Following Rain 1, the relative abundance of 44 OTUs changed significantly on the tomato fruit surface, including 10 members of the Enterobacteriaceae and 3 members of the Xanthomonadaceae, all of which increased in average relative abundance. Tere were no significant differences at the family level between 9/9 and 9/13, however there were significant differences in 6 low-abundance bacterial families between 9/9 and 9/25. For example, relative abundance of the family Rhodobacteriaceae increased steadily over the study period , from ~0.35% to 3.5%. Te most abundant taxa tended to fuctuate throughout the study period irrespective of proximity to rainfall. On average, the Pseudomonadaceae increased in relative abundance from 9/9 to 9/13 , decreased on 9/17 , only to increase again after Rain 2 and decrease again by 9/25 . Te increase from 9/9 to 9/13 was not statistically supported, however the increase from 9/17 to 9/22 was statistically significant . Relative abundance of the Oxalobacteriaceae also increased significantly following Rain 2, from 0.5% to 2.9% , following a small borderline insignificant increase after Rain 1, from 9/9 to 9/13 . This temporary post-rainfall increase in relative abundance of the Oxalobacteriaceae mirrored results seen on cucumber.On tomato leaf surfaces, increases in α-diversity around rainfall were not discernible, plastic gutter with an average of 150 OTUs per sample. Tough not significant, the pre-rain timepoint had the highest median number of OTUs per sample .
A core microbiome of 27 OTUs was observed in all leaf samples. Twenty-one OTUs common to all pre-rain samples were not observed 1 day post-rain, and only 6 new OTU introductions common to all 1 day post-rain samples were detected. No OTUs were found to be common across all samples between 1 day and 4 days post-rain timepoints other than the core 27 common to all 3 dates, while 7 OTUs shared among all 4 days post-rain and pre-rain samples were recovered. On the other hand, shifs in β-diversity were detected across the full sampling period; timepoint significantly influenced bacterial community structure at a rarefaction level of 8,200 sequences per sample when analyzed with unweighted UniFrac distance and Bray-Curtis dissimilarity . However, weighted UniFrac analysis did not reveal a significant effect . No taxa shifted in relative abundance between the pre-rain and 1 day post-rain timepoints, but changes in the average relative abundance of some low-abundance families were detected between the pre-rain timepoint and 4 days post Rain 2. Te Unweighted UniFrac distances between averaged sample groups from each timepoint suggested a weak but consistent shift in community structure over the course of the sampling period, but also considerable variation within samples from the last timepoint .Cucumber fruit surfaces yielded a higher number of bacterial OTUs, 281 , compared to tomato fruit surfaces, 232 . Similarly, bacterial community structure differed significantly between the 2 fruit types .
Several of the most dominant taxa differed in average relative abundance between cucumber and tomato fruit surfaces, including Pseudomonadaceae , Xanthomonadaceae , Methylobacteriaceae , and Microbacteriaceae . Enterococcaceae were higher in average relative abundance on cucumber compared to tomato fruit . Taxa of the Rhizobiaceae family were prevalent on both tomato and cucumber fruit surfaces and were not significantly different . Similarly, members of the Enterobacteriaceae were dominant on both tomato and cucumber fruit surfaces . .In order to more directly address the food safety implications of potential bacterial community changes in response to rainfall, samples were screened for presumptive generic E. coli, frequently used as an indicator of fecal contamination. In total, tomato leaf samples had significantly higher levels of presumptive E. coli compared to tomato and cucumber fruit samples . No significant differences were observed between sampling dates for any sample type, although presumptive E. coli counts and variability in the data increased in all samples following Rain 1 and the highest individual values within each sample type were observed on 9/13, the day following the first rain event .Although the impact of precipitation on fruit and vegetable crop microbiomes has not been directly investigated, it has long been understood that increased plant disease and food safety risks can succeed rain events by enhanced dissemination of pathogens, splash or fooding and by provision of more favourable growth conditions. Shifts in precipitation and periods of drought could also affect bio-control microorganisms, altering the suppressive potential of microbiomes. Reaching a more nuanced understanding of the precise effect of precipitation on crop microbiomes and the microbial dynamics that ensue is valuable to the application of systems thinking and approaches to crop protection, especially when confronted with climate change and increasing severity and duration of rainfall and drought periods. Te consistent seasonal fuctuations recently described in the airborne microbiome captured in rain and snow imply that bacterial introductions via rain could be predictable during sequential rain events in a given agricultural region, hence allowing for crop management decisions to be made with regards to anticipated microbial dynamics in response to drought or precipitation.
