNA was set at 0·005% as a default value.

Regression analysis was used to investigate the relationship of mycotoxin concentrations with disease incidence, fungal DNA, temperature, duration of wetness and grain moisture (%). Mycotoxin concentration (DON or NIV) was logarithm-transformed before analysis to reduce the degree of variance heterogeneity. In all experiments, only one toxin (either NIV or DON) was predominant; the minor toxin was highly correlated with the predominant toxin. Therefore, all regression analyses on toxins were carried out only for the predominant toxin.

A generalized nonlinear mixed-model approach was adopted in all analyses; replicate inoculation of the same pathogen or pathogen combination was treated as a random factor. To avoid small values of parameter estimates, both temperature (t, °C) and duration of wetness (w, h) was divided by 10 before analysis. All statistical analyses were done using Genstat™ (Payne, 2002).

Results

A total of nineteen inoculation experiments were conducted (Table 1). With the exception of the co-inoculation with F. graminearum, F. poae, and F. culmorum, all inoculations with single or mixed species were repeated once. The summary of disease, fungal DNA and mycotoxin data for each inoculation is given in Table 1. The predominant toxin produced by the toxigenic species was either DON or NIV. In all inoculations, there were considerable variations in the FHB disease complex (incidence of diseased spikelets and fungal DNA) and mycotoxin production between replicate inoculations; this variability is not commented on further.

Infection by F. avenaceum

Average incidence over two replicate inoculations increased from 10 to 32% when wetness duration increased from 6 to 48 h, and from 11 to 38% when temperature increased from 10 to 25°C (Fig. 1a). This relationship was described by a linear relationship of the logit of disease incidence with the product of t and w (tw) (Table 3). Correlation between fitted and observed incidences was 0·92. Fusarium avenaceum DNA was quantified in all treatments, ranging from 0·04 to 1·21% (median = 0·18%). The fungal DNA content was positively correlated with disease incidence (r = 0·46; P < 0·01) (Fig. 1b). The logit of the fungal DNA was linearly related to inline image (Table 3); correlation between fitted and observed DNA values was 0·69.

Table 3.  Summary of regression analysis of incidence of diseased spikelets (p), fungal DNA (D) and mycotoxin concentrations (ppb) involving four Fusarium species on temperature (t), duration of wetness (w) and grain moisture (GM, %)
SpeciesaDiseaseFungal DNAMycotoxin
  • a

    Fa –Fusarium avenaceum, Fc –F. culmorum, Fg –F. graminearum, Fp –F. poae.

  • b

    Both temperature (t) and duration of wetness (w) were divided by 10 before analysis to avoid very small regression coefficients. The r is the correlation between predicted and observed values on the transformed scale; numbers in the brackets are the standard errors of the corresponding parameter estimates.

