Sensitivity of Zasmidium citri-griseum to Fenbuconazole and Control of Citrus Greasy Spot in Panama

: Citrus greasy spot, caused by Zasmidium citri-griseum ( = Mycosphaerella citri ) is the most prevalent fungal disease of citrus in Panama, causing premature defoliation and yield reduction. Fungicide applications are generally needed for the control of greasy spot. In this study, the sensitivity to fenbuconazole of 34 isolates of Z. citri-griseum from Panama was determined by calculating the effective concentration needed to reduce mycelial growth by 50% (EC 50 ). Two ﬁeld trials were conducted from 2011 to 2013, to evaluate the efﬁcacy of fenbuconazole to reduce disease severity and yield loss. The EC 50 values for fenbuconazole ranged from 0.09 to 7.62 µ g mL − 1 , with a mean EC 50 value of 2.66 ± 0.36 SE µ g mL − 1 for mycelial growth. These data can be used for monitoring sensitivity shifts in Z. citri-griseum to fenbuconazole and reduce risk of fungicide resistance in Panama. In the ﬁeld trials, sprays with fenbuconazole signiﬁcantly reduced ( p < 0.0001) the severity of greasy spot on leaves compared with the non-treated control. Greater disease control was obtained when three sprays of fenbuconazole were applied instead of one. Nevertheless, no signiﬁcant differences ( p > 0.05) were detected in yield.


Introduction
Citrus greasy spot, caused by Zasmidium citri-griseum (F.E. Fisher) U. Braun and Crous (=Mycosphaerella citri Whiteside), is one of the most important fungal diseases of citrus in the Caribbean Basin, including Florida (US), the Caribbean islands, Mexico, and Central America [1][2][3]. Greasy spot is widely distributed in Panama affecting mainly sweet orange [Citrus sinensis (L.) Osbeck] and Tahiti lime [C. latifolia (Yu.Tanaka) Tanaka] [4]. Trees affected by greasy spot typically show yellow mottle on the adaxial surface of leaves and yellow-brown spot, slightly raised pustules on the abaxial surface, resulting in premature defoliation and yield reduction. In Florida, fruit symptoms have been reported mainly in grapefruit (C. paradisi Macf.), consisting of minute black flecks on the peel which coalesce to form rind blotch [3]. In Florida, losses up to 45% have been reported in grapefruit and 25% in sweet orange [2]. In Cuba, yield losses up to 5 tons ha −1 were reported in sweet orange associated with greasy spot [5].
The main inoculum source of Z. citri-griseum are the ascospores formed in pseudothecia on decaying leaves on the orchard floor. Ascospores are discharged in response to wetting and dispersed by wind currents [6]. Once deposited on the leaf surface, ascospores germinate and the resulting mycelia grow epiphytically [7]. Conidia of Z. citri-griseum have also been reported, but they are considered of minor epidemiological relevance [2].The disease is characterized by a relatively long Table 1. Isolates of Zasmidium citri-griseum included in the sensitivity study for fenbuconazole with their corresponding values of the effective concentration needed to reduce mycelial growth by 50% (EC 50 ).

Isolate
Year A commercial formulation of fenbuconazole (Indar 25% w/v OF; Dow Agrosciences, US) was used in the mycelial growth inhibition assay of Z. citri-griseum. Appropriate volumes of fungicide were added to PDA at about 50 • C to obtain final concentrations of active ingredient (a.i.) of 0.01, 0.1, 1, 10 and 100 µg mL −1 . Non-amended PDA medium was used as the control. Amended and non-amended PDA (20 mL) was poured into 90-mm-diameter Petri dishes. Mycelial plugs, 5 mm in diameter, from the actively growing area of the fungal colonies were placed in the center of each plate. A sample size of four replicate plates was used for each combination of isolate and fungicide concentration. Plates were incubated at 25 • C in the dark for 30 days until the control plates were 50% covered with mycelium. Colony diameter was determined as the average of two perpendicular measurements, substracting the 5 mm of the mycelial plug. Growth inhibition (%) was calculated relative to the non-amended control for each isolate. Dose-response curves were obtained and the effective concentration needed to reduce mycelial growth by 50% (EC 50 ) was calculated with the nplr package for R [32]. This package is commonly used for n-parameter logistic regression models, in our case with four-parameter logistic regressions. The box-whisker plot and histogram of EC 50 values were obtained. The shape of frequency distribution was analyzed by examining curve shape, range, and mean values of EC 50 . The R version 3.2.5 was used in all statistical and graphical analyses [33].

