Evaluation of fungicides for management of boxwood blight caused by Calonectria spp. under field conditions in Northern Germany

Fungicide protection is the current approach for management of boxwood blight caused by Calonectria pseudonaviculata (Cps) and C. henricotiae (Che). However, published studies evaluating fungicides under field conditions have been focused on Cps. The objective of this study was to evaluate fungicides in Northern Germany where both Cps and Che were present. Three trials were conducted between 2006 and 2016. In 2006, plants were artificially inoculated with a conidial suspension, while infested soil and plant debris were added to a different field as inoculum for the 2012 trial and this field was used again without further addition of inoculum in 2015. Fungicides were applied one to five times and assessments were done three to six times, depending upon the trial. The highest level of disease severity occurred in 2015 (0.91), while the lowest occurred in 2012 (0.01). Among the fungicides evaluated in 2006, preventive sprays of Cercobin FL, Switch, Harvesan, Pugil 75 WG, Dithane NeoTec and Euparen M WG were most effective, with blight control above 65%. In 2012, all treatments including Askon, Cabrio Top, Malvin WG, Dithane NeoTec and Osiris showed > 75% blight control. In 2015, Bayer Rosen-Pilzfrei Baymat and Switch were the most effective (> 82%). Extended in-season blight control was also observed with some fungicides. Additionally, a few fungicides that were evaluated in more than 1 year showed reduced effectiveness over time. This study filled several major knowledge gaps especially regarding fungicide efficacy against Che under field conditions and thus provides crucial information for developing chemical control strategies.

Boxwood blight caused by Cps has spread throughout Europe (Palmer and Shishkoff 2014;LeBlanc et al. 2018;Daughtrey 2019), parts of Asia (Akilli et al. 2012;Mirabolfathy et al. 2013), New Zealand (Ridley 1998) and North America (Ivors et al. 2012;Elmhirst et al. 2013), jeopardizing the supply chain of a major evergreen ornamental plant (Hall et al. 2021). In the USA alone, this disease has spread to over 30 states, putting at risk more than 90% of the nation's boxwood production (Hall et al. 2021). This pathogen threatens boxwood crops, plantings and forests globally (Barker et al. 2022).
The other species, Che, has been reported for a few European countries, including Germany, Belgium, the UK, Slovenia, The Netherlands (Gehesquière 2014) and The Czech Republic (Bartíková et al. 2020). It is likely that Che could spread and establish outside Europe. This pathogen may pose a more severe threat to the global boxwood industry if it spreads to other countries, due to its greater tolerance of high temperature and reduced sensitivity to some of the most efficacious fungicide groups (Gehesquière et al. 2016).
The current management approach largely depends upon the use of tolerant cultivars and fungicide applications. Although a handful of cultivars are less susceptible to boxwood blight, no cultivar is completely resistant to the disease (Brand et al. 2022;Kramer et al. 2020;LaMondia and Shishkoff 2017;Yoder et al. 2022). In order to maintain a broad spectrum of economically important Buxus species, varieties and cultivars, disease management is heavily dependent upon chemical control strategies.
While fungicide evaluation studies against Cps have been ample, studies pertaining to its sister species, Che, have been limited. Several fungicides tested in vitro, alone or in combinations, have been shown to work against different stages of the Cps life cycle (Brand 2006;Henricot et al. 2008;LaMondia 2014). Brand (2006) found that prochloraz, propiconazole, thiophanate-methyl and carbendazim + flusilazole inhibited mycelial growth, whereas tolylfluanid, mancozeb, chlorothalonil and fludioxonil + cyprodinil were effective against conidial germination. Later, Henricot et al. (2008) reported that prochloraz, carbendazim, kresoxim-methyl and epoxiconazole + pyraclostrobin + kresoxim-methyl were effective in inhibiting mycelial growth while azoxystrobin, chlorothalonil, kresoxim-methyl, mancozeb, boscalid + pyraclostrobin, epoxiconazole + pyraclostrobin and epoxiconazole + pyraclostrobin + kresoxim-methyl limited conidial germination. Similarly, LaMondia (2014) concluded that propiconazole, triflumizole, tebuconazole and fludioxonil + cyprodinil were effective against mycelial growth but not conidial germination, highlighting efficacy against Cps under in vitro conditions. One field study conducted in the UK found propiconazole + pyraclostrobin, myclobutanil + fludioxonil, kresoxim-methyl and chlorothalonil to have greater efficacy against Cps (Henricot and Wedgwood 2013). The major gap in knowledge is with regard to fungicide efficacy against Che under field conditions. Considering its lower sensitivity to tetraconazole and kresoxim-methyl compared to Cps observed in vitro and in vivo (Gehesquière et al. 2016), evaluating fungicides against Che under field conditions is long overdue.
The objective of this study was to fill this knowledge gap by evaluating fungicides with 19 different active ingredients from 9 different Fungicide Resistance Action Committee (FRAC) groups in fields where Cps and Che were both present (Brand et al. 2022). Results from the study will enable growers and landscapers to better manage boxwood blight.