In this study, assessment of microbiomes associated with cucumber and tomato surfaces using high quality amplicons of the 16 S rRNA gene provided a novel contribution to our understanding of the impact of rainfall on epiphytic bacterial communities at the time of crop harvest. This work demonstrated an increase in bacterial species diversity on cucumber and tomato fruit surfaces following rain events, at times accompanied by shifts in bacterial community structure. Several new bacterial taxa were introduced to the cucumber and tomato carpoplanes following rainfall and persisted at low abundance in the days following precipitation. This points to a window of time following precipitation during which introduced taxa may become newly established in the phyllosphere, potentially constituting a period of high risk for plant disease and food safety. On cucumber fruit surfaces, the relative abundance of several of the most dominant taxa changed following rainfall, often fully or partially returning to pre-rain proportions within 4 days. Furthermore, overall bacterial community structure on cucumber fruit shifed significantly in response to rain as measured both when incorporating phylogenetic relatedness of bacteria present and including the relative abundance and phylogenetic relatedness of bacterial OTUs . Tomato fruit-associated bacterial communities shifted throughout the study period when assessed by OTU richness and the unweighted UniFrac metric. By contrast, this observed sustained increase in diversity on the tomato fruit surface was not clearly accompanied by shifts in measures that take relative abundance into account, being only detected using Bray-Curtis dissimilarity but not weighted UniFrac. An additive effect considering multiple rain events could explain these differences, however other drivers cannot be ruled out. Phyllosphere microbiome responses to rain have not been investigated in crop systems but reports from rhizobacterial communities suggest that below ground microbiota may be more resilient. Studies on the effect of precipitation on the rhizosphere microbiome appear to reveal that rain has a limited impact on bacterial community composition but can drive changes in relative abundance of taxa. Our study indicated that phyllosphere bacterial assemblages of cucumber and tomato are responsive to precipitation events. This is turn suggests thatphyllosphere microbiomes could be amenable to modulation with the aim of achieving desired outcomes such as disease resistance, enhanced food safety and stress tolerance. Unlike fruit surface communities, blueberry container tomato leaf surface community α-diversity remained largely consistent across all sampling dates; a decrease in the number of OTUs observed 1 day post-rain was not statistically supported. A shift in community structure was observed in the phylloplane over the course of the sampling period, however, in concordance with results from tomato fruit, the incorporation of abundance data into the UniFrac metric diluted the effect. For both tomato fruit and leaves, this indicates that any changes in community structure could likely be attributed to shifts in low-abundance taxa. In tomato fruit, these changes in low-abundance taxa could be observed both through shifs in relative abundance and introduced taxa.
The diminished effect on leaves, on which little post-rain seeding of novel taxa was detected, could be the result of the strong influence of the plant host on species recruitment. However, abiotic conditions on the leaf could also be a factor, with higher relative humidity surrounding leaves as a result of transpiration and trapped moisture within the layer of abundant trichomes. By contrast, mature tomato fruit lack stomata and trichomes, such that moisture from rain could have significantly changed growth conditions by increasing the amount of free water available to microorganisms. Compared to results seen on tomato, microbiota associated with the cucumber carpoplane were more responsive to weather-related changes, although only one rain event was evaluated in the case of cucumber. Fruit and vegetable crops harbor distinct bacterial communities that could be based on inherent differences in plant surface topography and nutritional profile. Differences in cropping practices could also partly explain differences between the two fruit crops. Cucumbers were grown on plastic culture on the ground, while tomatoes were staked upright. Cucumber fruit lying directly on plastic mulch were lef both more exposed to direct rainwater contact and closer to the soil, increasing the potential for splash. Newly introduced taxa may have originated from rain or transferred via splash or wind from soil or nearby plant parts. While the sampling dates were selected to surround rain events, other weather dynamics during the sampling period could not be controlled for and likely influenced plant-associated microbiomes as well. Differences in barometric pressure and wind speed or reduced UV stress due to cloud cover could have influenced crop phytobiome dynamics or interacted with the factor of rain. Furthermore, rainfall may have been correlated with larger scale ecosystem changes. For example, insect visitation, which can affect plant microbiomes may have been limited during the rain event but elevated in the days following precipitation. Pesticides were applied to tomato plots during the sampling period, on the evenings of 9/9 and 9/16. It is possible that these applications could have influenced microbiome structure and diversity, however phyllosphere bacterial communities tend to be fairly robust in the face of pesticide application. Pesticides were not applied to the cucumber plot and similar but more discernable responses to rainfall were detected on cucumber fruit. Prior to Rain 1, the region experienced a long drought with negligible rainfall since the previous major rain event 3 weeks before the study began. The increase in diversity observed following Rain, 1 but not Rain 2, could be explained by drought-induced suppression of bacterial diversity at the start of sampling, not replicated prior to Rain 2, which occurred only 9 days later. Due to the close proximity of the rain events, it is possible that bioaerosols were less prevalent during the second rain. Plants release microbes into the atmosphere preferentially on sunny, dry days, and there were few of those between the two rain events. Unfortunately, cucumber data for Rain 2 were not collected due to seasonality and a lack of availability of high-quality fruit samples, such that we cannot address whether the difference between Rain 1 and Rain 2 is mirrored on cucumber. Te shifts in bacterial OTU richness and in some cases community structure that we did observe following rain events could have been the result of direct inoculation by rainwater-associated microbiota or by other factors associated with rainfall. Rain could physically remove microbes from the plant surface, opening up a niche for others to fill. Alternatively, increased moisture and relative humidity in the air before, during and after rain events could favor rapid growth of certain taxa at the expense of others. In previous work we noted that E. coli levels on lettuce spiked after moderate precipitation but plummeted after heavy rainfall, suggesting that the effect of rain is a balance between new species introductions, stimulating growth conditions as a result of enhanced moisture, and a depleting effect, depending on rainfall depths. In this study, we saw an increase in variation of E. coli on fruit and leaves, but no significant hike in population levels, suggesting that this taxon is not an adequate indicator of bacterial community shifts in the phyllosphere. While it is important to understand the local influence of isolated rainfall events on microbial dynamics in agriculture, in the future it will also be important to consider the influence of weather patterns on a larger scale. In addition to the direct impact of rainfall on phytobiomes, prolonged wet or dry periods could influence plant health and immune responses, and storms could lead to wounding, creating opportunities for pathogens to infltrate plant tissues.