Field Trials
To evaluate the efficacy of fenbuconazole for the control of greasy spot, two field experiments were conducted in commercial citrus orchards severely affected by the disease. Experiment 1 was located in a 'Star Ruby' grapefruit orchard at Tambo with 6-year-old trees grafted on 'Swingle' citrumelo [C. paradisi x Poncirus trifoliata (L.) Raf.] rootstock and with 5 x 10 m tree spacing. Experiment 2 was located in a 'Valencia' sweet orange orchard at Miraflores with 5-year-old trees grafted on 'Swingle' citrumelo and with 3.5 x 7 m tree spacing. Both orchards were rainfed. The same commercial formulation of fenbuconazole indicated for the mycelial growth inhibition assay was used. According to the label rate, a concentration of fenbuconazole of 0.12 g a.i. L −1 was compared with a non-treated control. Treatments were arranged in a completely randomized design with six replications of 10 trees each. A guard tree was located between plots within rows. Trees were sprayed with a volume of ∼ 6.5 L tree −1 in experiment 1 and ∼ 3 L tree −1 in experiment 2 using a hydraulic sprayer (Arbus 2000 Jacto, Brasil) at 300 psi. In 2011, spray applications were performed on 20 May, 7 June and 30 June in experiment 1 and 9 June, 1 July and 21 July in experiment 2. In order to evaluate a more cost-effective program, in 2012 only one spray application was carried out on 18 May in experiment 1 and 15 May in experiment 2. In 2013, one spray application was made on 21 May in experiment 1 and no fungicides were applied in experiment 2. In all cases, the first spray coincided with the appearance of new shoots at phenological stage BBCH 32-39 (between 20% and 90% of final shoot length) [34].
Disease severity was assessed in December 2011-2013 in experiment 1 and December 2011-2012 in experiment 2. Ten shoots were arbitrarily selected in each tree (∼ 70 leaves tree −1 ) and tagged before the first spray. All leaves in the tagged shoots were assessed according to the following rating scale: 0 = no lesions observed; 1 = 1-10 leaf spots; 2 = 11-20 leaf spots; 3 = more than 20 leaf spots per leaf [31]. Descriptive statistics were obtained for disease severity categories in each experiment. Disease severity categories were treated as an order factor, which was modelled with a proportional odds logistic regression model using the polr function of the MASS package for R [35].
where P(Y i ≤ j ) is the cumulative probability between 0 and 1 for category j of Yi, j = 0, . . . , 2 are categories of greasy spot severity, α j0 are the thresholds for greasy spot severity categories, β tr , is the coefficient for the fungicide treatment evaluated (I tr ), and n is the total number of ten-tree plots. The proportional odds model assumes that the logit of the cumulative probabilities changes linearly as the explanatory variables change, and also that the slope of this relationship is the same regardless of the disease severity category [36]. In our case, the non-treated control was used as the reference level and the odds ratio for the fungicide treatment was calculated as e −β based on the cumulative probabilities. Owing to the proportional odds assumption, the odds ratio for the fungicide treatment stays the same no matter how disease severity is dichotomized into two levels. This proportional odds assumption was evaluated by the Likelihood Ratio Test (LRT) between the proportional odds model and the non proportional odds model [37] using the VGAM package for R [38]. The goodness of fit was assessed by a χ 2 -test of the residual deviance against the null model (i.e., including only the intercept).