Faln(p/(1 − p)) = −1·14 + 0·27tw; r = 0·92; (0·126, 0·022)bln(D/(1 − D)) = −6·06 + 0·754inline image; r = 0·69; (0·31, 0·126)Insufficient flour for quantification
Fcln(p/(1 − p)) = −1·11 + 0·036t2w2; r = 0·85; (0·175, 0·005)ln(D/(1 − D)) = −3·63 + 2·55inline image; r = 0·70; (0·49, 0·357)ln(NIV) = 8·12 + 0·093t2w + 0·110D; r = 0·96; (0·325, 0·012, 0·030)
Fgln(p/(1 − p)) = −0·645 + 1·84inline image; r = 0·999; (0·02, 0·03)ln(D/(1 − D)) = −5·53 + 0·132tw − 0·0028t2w2; r = 0·87; (0·06, 0·032, 0·0012)ln(DON) = 8·17 + 0·88inline image + 1·05D; r = 0·88; (0·08, 0·14, 0·43)
Fpln(p/(1 − p)) = −1·42 + 0·066t2w; r = 0·66; (0·42, 0·008)ln(D/(1 − D)) = −6·94 + 0·205t2w2 − 0·053t2w; r = 0·34; (1·68, 0·035,0·010)ln(NIV) = 5·08 + 1·23D+ 0·056t2w; r = 0·97; (1·22, 0·58, 0·015)
Fc, Fpln(p/(1 − p)) = −1·04 + 0·081t2w; r = 0·88; (0·10, 0·008)Only a few samples had quantifiable amount of F. culmorum DNA.ln(NIV) = 7·19 + 0·187t2w – 0·935GM; r = 0·82; (0·15, 0·024, 0·354)
Fg, Fpln(p/(1 − p)) = −1·33 + 0·105t2w; r = 0·91; (0·27, 0·011)ln(D/(1 − D)) = −4·79 + 0·153tw; r = 0·55; (0·16, 0·045) (F. graminearum)Samples were destroyed in a lab fire in University of Naples
Fa, Fpln(p/(1 − p)) = −0·87 + 0·081w2 + 3·33t − 0·698t2; r = 0·77; (0·19, 0·017, 1·37, 0·30)None of the treatments had quantifiable amounts of F. poae or F. avenaceum DNA.ln(NIV) = 8·84 + 1·37inline image + 3·48t − 0·73t2;r = 0·97; (0·04, 0·07, 0·73, 0·16)
Fa, Fg, Fpln(p/(1 − p)) = 0·014 + 0·071t2w − 0·62w + 0·093tw2; r = 0·96; (0·006, 0·035, 0·18, 0·038)ln(D/(1 − D)) = −5·08 + 0·272tw; r = 0·70; (0·10, 0·041) (F. graminearum)Samples were destroyed in a lab fire in University of Naples
Fa, Fc, Fpln(p/(1 − p)) = −0·204 + 0·082t2w; r = 0·89; (0·049, 0·011)Only a few treatments had quantifiable amounts of fungal DNA.ln(NIV) = 8·63 + 3·67inline image + 1·04inline image; r = 0·98; (0·052, 0·262, 0·093)
Fc, Fp, Fgln(p/(1 − p)) = 1·08 + 0·210t2w − 1·492inline image; r = 0·90; (0·32, 0·029, 0·266)Only a few treatments had quantifiable amounts of F. graminearum DNA.ln(DON) = 7·50 + 0·162t2w; r = 0·91; (0·164, 0·017)

Infection by F. culmorum

Average incidence increased from 11 to 47% when duration of wetness increased from 6 to 36 h, and from 14 to 30% when temperature increased from 15 to 25°C (Fig. 1c). The logit of the incidence was linearly related to one term only (t2w2) (Table 3); correlation between fitted and observed incidences was 0·85. All samples had a quantifiable amount of F. culmorum DNA, ranging from 0·05 to 9·38% (median = 0·37%). There was significant correlation between fungal DNA and disease incidence (r = 0·63; P < 0·01) (Fig. 1d). The logit of the fungal DNA was positively related linearly to inline image (Table 3); correlation between fitted and observed DNA values was 0·70.

Mycotoxin (NIV) concentration varied greatly between treatments, ranging from 798 to 13 183 ppb, and was correlated (P < 0·01) with both disease incidence (r = 0·85) (Fig. 1d) and fungal DNA (r = 0·76). It increased with increasing temperature and wetness duration. In addition to the amount of fungal DNA, NIV was also positively related to t2w (Table 3); the correlation between fitted and observed NIV values on the logarithm scale was 0·96.

Infection by F. graminearum

Average incidence appeared to increase with increasing duration of wetness and, to a lesser extent, with increasing temperature (Fig. 1e); a wetness period as short as 6 h resulted in 20% of spikelets with FHB symptoms. Average incidence increased from 19 to 42% when duration of wetness increased from 6 to 36 h; average incidence was 24, 35 and 34% at 15, 22 and 30°C, respectively. Only a single variable (inline image ) was needed to predict disease incidence under the present experimental conditions (Table 3); correlation between fitted and observed incidences was close to 1·0 (0·999).