Materials and methods
All field trials were conducted at the Research and Teaching Institute for Horticulture Bad Zwischenahn, Germany. Buxus sempervirens 'Suffruticosa' was used as the test plant in all trials due to its high susceptibility to boxwood blight. Three trials were performed in 2006, 2012 and 2016 (Table 1). Each trial started with a new planting of about 2-year-old Buxus sempervirens 'Suffruticosa', propagated by cuttings with a height of about 10 cm in 0.5 l pots. Planting dates are listed in Table 2. Each treatment had three (2006) or four (2012 and 2015) replicate rows with ten plants per row (with 15 cm planting distance in the row and 50 cm between rows, resulting in a row length of 1.5 m and a plot size of 0.75 m 2 ), and they were arranged in a randomized complete block (RCB) design within each trial.

Fungicide efficacy field trials-inoculation and treatments
Plants of the 2006 trial were inoculated on 14 September by spraying a conidial suspension of Cylindrocladium pseudonaviculatum at 10 4 -10 5 conidia/ml until runoff (150-200 mL/ m 2 ) using a handheld pressure sprayer. After inoculation, the entire field was covered for 36 h with white polyethylene film to assure sufficient leaf wetness duration for infection. The 2006 trial evaluated nine fungicides (Table 1) and compared to a nontreated control. All test fungicides were applied once before inoculation to determine their performance as protectants, while seven were also applied once 2 days after inoculation to determine their curative effects ( Table 2). All fungicide applications were made with a backpack sprayer with a water volume of 100 mL/m 2 at the application rates mentioned in Table 1. At the end of the trial, fallen leaves and infected plants were shredded and incorporated into the topsoil to acquire an area contaminated with the pathogen for succeeding trials (Brand et al. 2022).
The 2012 trial was conducted in a nearby field, where plants were inoculated by scattering infested soil and fallen leaves collected from an infested field where blight trials had been performed since 2006 (Brand et al. 2022). Inoculum was added to the trial field on August 20, 2012, immediately after the first treatment with fungicides, earlier the same day (Table 2). This trial included five fungicides (Table 1) and a nontreated control. Each fungicide was applied to the same plants twice. Due to slow disease development, the field was sprayed with water and covered with white polyethylene film from 9 to 10 September to improve infection conditions.
The 2015 trial was done in the same field as the 2012 trial evaluating six fungicides (Table 1) compared to a nontreated control. No inoculation was made, but inoculum was present in organic debris from earlier studies. Each fungicide was applied five times at 2-week interval starting 15 July (Table 2).

Disease assessment
Disease severity in each replicate row was assessed by estimating percentage of area with leaf spots and fallen leaves, recorded as an estimate of the percentage of total leaf area affected. Later, mean severity for each replicate row was calculated and expressed as proportion ranging between 0 (exclusive) and 1 (exclusive). Three assessments were made in 2006 and 2012, while six were made in 2015 (Table 2).

Comparative analysis of fungicides evaluated in multiple years
To strengthen the comparative analysis of fungicide performance across several years, results from two additional trials that are presented in supplemental Figs 4 and 5 were included in the analyses. In 2015, a separate field trial (referred to as 2015_a) was conducted evaluating three fungicide products: Delan Pro (dithianon + potassium phosphate), Geoxe (fludioxonil) and Switch (fludioxonil + cyprodinil). Planting for the 2015,_a trial was done on 22 August and each fungicide was applied twice on 21 August and 31 August; three disease assessments were made. An additional trial was done in 2016 evaluating Delan Pro (dithianon + potassium phosphate), Sunjet Flora (azoxystrobin + isopyrazam), Dagonis (fluxapyroxad + difenoconazole), Ceriax (fluxapyroxad + epoxiconazole + pyraclostrobin), Switch (fludioxonil + cyprodinil) and Geoxe (fludioxonil). Planting was done on 19 August with two treatment applications (18 August and 1 September) and two disease assessments (23 September and 10 October).

