For all experiments, single colonies of B91 and Y126 were grown in liquid media - B91 in LB broth (ThermoFisher) and Y126 in potato dextrose broth - for 24 hours on a shaking incubator (180 rpm, 25°C). The concentration was measured using a spectrophotometer and adjusted to OD600 0.2 (B91) and 0.01 (Y126), equivalent to 1x10⁷ CFU/ml for inoculation. This concentration was determined from a preliminary experiment where B91 and Y126 were applied at four doses (10⁸, 10⁷, 10⁶, 10⁵ CFU/ml) onto cherries that were inoculated with M. laxa 24 h later. The biocontrol suspensions, along with fungicide or the sterile distilled water control, were applied to mummified fruit or blossoms using handheld sprayers. To produce M. laxa inoculum for inoculum, a M. laxa isolate from the East Malling collection was first inoculated onto wounded ripe plums and then spores were collected by washing the sporulating fruit 2 weeks after. A haemocytometer was used to adjust the M. laxa concentration to 1x10⁵ conidia/ml; M. laxa inoculum was applied similarly to the biocontrol suspensions via a hand-held sprayer.

In 2017, mummified damsons from Damson plums of an unknown cultivar from the University of Reading were used, whereas in 2018 mummies of the cherry variety “Kordia” were collected from a non-commercial orchard at East Malling, Kent, England. The fruits were kept in a cold store at 4°C until needed. Single colonies of B91 and Y126 were grown in liquid media for 24 hours on a shaking incubator (180 rpm, 25°C). A commercial fungicide, Luna^® Sensation (250 g/L fluopyram + 250 g/L trifloxystrobin), was prepared according to the manufacturer’s instructions. Mummified fruits were sprayed until runoff with a handheld sprayer, within polythene bags to avoid cross-contamination between treatments. For biocontrol experiments, each mummified fruit was estimated to have received ca. 4 x 10⁷ CFU for the biocontrol microbe. In addition, a control treatment was included where mummified fruits were sprayed with distilled water.

Treated fruits were placed in green mesh (ca. 0.5 cm) bags (10 fruits per bag) and placed in an orchard at East Malling. For each treatment, treated fruits were sampled at six time points ( Table 1 ) to quantify viable population sizes of B91, Y126 and M. laxa. Four cherry trees in a non-commercial orchard were used, each tree as an experimental block; mesh bags of different treatments were hung on branches at similar heights. On each tree, there were 24 mesh bags (six bags for each treatment, one for each sampling time point). Immediately after removal from the orchard, all fruit from a single bag was washed before PMA treatment and DNA extraction as described above.

For this study, plums of variety “Victoria” were purchased in the early winter of 2021, surface sterilised. Each fruit was then inoculated with an M. laxa mycelial plug placed into a freshly created wound, and then placed on plastic trays in a glasshouse compartment to mummify for 6 weeks.

As for the single application in autumn, there were four product treatments: B91, Y126, fungicide and control (distilled water). Each product was applied on the following three dates: 17/01/2022, 24/01/2022 and 01/02/2022; any given fruit only received one application. Thus, there were in total 12 treatments: four products each applied on three different dates. All experimental fruit, irrespective of application timing, were placed into mesh bags and hung on branches of cherry trees in a netted orchard on 14/01/2022, three days before the first application. All fruit were taken back to the laboratory on 07/02/2022 and incubated under 100% moisture conditions for one week before estimation of M. laxa conidia density with a haemocytometer. Treated mummified fruits were washed in 200 ml of Maximum Recovery (Sigma-Aldrich) for one hour on a shaking incubator at 180 rpm. The wash was then filtered through a sterile muslin. A serial dilution was then performed before a haemocytometer was used to estimate the number of M. laxa spores in each aliquot.