Fusarium graminearum DNA was quantified in all treatments, ranging from 0·05 to 0·85% (median = 0·31%) and correlated with the disease incidence (r = 0·83, P < 0·01, Fig. 1f). The logit of the fungal DNA was positively related linearly to two derived variables (tw, t2w2) (Table 3); the correlation between fitted and observed DNA values was 0·87. Mycotoxin (DON) concentration varied greatly between treatments, ranging from 522 to 11 962 ppb, and was correlated with disease incidence (r = 0·63, P < 0·01) (Fig. 1f) and fungal DNA (r = 0·69, P < 0·01). In addition to the amount of fungal DNA, DON was also log-linearly related to (inline image) (Table 3); the correlation between fitted and observed toxins was 0·88.

Infection by F. poae

Overall incidence generally increased with increasing duration of wetness, particularly at high temperatures, and with increasing temperature (Fig. 1g). Average incidence increased from 13 to 25% when duration of wetness increased from 6 to 36 h, and from 14 to 26% when temperature increased from 10 to 25°C. The logit of the incidence was positively related linearly to t2w only (Table 3); correlation between fitted and observed incidences was 0·66. Fusarium poae DNA was detected in all 30 treatments and quantified in 22 out of 30 samples (Table 1), ranging from 0·01 to 1·22% for the samples with a quantifiable amount of DNA (median = 0·32%). The correlation (r = 0·57) between disease incidence and fungal DNA was significant (P < 0·01) (Fig. 1h). The logit of the fungal DNA was linearly related to t2w and t2w2 (Table 3); however, the correlation between fitted and observed DNA values was low (r = 0·34). This low correlation mainly resulted from the fact that in eight treatments the fungal DNA detected was below the limit of quantification.

There were significant (P < 0·01) positive correlations of mycotoxin (NIV) with disease incidence (r = 0·82) (Fig. 1h) and fungal DNA (r = 0·68). NIV concentration varied greatly between treatments, ranging from 17 to 3834 ppb. As for disease incidence, NIV generally increased with increasing temperature and duration of wetness. In addition to the amount of fungal DNA, NIV was also log-linearly related to t2w (Table 3).

Infection by F. culmorum and F. poae

Overall incidence increased with increasing temperature and increasing duration of wetness when temperatures were ≥ 18°C (Fig. 2a). Incidence increased from 17 to 33% when duration of wetness increased from 6 to 36 h, and from 19 to 37% when temperature increased from 10 to 25°C. Based on the dPCR and qPCR results, F. culmorum and F. poae were similarly competitive (Table 1). The logit of disease incidence was linearly related to t2w only (Table 3); correlation between fitted and observed values was 0·88. The main mycotoxin (NIV), ranging from 59 to 22 777 ppb, was positively correlated (r = 0·86, P < 0·01) with disease incidence (Fig. 2b). It was positively related log-linearly to t2w but negatively to grain moisture content (Table 3). Correlation between fitted and observed NIV values was 0·82.

Infection by F. graminearum and F. poae

Overall disease incidence generally increased with increasing temperature and increasing duration of wetness although the incidence at 30°C was less than at 25°C (Fig. 2c). Fusarium graminearum was more competitive than F. poae: only one grain sample contained a detectable amount of F. poae DNA whereas all samples except one contained quantifiable amounts of F. graminearum DNA, ranging from 0·02 to 3·0% (median = 0·54%). The logit of disease incidence was related to t2w only (Table 3) and correlation between fitted and observed values was 0·91. Fusarium graminearum DNA was significantly correlated (r = 0·54, P < 0·01) with disease incidence (Fig. 2d) and was related to tw only (Table 3).