Data analysis
All disease severity data were analyzed using proc GLIM-MIX in SAS (v9.4; Statistical Analysis Software, Cary, NC). Analysis of variance was used to test for the overall effect of treatment, disease assessment time and their interactions on disease severity with a 'beta' distribution specified in the model statement. Due to presence of 0 s (zeros) and 1 s (ones) in the disease severity data, severity values were adjusted using the following equation (Smithson and Verkuilen 2006): where y′ is the new disease severity value; y is the original disease severity; and N is the number of observations or the sample size. This resulted in the 0 s being a very small decimal and the 1 s being very close to 1. The effect of the fungicides on disease severity for each assessment date was modeled in a similar manner as described above with treatment as fixed and replicate as random effect. Treatment LSMEANS were estimated using the maximum likelihood approach, while differences among treatments were compared using t-grouping. Later, data from the model scale were converted to the original scale using an 'ilink' option.
Fungicide efficacy of each treatment was calculated as described by Paul et al. (2007): The higher the percentage, the better the treatment efficacy. Treatment efficacy among assessment dates within a given year was compared using t-grouping. A nonparametric student's t test was used to compare efficacy of fungicides tested across trials/years where applicable.

Significance of treatment and assessment time
Boxwood blight was observed in all field trials, with the highest blight severity in the nontreated control ranging from 0.16 in 2012 to 0.91 in 2015. There was a significant effect of treatment and disease assessment time on blight severity in all trials (Table 3). Similarly, the treatment × time effect was significant for all years except for 2006.

Fungicide performance
Fungicide performance close to 2 weeks after the last treatment is presented here, while data are provided in supplemental materials.

The 2006 trial
A clear distinction in the proportion of blighted leaves was seen among protective and curative sprays (Fig. 1). Six fungicides applied protectively, including Cercobin FL (thiophanate-methyl), Switch (fludioxonil + cyprodinil), Harvesan (carbendazim + flusilazole), Pugil 75 WG (chlorothalonil), Dithane NeoTec (mancozeb) and Euparen M WG (tolylfluanid), were the most effective, with the proportion of blighted %Efficacy = Severity of untreated control − Severity of treatment Severity of untreated control × 100 leaves in plants below 0.01. In contrast, the proportion of blighted leaves in the control and the corresponding curative treatments was at least twice as high. Percentage blight control achieved by those 6 preventive sprays remained above 65% even 18 days after the treatment application (Table 4). Curative sprays of Mirage 45 EC (prochloraz) and Ortiva (azoxystrobin) were the least effective, with blight control ranging between 17 and 19%. Tilt 250 EC (propiconazole) was not effective as a curative or a protectant.

The 2015 trial
Of the six treatments tested, Bayer Rosen-Pilzfrei Baymat (tebuconazole) and Switch (fludioxonil + cyprodinil) were ; the other two are presented as supplemental materials. Each column represents the mean of three replicates and is topped with a standard error bar. Columns topped with a shared letter(s) did not differ according to t-grouping at α = 0.05 the most effective, with the proportion of blighted leaves below 0.20, while the proportion of blighted leaves in the nontreated control remained above 0.85 (Fig. 3). The corresponding efficacy 18 days after the last treatment was 82 and 87%, respectively (Table 4). Products that were least effective with blight proportion above 0.60 included Duaxo Universal Pilzfrei (difenoconazole), Pilzfrei Ectivo (myclobutanil), Polyram WG (metiram) and Ortiva (azoxystrobin).

Extended efficacy after the last treatment
While maximal fungicide protection was observed within the first 2 weeks of the last treatment, most products retained their efficacy throughout the trial. For example, in 2006, efficacy of preventive application of Cercobin FL, Switch, Harvesan, Pugil 75 WG, Dithane NeoTec and Euparen M WG increased from 18 to 38 days after the last treatment, with blight control ranging between 87 and 90% (Fig. 4). While blight control efficacy of Euparen M WG and Dithane Neo-Tec was reduced significantly 71 days after the last treatment application, the percentage blight control of Cercobin FL, Switch, Harvesan and Pugil 75 WG did not change (Fig. 4). Similarly, in 2012, while percentage blight control of Osiris and Cabrio Top increased significantly from 7 to 17 days after the last treatment, their efficacy did not alter 25 days from the last treatment (Fig. 5). Other fungicides, including Askon, Malvin WG and Dithane NeoTec, provided a consistent blight control throughout the trial, ranging between 47 and 84% (Fig. 5).
In 2015, performance of Bayer Rosen-Pilzfrei Baymat and Switch increased significantly from the first to the second disease assessment time with blight control ranging between 79 and 83%. These products continued to provide blight protection until the final assessment (Fig. 6).