In spring 2019, an orchard trial was conducted to assess the survival of B91 and Y126 on blossoms of cherry variety “Kordia” from late spring through to the summer/fruit harvest. The experiment included five treatments: B91, Y126, a fungicide (Luna Sensation), Serenade (a commercially formulated B. subtilis QTS713 strain), and a sterile distilled water control. These products were applied to all blossoms/fruits in selected and marked branches of cherry trees in an orchard at East Malling just a few days after full bloom. Following application, blossoms/fruit treated with B91, Y126 and distilled water were taken at six time points (1 [blossom], 8 [blossom], 34 [green fruit], 45 [green fruit], 79 [ripening fruit] and 92 [ripe fruit] days after application) to assess the viable population size of B91, Y126 and M. laxa. For comparison, fruits or blossoms of Serenade and Luna Sensation treatments were only sampled at the first and last time points since our focus was on the dynamics of M. laxa in relation to B91 and Y126.

A randomised block design was used for the experiment. There were three blocks; within each block, there were four trees - a single tree was randomly assigned to the B91, Y126, or water control treatment, and the remaining tree was used for both Luna Sensation and Serenade. Thus, there were a total of 12 trees. On each tree assigned to the B91, Y126, or water control treatment, one of six branches was randomly assigned to one of the six sampling time points. On each tree allocated to Luna Sensation and Serenade, two branches were first randomly allocated to one product and then a single branch was assigned to either the first or last sampling time point.

At each sampling time point, all treated fruits/blossoms on each branch allocated to the time point were taken back to the lab. For the ripening and ripe fruit samples, 10 similar-sized fruits were selected from each branch; for all the other four sampling points, 20 similar-sized flowers/green fruits were taken from each branch. These selected flowers/fruits were immediately washed in Maximum Recovery (Sigma-Aldrich) for one hour on a shaking incubator at 180 rpm. The washes were collected, spun down and the pellet resuspended in 500 µl of sterile distilled water. Washes were treated with PMA before subsequent DNA extraction and qPCR analysis. Estimated viable CFU were expressed as number per fruit/blossom.

In 2020, an inoculation experiment was conducted on potted trees of two cherry cultivars (“Kordia” and “Sweetheart”) in a polythene tunnel at East Malling to study BCA application timing relative to inoculation with M. laxa. The three products (B91, Y126 and distilled water as control) were applied on the same day but blossoms were inoculated with M. laxa at two different time points: 4 days before and 1 day after product application. To assess the background level of M. laxa, fruits or blossoms were not inoculated with M. laxa on one randomly selected branch on each tree, namely the control treatment.

A random block design with four blocks was used for this experiment. Within each block, there were three potted trees for each of the two cultivars; a single tree of a given cultivar was randomly allocated to one of the three product treatments. On each tree, three branches were selected, two of which randomly assign to the two M. laxa inoculation times (one for each inoculation time); the remaining branch was used as the control to assess the background level of M. laxa. Branches were prepared by removing old and new (un-opened) blossoms, so all inoculated and treated blooms were open and of similar age. M. laxa spore suspension was adjusted to 1x10⁴ spores ml^-1 with a haemocytometer for inoculation. Both pathogen inoculation and microbial applications were via a hand-held sprayer sprayed until runoff. On each tree, all experimental blossoms were sampled 5 days after the application of treatment products, and PMA-qPCR used to estimate viable population sizes of B91, Y126 and M. laxa as for the field experiment.

In all analyses, all CFU data were log₁₀ transformed before statistical analysis. In many cases where PMA-qCPR failed to generate an estimate, we used the detection/quantification limit instead (which was equal to 3.63 on the log₁₀ scale).

Applying B91 resulted in a large and highly significant (P < 0.001) increase in the viable B91 population size at the first sampling. Overall, the application of B91 in the autumn led to an increase in the viable population size of B91 over the winter when compared to the control: 5.95 (± 0.174) vs. 4.69 (± 0.136). Although the overall B91 population size did not change much over time (from 4.97 ± 0.191 to 5.52 ± 0.237), there were significant interactions between treatments and time (P < 0.001, Figure 2A ). Specifically, there were no differences between the two treatments on the 3^rd and last sampling points, but applying B91 led to a much higher B91 population size for the other four time points.