Infection by F. avenaceum and F. poae

While average disease incidence increased with increasing duration of wetness, the overall temperature effects were relatively small (Fig. 2e). Average incidence increased from 20 to 42% when duration of wetness increased from 6 to 36 h; average incidence was 24, 34 and 29% at 15, 23 and 30°C, respectively. Fusarium avenaceum and F. poae appeared to be similarly competitive. Most samples contained detectable amounts of fungal DNA of both species, but there was sufficient fungal DNA for quantification in only nine samples (all F. avenaceum) (Table 1).

The logit of disease incidence was related to t, t2 and w2 (Table 3); correlation between fitted and observed values was 0·77. However, because only three temperatures were examined, the reliability of conclusions on the effects of temperature was debatable. There was a correlation of disease incidence with mycotoxins (NIV) (r = 0·61, P < 0·01) (Fig. 2f). NIV, ranging from 851 to 31 521 ppb, generally increased with increasing duration of wetness of temperature and was log-linearly related to several weather variables (Table 3).

Infection by F. graminearum, F. avenaceum and F. poae

Overall incidence generally increased with increasing temperature; there was no effect of wetness duration at low temperatures but there was a marked and positive effect of wetness duration at higher temperatures (Fig. 3a). Average incidence increased from 44 to 61% when duration of wetness increased from 6 to 36 h, and from 36 to 63% when temperature increased from 10 to 25°C. The diagnostic PCR and competitive PCR results indicated that F. graminearum and F. poae were the most and least competitive species, respectively (Table 1). None of the grain samples contained quantifiable amounts of F. poae DNA. All the samples contained detectable amounts of F. avenaceum DNA but only about half contained quantifiable amounts. All samples contained sufficient F. graminearum DNA for quantification, ranging from 0·09 to 2·4% (median = 0·29%). In 14 samples with sufficient DNA of both F. graminearum and F. avenaceum to permit quantification, there was a greater (P < 0·01) amount of F. graminearum DNA (0·54%) than that of F. avenaceum (0·04%) on the basis of a simple logistic regression analysis.

Figure 3.

 Incidence of spikelets with fusarium head blight symptoms on wheat ears inoculated with a mixture of three isolates of: Fusarium graminearum, F. avenaceum and F. poae (a, b), F. culmorum, F. avenaceum and F. poae (c, d) and F. culmorum, F. graminearum and F. poae (e, f) in controlled environment cabinets plotted against (a, c, e) the temperature and duration of wetness and (b, d, f) mycotoxins or amount of fungal DNA. Different symbols represent different replication inoculation experiments.

The logit of disease incidence was related to w, tw2 and t2w (Table 3); correlation between fitted and observed values was 0·96. There was a positive correlation of disease incidence with F. graminearum DNA (r = 0·43, P < 0·01) although the magnitude of this correlation varied considerably between two inoculations (Fig. 3b). The logit of F. graminearum DNA was positively related linearly to tw only (Table 3).

Infection by F. culmorum, F. avenaceum and F. poae

Overall disease incidence increased with increasing temperature; at low temperatures the duration of wetness had no consistent effect on disease but at higher temperatures disease incidence increased with greater wetness duration (Fig. 3c). Fusarium culmorum appeared to be most competitive on the basis of diagnostic PCR and competitive PCR results; its DNA was detected in all samples and quantifiable in 20 out of 35 samples. Fusarium avenaceum was least competitive in one replicate inoculation and F. poae was least competitive in the other (Table 1). Fusarium avenaceum DNA was quantified in 13 samples in which F. culmorum DNA was also quantified; average F. avenaceum DNA in these samples was 0·02%, which was significantly less than that of F. culmorum (0·25%) based on a simple logisitic regression analysis.

The logit of disease incidence was related to t2w only (Table 3); correlation between fitted and observed incidences was 0·89. There was a significant correlation between disease incidence and mycotoxins (NIV) (r = 0·70, P < 0·01) (Fig. 3d). NIV varied from 541 to 48 991 ppb and was positively related log-linearly to inline image and inline image (Table 3).