Discussion
This study provided field-based evidence that a number of fungicides were effective against both pathogens, Cps and Che, under varying levels of disease pressure in northern Germany. Among the most effective products with single chemistry were Malvin WG (captan), Bayer Rosen-Pilzfrei Baymat (tebuconazole) and prophylactic sprays of Pugil 75 WG (chlorothalonil), Cercobin FL (thiophanate-methyl), Dithane NeoTec (mancozeb) or Euparen M WG (tolylfluanid). Likewise, effective products with two active ingredients Fig. 2 Proportion of blighted leaves of Buxus sempervirens 'Suffruticosa' with and without fungicide treatments in 2012. Shown is the second of three assessments (10 September, 20 September, 28 September), the other two are presented as supplemental materials. Each column represents the mean of four replicates and is topped with a standard error bar. Columns topped with a shared letter did not differ according to t-grouping at α = 0.05 Fig. 3 Proportion of blighted leaves of Buxus sempervirens 'Suffruticosa' with and without fungicide treatments in 2015. Shown is the fourth of six assessments (12 August,25 August,8 September,25 September,9 October and 30 October); the other five are presented as supplemental materials. Each column represents the mean of three replicates and is topped with a standard error bar. Columns topped with a shared letter(s) did not differ according to t-grouping at α = 0.05 included Askon (difenoconazole + azoxystrobin), Cabrio Top (metiram + pyraclostrobin), Osiris (epoxiconazole + metconazole), Switch (fludioxonil + cyprodinil) and Harvesan (carbendazim + flusilazole). These active ingredients belong to FRAC groups 1, 3, 9, 11, 12, M 03, M 04, M 05 and M 06, offering a variety of options for scheduled spray programs against boxwood blight and for product rotation to slow fungicide resistance development. This study also demonstrated extended in-season blight control of some fungicide products. Additionally, it showed decreasing performance of some fungicides over time.
The study confirmed the efficacy of fungicide products previously evaluated against the blight pathogens. For example, chlorothalonil and cyprodinil + fludioxonil that were earlier shown to provide control against Cps (Cinquerrui et al. 2017;Henricot and Wedgwood 2013) were among the most effective products in our study. Cyprodinil + fludioxonil (Switch) treatment was significantly better than controls when tested over multiple years under our field conditions. More importantly, this study expanded our understanding of product performance from Cps to Che. For example, azoxystrobin and propiconazole, reported to be effective against Cps (Henricot et al. 2008;LaMondia 2014), were ineffective against the combination of the two blight pathogens in our study, which is in good accordance with the in vitro study of Gehesquière et al. (2016). Another active substance from the azoles (FRAC Group 3), myclobutanil (Pilzfrei Ectivo), also had limited effectiveness in our study. This might be due to the dominance of Che at the research site (Brand et al. 2022). These results highlight the importance of monitoring for the presence or dominance of Che when selecting fungicides for blight mitigation.
Our findings demonstrated that preventive applications were more effective than curative applications. In 2006, preventive sprays were significantly more effective than curative treatments of the same fungicides. In subsequent trials, the first spray was made prophylactically followed by succeeding sprays, making it harder to make comparisons between preventive and curative approaches. However, our results reflect that fungicides with greater activity are best applied prophylactically prior to infection. This result is in agreement with the observation of LaMondia (2015) that fungicide products containing propiconazole, thiophanate-methyl, pyraclostrobin, fludioxonil and kresoxim-methyl, applied as protectants, had the best efficacy. In practice, however, it is very difficult to apply treatments immediately prior to infection because predictive models of infection and microclimatic influences on infection-especially temperature and humidity as seen in other pathosystems (Chaulagain et al. 2020;De Wolf et al. 2003)-are not always available and applicators are not always able to make treatments on short notice. As a result, periodic treatment with fungicides currently is a common practice in nurseries and landscapes.