A significant effect for the fungicide treatment was considered when the 95% confidence interval of the odds ratio did not overlap the null value of 1 [36]. All fruit in each tree were harvested and weighed. In each experiment, descriptive statistics and analysis of variance (ANOVA) were performed for yield data (kg tree 1 ) using the stats package for R [39]. Prior to ANOVA the normality assumptions were verified by the test by Shapiro-Wilk and the homogeneity of variance using the Bartlett test. A diagnosis of the studied residues was made applying the test of Breusch-Pagan to determine the homocedasticity of the variance.

Mycelium Growth Inhibition Assay
The EC 50 values of 34 isolates of Z. citri-griseum collected from different provinces of Panama exposed to fenbuconazole ranged from 0.09 to 7.62 µg mL −1 (Table 1), with a mean ± standard error (SE) of 2.66 ± 0.36 µg mL −1 . The box-whisker plot ( Figure 1) showed a median value of 1.98 µg mL −1 with a interquartile range from 1.11 to 4.04 µg mL −1 and no outliers were observed. The frequency distribution of EC 50 to fenbuconazole of the 34 isolates of Z. citri-griseum included in the assay showed a unimodal curve with a positive skew ( Figure 1). The frequency of isolates with EC 50 values <1 µg mL −1 represented 20.6% (n = 7). For EC 50 from 1 to 5 µg mL −1 the frequency was 64.7% (n = 22) and for EC 50 values >5 µg mL −1 represented 14.7% (n = 5).

Field Trials
In experiment 1 in 2011, 57.22% of the leaves were in the severity category 0 (no lesions) in the trees treated with three fenbuconazole sprays while only 3.95% in the non-treated control ( Figure 2). The non-treated control showed 81.39% of the leaves in the severity category 3 (>20 leaf spots) in contrast to the 10.82% in the trees treated with fenbuconazole. In 2012 and 2013, the percentages of leaves in the severity category 0 in the trees with one spray of fenbuconazole decreased to 5.83% and 3.14%, respectively. The percentage of leaves in category 0 in the non-treated control was 0.54% in 2012 and 0.27% in 2013. Values for category 3 in the non-treated control were 74.96% in 2012 and 74.26% in 2013. In the trees with one spray of fenbuconazole, the percentage of leaves in category 3 was 33.38% and 33.15% in 2012 and 2013, respectively.
In experiment 2 in 2011, the percentage of leaves in the severity category 0 (no lesions) in the trees with three sprays of fenbuconazole was 51.55% whereas only 2.62% in the non-treated control ( Figure 3). The percentage of leaves in the severity category 3 (>20 leaf spots) was 14.66% in the trees treated with fenbuconazole and 64.31% in the non-treated control. In 2012, the percentage of leaves in category 0 decreased to 8.34% in the trees treated with one spray of fenbuconazole and 0.82% in the non-treated control. On the contrary, the percentage of leaves in category 3 increased to 42% in trees treated with fenbuconazole and 66.12% in the non-treated control.
In experiment 1, the proportional odds logistic regression model applied to the disease severity categories showed a significant effect (p < 0.0001) of the factor treatment in the three years of study (2011,2012 and 2013) ( Table 2). The negative values of the estimate for the treatment factor in the years 2011 (−3.55), 2012 (−1.79) and 2013 (−1.80) indicated a higher probability of obtaining lower values of disease severity in trees treated with fenbuconazole. In experiment 2, the results of the proportional odds logistic regression model also showed a significant effect (p < 0.0001) of the factor treatment in the two years of study (Table 3)  Odds ratio and 95% confidence interval values for the trees treated with fenbuconazole in experiment 1 are shown in Table 4. As it is observed, for non-treated trees, the odds of presenting higher disease severity were 34.81, 5.99 and 6.05 times than for treated trees in 2011, 2012 and 2013, respectively. Fenbuconazole applied 1 or 3 times reduced disease severity compared with the non-treated control in all years as the 95% confidence interval of the odds ratio did not overlap the null value of 1. The predicted probabilities for disease severity categories in the trees treated with fenbuconazole and the non-treated control in experiment 1 are shown in Table 4. In 2011, a high probability of occurrence of severity category 0 (no lesions) was observed in the trees treated with three sprays of fenbuconazole (0.57), in contrast with the low probability (0.04) of occurrence of this severity category in the non-treated control. On the other hand, a high probability (0.81) of occurrence of severity category 3 (>20 leaf spots) was obtained for the non-treated control compared with a low probability (0.11) for the trees treated with fenbuconazole. In 2012 and 2013, the predicted probability for the severity category 3 in the non-treated control was higher (0.74) than in the trees treated with one spray of fenbuconazole (0.33).