Infection by F. culmorum, F. graminearum and F. poae

The effect of wetness duration was minimal except at 25°C (Fig. 3e). A wet period as short as 6 h led to an average of nearly 45% of diseased spikelets. The diagnostic PCR and competitive PCR results suggested that F. graminearum was most competitive whereas F. poae and F. culmorum were similar in their competitiveness (Table 1).

The logit of disease incidence was related to inline image and t2w (Table 3); correlation between fitted and observed incidences was 0·90. DON ranged from 798 to 52 421 ppb and, on average, was about six times more abundant than NIV. The greatest amount of DON was produced at 20 and 25°C. DON concentration was correlated with disease incidence (r = 0·80, P < 0·01) (Fig. 3f) and log-linearly related to t2w only (Table 3). Correlation between fitted and observed DON values was 0·91.

Discussion

This study has for the first time obtained a considerable amount of data under controlled environment conditions on the relationships between visual disease, fungal development and mycotoxin accumulation as well as species competition. A few points should be borne in mind when interpreting the results. Due to the limited number of CE cabinets available, it was not possible to repeat the treatments or to conduct single and mixed-isolate inoculations simultaneously. Observed variability between replicate inoculations over time might also be partly due to differences in inoculum quality/strength or differences in the condition of the wheat plants. Furthermore, because only a single isolate from each FHB species was used, the competition results in fungal biomass and mycotoxin production may be interpreted as specific to either the isolates or the species.

The single-isolate inoculations yielded several important results and corroborates many previous findings from field studies. Fusarium poae is generally believed to be less pathogenic and aggressive than other FHB pathogens (Pettersson & Olvång, 1995; Brennan et al., 2003). The present study supports this view as the overall incidence of diseased spikelets was the lowest and, in many cases, while F. poae DNA could be detected in grain at harvest, there was insufficient DNA present to quantify. Of the four species, inoculation with F. graminearum resulted in the highest overall disease incidence but inoculation of F. culmorum led to the greatest amount of fungal DNA accumulating in grain. This suggests that F. culmorum may colonise infected grain more extensively than the other FHB species.

Overall the incidence of spikelets with FHB symptoms increased with increasing temperature and duration of wetness, as found in previous field-based studies (Lacey et al., 1999; Rossi et al., 2001; De Wolf et al., 2003). Models have been developed from the single-isolate inoculation data to describe the disease incidence, fungal DNA and mycotoxin in relation to temperature and wetness duration during the initial infection period. The exact model form differed between the four pathogens. Disease incidence can be well predicted from the temperature and duration of wetness except for F. poae. These models differ considerably from those reported previously (Rossi et al., 2001) using detached spikelets, probably because of the differences in the temperature range used. However, the value of these models for practical applications may be limited because of the presence of more than one FHB species in field situations as well as the presence of other components of the microflora of wheat heads. In such situations, estimating disease incidence is complicated, and will depend on the particular species present and their relative inoculum potential as well as their relative spatial patterns within the crop.

The inter-relationships between incidence of diseased spikelets, fungal DNA and mycotoxins can be better understood from the single-isolate inoculation data. There were consistent positive correlations among the disease incidence, fungal DNA and mycotoxins. Correlation between fungal DNA and disease incidence was weaker than the other two pairwise correlations for F. culmorum and F. poae whereas the opposite was true for F. graminearum. This may be because the visual symptoms are those on the floral tissues rather than the grain, which is the tissue in which fungal DNA is measured. Thus, it could be that F. culmorum and F. poae colonise floral tissues to a greater extent than they do grain. In contrast, F. graminearum may colonise both tissue types to a similar extent. The stronger relationship between visual symptoms and mycotoxin content of grain may, in part, be because mycotoxin can be translocated from floral tissues to the grain, as suggested by Snijders & Kretching (1992) and supported by Draeger & Nicholson (unpublished results). Thus, the level of disease on the outer tissues may provide a good indicator of the overall toxin accumulation. In accordance with its low aggressiveness, F. poae produced the least amount of mycotoxin (average 532 ppb per treatment); the other two species had similar toxin-producing capability (ca. 4800 ppb per treatment).