Our preventive + curative treatment applications appear to have worked even better to limit blight spread under higher disease pressure condition. For example, in 2015, five applications of tebuconazole and fludioxonil + cyprodinil, applied at 2-week intervals, were sufficient to control blight severity by 50-87%, when maximum blight severity in that year was very high (severity proportion > 0.85). The preventive-only treatments, including Opera (pyraclostrobin + epoxiconazole) and Bravo (chlorothalonil), however, were not as effective against boxwood blight under high infection pressure as they were under low infection pressure (Henricot and Wedgwood 2013). Besides disease pressure and treatment time, other factors such as a fungicide's mode of action, application method, pathogen populations and their fungicide sensitivities, infection site, dose, active ingredient and the formulation of the fungicide, in addition to weather, may play an important role in determining the extent of fungicide efficacy.
The observation of extended in-season performance of some fungicides was not expected, but it is intriguing and potentially of practical importance. It raised a set of important questions. Are the recommended treatment intervals set too short for these products? What additional studies are needed to re-evaluate current recommendations on treatment interval? And how might the extended performance observed be related to the number of treatment times? Any new recommendations on extending treatment intervals would reduce the number of treatments required per growing season, saving growers both chemical and labor cost and curtailing their environmental footprint. If extended treatment interval were to come with a cost of faster development of fungicide resistance in the pest population, however, the temporary savings would not be worth the loss of effective active ingredients.
While not all fungicides were tested each year, a few fungicides including Ortiva, Switch, Askon, Delan Pro, Geoxe and Dithane NeoTec were evaluated for at least 2 years (Supplemental Fig. 6). We noted that performance of these fungicides was reduced significantly in later years compared to their initial trial years. This was most striking in strobilurin and azole groups: efficacy of Ortiva (azoxystrobin) decreased by 51 percentage points from 2006 to 2015_a and that of Askon (difenoconazole + azoxystrobin) reduced by 98 percentage points from 2012 to 2016. We speculate that this reduced sensitivity could be a result of a shift in pathogen population structure from Cps to Che: for the first inoculation in 2006, an isolate of Cy. pseudonaviculatum (syn. Cy. buxicola) was used, while later, in 2016, Che was detected in our field and largely predominated by 2021 (Brand et al. 2022). Our results validated the in vitro and in vivo observation on Che made by Gehesquière et al. (2016), suggesting its risk of developing fungicide resistance as noted by Maurer et al. (2017). A continuous evaluation of fungicide products involving similar active ingredients over the period since Cps and Che were first introduced to Europe could have better presented the occurrence and evolution of fungicide resistance, but our results suggest changes in fungicide susceptibility over time. To properly use and refine best management practices, it is crucial to follow research-based fungicide use recommendations, including rotation of products with different modes of action, and to monitor shifts in fungicide sensitivity.
Some of the fungicides that show great efficacy in our studies are no longer available in Europe or the USA, but may remain on the market elsewhere. For example, thiophanatemethyl, carbendazim, flusilazole, chlorothalonil, mancozeb, tolylfluanid and prochloraz are no longer approved in the European Union (European Commission 2022); therefore, no corresponding fungicides are registered in Germany (Bundesamt für Verbraucherschutz und Lebensmittelsicherheit, 2022). According to our results, among the fungicides currently registered for use in nurseries, only fludioxonil + cyprodinil (Switch) is effective. Only two fungicides are registered for use in private gardens against leaf spot pathogens or boxwood blight, of which only Bayer Garten Rosen-Pilzfrei Baymat (tebuconazole) showed sufficient blight control. The same product is also approved for public greens, but application in larger boxwood plantings is hardly practical because it is available only in very small and comparatively expensive packages. Other fungicides approved for public greens (Duaxo Universal Pilzspritzmittel, Ortiva) did not suppress disease effectively in this study. From the American perspective, products such as flusilazole + carbendazim, flusilazole, tolylfluanid, metiram and metiram + pyraclostrobin are not registered for use by the United States Environmental Protection Agency (2022). In the absence of effective registered fungicides, alternative products with relatively low toxicity will be needed to minimize blight development, particularly if Che with its smaller spectrum of effective chemistries is introduced to North America. However, the effect of materials with lower risk potential for environment and health on the blight pathogens remains to be studied.