Odds ratio and 95% confidence interval values for the trees treated with fenbuconazole in experiment 2 are shown in Table 4. As it is observed, for non-treated trees, the odds of presenting higher disease severity were 14.88 and 2.86 times than for treated trees in 2011 and 2012, respectively. Fenbuconazole applied 1 or 3 times reduced disease severity compared with the non-treated control in all years as the 95% confidence interval of the odds ratio did not overlap the null value of 1. The predicted probabilities for disease severity categories in the trees treated with fenbuconazole and the non-treated control in experiment 2 are shown in Table 4. In 2011, the trees with three sprays of fenbuconazole showed a high probability (0.50) of occurrence for the severity category 0 (no lesions) in contrast with the non-treated control that showed a very low probability (0.06). On the other hand, the non-treated control showed a high probability (0.66) of occurrence of severity category 3 (>20 leaf spots) compared with the trees treated with fenbuconazole that showed a low probability (0.11) for this same severity category. In 2012, the predicted probability for the severity category 3 in the non-treated control was still high (0.67) compared with the trees with one spray of fenbuconazole (0.41) ( Table 4).      Yield data (kg tree −1 ) from experiments 1 and 2 were normally distributed (p > 0.05) in all years, with homogeneous variances (p > 0.05). In experiment 1, no significant differences in yield were detected between the grapefruit trees treated with fenbuconazole and the non-
The reduced sensitivity of Z. citri-griseum isolates to fenbuconazole detected in our study suggest that the isolates tested were previously exposed to this fungicide and/or other DMIs. Corio-Costet [41] indicated that the resistance to one DMI fungicide can confer positive cross-resistance to other molecules in the same group. Nevertheless, to elucidate the presence of resistance to DMIs in Z. citri-griseum populations in Panama, it would be necessary to detect point mutations or overexpression of the C14α-demethylase (CYP51) by molecular analysis [16]. Another important aspect to consider when interpreting our results is the sample size used to determine the EC 50 . Franke et al. [42] indicated that is important to define the appropriate sample size to detect differences in fungicide sensitivity. In our study, a total of 34 isolates of Z. citri-griseum from different geographic areas of Panama were collected to calculate the EC 50 . However, in the study carried out by Mondal et al. [25] only five isolates were evaluated. The greater variability in EC 50 values of Z. citri-griseum to fenbuconazole detected in our study might be associated with the greater sample size used, capturing differences associated with the origin of the isolates and potential previous exposure to DMIs. Actually, Franke et al. [42] indicated that the minimum sample size depends on the variation of the EC 50 values, as a number of individuals less sensitive to the fungicides could significantly change the variance. Our results provide important data for monitoring sensitivity shifts in Z. citri-griseum to fenbuconazole and reduce the risk of fungicide resistance in Panama.