Previous work has shown that visual disease and DON contents may (Boyacioglu et al., 1992; Lemmens et al., 1997) or may not (Liu et al., 1997; Edwards et al., 2001) relate to each other. This inconsistency might be due to variability in competitiveness (Gang et al., 1998; Miedaner et al., 2000; Brennan et al., 2003) and toxin-producing capability (Gang et al., 1998; Miedaner et al., 2000; Chandler et al., 2003; Ludewig et al., 2005) between and within pathogen species in field samples . This kind of variability was eliminated in the present CE inoculations in which only a single isolate was used. A meta-analysis of previously published and unpublished data suggested that there is an overall positive correlation between FHB incidence/severity and DON accumulation but such a relationship is critically influenced by other factors (Paul et al., 2005).

FHB incidence in a mixed inoculation is expected to be greater than that from any single-isolate inoculations, assuming mutual independence between isolates and that diseased spikelets result only from initial infection. The expected overall incidence under the assumption of mutual independence was estimated for each co-inoculation, disregarding some differences in environmental conditions between single- and mixed-isolate inoculations. The predicted overall incidences were 38%, 46%, 37%, 57%, 51% and 58% for F. culmorum/F. poae, F. graminearum/F. poae, F. avenaceum/F. poae, F. graminearum/F. poae/F. avenaceum, F. avenaceum/F. poae/F. culmorum and F. graminearum/F. poae/F. culmorum, respectively. These predicted values based on the assumption of independence were all greater than the corresponding observed values (25%, 19%, 29%, 52%, 46% and 47%), particularly for the two-isolates co-inoculations. Thus, present results suggest that there is some degree of competitive interaction between isolates that influences the development of visual symptoms. Furthermore, the data did not indicate the presence of any synergistic interactions among the FHB pathogens.

Both diagnostic and competitive PCR results provide evidence for competitive interactions between isolates. In all inoculations with individual isolates, almost all grain samples contained quantifiable amounts of fungal DNA. However, in inoculations with a mixture of isolates, sufficient DNA for quantification was usually only present in one of the isolates. Furthermore, for several mixed inoculations it was not possible to detect fungal DNA for one or two of the isolates in many of the samples. Of all four isolates, the F. graminearum isolate was most competitive. In co-inoculations with other isolates, F. graminearum DNA was detected and quantified in almost all samples, whereas only a few samples had sufficient DNA of other isolates for quantification. Except for F. graminearum, the amount of fungal DNA was far less than that in the inoculation with individual isolates alone. For the majority of samples, the amount of F. graminearum DNA was similar in single- and mixed-isolate inoculations. In pair-wise inoculations, the F. poae isolate appeared to be slightly less competitive than the F. culmorum and F. avenaceum isolates, whilst F. culmorum appeared to be slightly more competitive than F. avenaceum (data not shown). When the three isolates were co-inoculated, F. culmorum appeared to be the most competitive. Recent studies have shown that the outcome of competition between F. culmorum, M. nivale and M. majus depends critically on temperature and the order of fungal establishment (Simpson et al., 2004). In the present study, interactions appeared to be broadly consistent over the environmental conditions tested.

Several studies on maize have also shown that F. graminearum is more competitive than other toxigenic species, including F. proliferatum, F. moniliforme (Velluti et al., 2000) and A. parasiticus (Etcheverry et al., 1998). Fusarium graminearum has gradually become more prevalent in cooler regions of Europe (Obst et al., 2002; Waalwijk et al., 2003; Xu et al., 2005). In regions where F. graminearum is present, incidence and severity of FHB are normally significantly greater when wheat follows maize than when wheat follows other crops (Dill-Macky & Jones, 2000; Cromey et al., 2002). Recently there has been an increased production of maize in rotation with wheat in Europe (Logrieco et al., 2002; Fox, 2004) and this is thought to have contributed to the increase in the predominance of this species. However, current findings suggest that F. graminearum might also be inherently more competitive on wheat spikes than other Fusarium species, and this may also be contributing to the increase in this species when it is adapting to the cooler regions.