Overall, applying Y126 did not consistently alter the Y126 population size over the following winter (ranging from 3.79 ± 0.138 to 4.42 ± 0.0.198; P = 0.097), although initially it led to a large increase in Y126 population ( Figure 2B ). Although the mean Y126 population size did not vary much over time, there were significant interactions between treatments and time (P < 0.05, Figure 2B ).

The average effect of treatment on the population size of M. laxa was affected by treatments (P < 0.05) but the interpretation of this was complicated by the significant interactions between treatment and assessment time (P < 0.01, Figure 2C ) – relative performance of treatments varied greatly with time. The average number of M. laxa CFU was, in decreasing order: control 4.72 (± 0.148); B91 4.51 (± 0.144); Y126 4.22 (± 0.136) and fungicide 4.14 (± 0.121). For individual samples there was no significant correlation between M. laxa CFU and either of the B91 and Y126 CFU counts.

Increasing the interval between applying products and assessment led to reduced (P < 0.001) M. laxa sporulation on mummified fruit. The average M. laxa spore counts were 4.16 (± 0.061), 5.27 (± 0.042) and 6.51 (± 0.069) for fruits receiving treatments on 17/01/2022, 24/01/2022 and 01/02/2024, respectively. Products also differed (P < 0.001) in their effect on M. laxa sporulation. The average spore counts of M. laxa were 5.23 (± 0.026), 5.55 (± 0.030), 5.38 (± 0.032) and 5.10 (± 0.028) for the B91, Control, Fungicide and Y126 treatment, respectively. There was also a significant interaction between treatments and the application timing (P < 0.001) ( Figure 3 ).

B91 was detected in several samples at the first two sampling points. The viable population size was similar to the water control at all time points. Applying B91 to cherry blossoms in an orchard did not result in appreciable differences in the viable population size of B91 ( Figure 4A ) over time. The viable B91 population size (P < 0.001) rose asymptotically between the six time points, primarily because of the higher level from the green fruit onwards ( Figure 4A ). For the Serenade treatment, B. subtilis was not detected on the first sampling date and was 9.59 (± 0.008) on the last date, compared to 9.59 (± 0.440) and 7.48 (± 1.930) for the B91 and control treatment, respectively.

Y126 was not detected in several samples at the first and second time points, particularly for the control treatment, and hence set to the detection limit. Applying Y126 to cherry blossoms appeared to increase the viable population size of Y126 initially ( Figure 4B ) when compared to the control treatment. Unlike B91, the differences in the viable Y126 population size between the six time points were not significant (P = 0.089). There were no significant interactions between treatment and assessment time.

M. laxa was not detected in any sample at the first time point, and in few samples at the second time point. Although Y126 appeared to reduce M. laxa CFU ( Figure 4C ), differences between treatments were not statistically significant because of the large variation between replicate samples. The average M. laxa CFU was 6.43 (± 0.681), 6.11 (± 0.547) and 5.18 (± 0.559 for the B91, Control, and Y126 treatment, respectively. Overall, M. laxa CFU increased with time (P < 0.01, Figure 4C ). On the final sampling date, the average M. laxa CFU was 7.57 (± 1.290), 7.54 (± 0.181), 9.88 (± 1.370), 8.38 (± 2.100) and 5.24 (± 1.111) for the B91, Control, Fungicide, Serenade and Y126 treatment, respectively.

M. laxa was only detected in six inoculated samples, all from the control treatment (namely not treated with biocontrol microbes; later inoculation with M. laxa led to higher M. laxa CFU values than earlier inoculation although this difference was statistically not significant (P = 0.21). As M. laxa was not detected in treated samples, no statistical analysis could be carried out. In all 16 samples taken from fruits treated with B91, B91 was above the detection level, ranging from 1.45x10⁵ CFU/fruit to 7.88x10⁷ CFU/fruit. Y126 was above the detection level in 14 of the 16 samples treated with Y126.