The efficacy of fenbuconazole to reduce the severity of foliar symptoms of greasy spot was evaluated on grapefruit (experiment 1) and sweet orange (experiment 2). In 2011, three sprays with fenbuconazole were applied from May to July resulting in lower disease incidence and severity than the non-treated controls. The reduction of greasy spot severity was illustrated by the 95% confidence intervals of the odds ratios in experiment 1 (30.75-39.47) and experiment 2 (13.21-16.85), not overlapping the null value of 1 in either case. Our results are in line with those obtained in Costa Rica, where two sprays with copper compounds applied in June and August coinciding with the first leaf flush of 'Valencia' sweet orange significantly reduced the incidence and severity of greasy spot and subsequent defoliation [1]. Similar results were reported by Timmer et al. [2] in experiments carried out in Florida on 'Marsh White' and 'Ruby Red' grapefruits with two copper applications in June and August during the spring flush. Mondal and Timmer [21] indicated that one or two fenbuconazole applications before the development of epiphytic mycelium in July completely controlled greasy spot in the spring flush. In our study, the first fenbuconazole spray was applied at the beginning of shoot emergence in May, coinciding with the onset of the infection period of Z. citri-griseum in Panama [43].
When a single spray of fenbuconazole was applied in 2012 and 2013 in experiment 1 and 2012 in experiment 2, disease incidence doubled that of 2011 when three sprays were applied. The 95% confidence intervals of the odds ratios for the experiment 1 in 2012 (5.50-6.55) and 2013 (5.47-6.70) were lower than those obtained in 2011, but did not overlap the null value of 1 in either case. Similar results were obtained in experiment 2, with a 95% confidence interval for the odds ratio in 2012 (2.85-3.16) lower than in 2011. The more effective disease control achieved in 2011 was likely due to the two additional sprays applied during the infection period, which probably inhibited the epiphytic growth of the pathogen for a longer period of time.
Although with lower efficacy, disease severity was significantly reduced with a single spray of fenbuconazole compared with the non-treated control. Our results are in line with those by Timmer et al. [2], who reported effective control of greasy spot with a single application of copper applied in June in three experiments carried out in Florida on grapefruit. Similarly, other experiments on grapefruit in Florida also reported reductions in greasy spot severity with a single spray of fenbuconazole applied in June or July [18][19][20]. Actually, a single spray with petroleum oil applied in mixture with copper from mid-May to June is recommended for greasy spot control in processing sweet oranges in Florida [44]. For processing grapefruits, two sprays are recommended instead [44]. QoI and DMI fungicides, including fenbuconazole, are also recommended for greasy spot control with or without petroleum oil [45]. In the case of fenbuconazole, it provides effective control of the disease on both leaves and fruit [44]. In our case, no rind blotch symptoms on fruit were observed in the field trials, nor in our previous surveys [4,31].
Despite the significant reduction of greasy spot severity on leaves in the trees treated with fenbuconazole, particularly in 2011 when three sprays were applied, no significant differences in yield were found in any of the experiments. Timmer et al. [46] reported similar results in Texas, where several fungicide spray programs were evaluated for greasy spot control on grapefruit during 4 consecutive years. Despite greasy spot severity on leaves and rind blotch incidence on fruit were significantly reduced, yields (kg tree −1 ) were not significantly improved. Likewise, Showler et al. [11] reported in Texas that sprays with aqueous organic mixtures applied for three consecutive years reduced disease incidence in grapefruit, but differences in yield (No. fruit tree −1 ) were only detected the first year. In our case, experiments for greasy spot control should probably have covered a longer period of time, allowing the treated trees to completely recover their leaf area and thus have an impact on yields. Furthermore, the higher control efficacy observed when three fenbuconazole sprays were applied instead of one indicates that intensive spray program would be probably needed under the rainfed conditions of Panama. In fact, in Florida two applications are needed when severe defoliation due to greasy spot occurred in the previous year [44]. In those cases, the first spray should be applied from mid-May to June and the second soon after the major summer flush has expanded [44]. In addition to the disease itself, it is important to consider the concurrence of other limiting agronomic factors. Rainfed conditions and typical crop management practices in Panama may result in nutritional deficiencies and seasonal water stress (January to March), which could seriously limit citrus yields and thus masking the effects of disease control.