Overall, mycotoxin accumulation was considerably greater following co-inoculations compared to the single-isolate inoculations. Only in one co-inoculation (F. culmorum and F. poae) was the overall level of the toxin (NIV) lower than the single-isolate inoculations with F. culmorum. Even for this co-inoculation, the productivity of mycotoxin per unit of fungal biomass was much greater (> 100 times) in the co-inoculation than in the single-isolate inoculation. In inoculations with F. poae only, the average NIV content of grain was only about 500 ppb whereas co-inoculations with F. avenaceum resulted in an average of more than 11 000 ppb of NIV even though the amount of F. poae DNA was less than in the single-isolate inoculations. Since F. avenaceum is not known to produce NIV, this increased NIV production is due to F. poae.

The mycotoxin productivity of F. graminearum in the co-inoculation with F. culmorum and F. poae (based on the 12 samples with quantifiable amounts of fungal DNA) was about 1550 times of that in the single-isolate inoculations. Even though the overall NIV production was reduced compared to the inoculation with F. culmorum alone, the NIV productivity was increased dramatically in the co-inoculations. Similarly, the average NIV concentration in grain following co-inoculation with F. poae, F. culmorum and F. avenaceum was nearly 12 000 ppb, compared to ca. 4900 ppb in the single-isolate inoculation of F. culmorum, even though the fungal biomass was much less in the grain from the co-inoculation. Several previous studies showed that mycotoxin production under competition may increase, decrease or remain at a similar level compared to no-competition (Ramakrishna et al., 1996; Etcheverry et al., 1998; Picco et al., 1999; Velluti et al., 2000; Valero et al., 2006). However, the mycotoxin productivity in these studies could not be estimated reliably because each individual fungal species was not estimated separately.

This apparent ‘synergistic’ effect on mycotoxin production may result from competition between isolates; toxigenic fungi may produce more toxins under stress, i.e. competition for resources. In addition, competition between pathogens in chaff tissues may lead to enhanced toxin production. Only the most aggressive isolate colonizes the grain, but toxin produced in the chaff may translocate to the grain as reported previously (Snijders & Krechting, 1992; Doohan et al., 1999). However, the increase in toxin accumulation was not universal, particularly for inoculations with isolates of different chemotype (DON producer –F. graminearum and NIV producer –F. poae and F. culmorum). In this case, only the amount of DON was increased (as F. graminearum is the most competitive) at the expense of NIV accumulation (however, NIV productivity per unit of fungal biomass was also greater in co-inoculations, but the amount of fungus was more greatly reduced). Thus the net result of mixed inoculation appears to be an increased production of toxins but the exact composition of the toxins in the grain depends on the relative competitiveness and the chemotype of the isolates.

This study demonstrated that FHB pathogens are likely to compete for nutrients or infection sites on wheat ears. The net results of this competition are decreased fungal biomass in grain (though not necessarily incidence of visual symptoms on floral tissues) and increased mycotoxin production. Disease development and mycotoxin production in field conditions are thus expected to be more complicated where more than one toxigenic species is usually present. In the present study, only a single isolate from each FHB toxigenic species was used; thus the results can be interpreted as competition between either individual species or isolates. Further research should be carried out to investigate whether such interactions differ significantly for within-species and between-species competitions.

Acknowledgements

This research was funded by the European Commission, Quality of Life and Management of Living Resources Programme (QOL), Key Action 5 on Sustainable Agriculture, Contract No. QLK5-CT-2000-01517 (RAMFIC). We thank Dr Simon Edwards of Harper Adams University College for providing the isolates.

Ancillary