Introduction

The Intergovernmental Panel on Climate Change (IPCC) conservatively estimates that global temperatures will increase by 1.4-4.4 °C by 2100 depending on greenhouse gas emissions (Masson-Delmotte et al., 2021). In Atlantic Canada, temperatures are predicted to increase from 2 to 4 °C in the summer and 1.5 to 6 °C during the winter (Vasseur & Catto, 2007). As a result, growing seasons will be longer and hotter, with shorter, warmer, and damper winters. These temperature changes will impact local agriculture by decreasing yields through heat stress, drought, frequent heavy rainfalls, and the increased populations of phytopathogens (Looby & Treseder, 2018; Jobin Poirier et al., 2020; Lesk et al., 2022; Singh et al., 2023).

Temperature is a key driver of grape development, and climate-induced temperature changes are threatening global viticulture and wine production (Jobin Poirier et al., 2020; Ausseil et al., 2021). In Atlantic Canada, these increased temperatures could affect grapevine phenology (i.e., developmental stages) and grape ripening, increase disease pressure and, ultimately, affect fruit quality. Grape varieties grown in the region are interspecific crosses between Vitis vinifera and Vitis species indigenous to North America (e.g., V. labrusca, V. riparia), which have been selected for strong cold resistance due to a unique biochemical profile and an early maturity (Pedneault et al., 2013). Typically in the region, berry fruit-set occurs in late June, with growth continuing until the end of August and the onset of veraison, the ripening stage. Berries will continue to ripen until harvested in mid to late October (Campos-Arguedas et al., 2022). During the entirety of berry development, including the berry growth phase, temperature affects the development of flavour, aroma, sugar, and alcohol content of berries (van Leeuwen & Seguin, 2006; Campos-Arguedas et al., 2022). For example, early development of berries is characterised by the accumulation of malic and tartaric acids, and proanthocyanidins. During late development (i.e., ripening), malic acid breaks down, while sugars and various secondary metabolites accumulate (Bonada et al., 2015; Campos-Arguedas et al., 2022). Our previous work has further experimentally confirmed in field experiments that increasing the temperature during early berry development changes their biochemistry at harvest (Campos-Arguedas et al., 2022). Unexpected changes to the biochemical profile of the grapes due to climate-induced temperature variation may thus impact the quality of the grapes and wine.

Temperature, a key factor in the “terroir” (van Leeuwen & Seguin, 2006; Di Paola et al., 2023; Johnston-Monje et al., 2023), also helps determine the composition and structure of resident microbial communities (Romero-Olivares et al., 2017; Looby & Treseder, 2018; Nottingham et al., 2018; Frindte et al., 2019; Malik et al., 2019; Jannson & Hofmockel, 2020; Tiedje et al., 2022). In soils, bacterial and fungal diversity generally tracks global abiotic patterns of temperature and precipitation (Kivlin et al., 2014; Leff et al., 2018; Nottingham et al., 2018; Malik et al., 2019; He et al., 2020; Koskella, 2020; Zhu et al., 2021; Tiedje et al., 2022). However, fungi do tend to make up more of the soil biomass across the boreal forests (50° N-70° N, Treseder et al., 2014; He et al., 2020). This appears to be due to the slower rates of decomposition, largely performed by free-living filamentous fungi (Treseder et al., 2014; Treseder & Lennon, 2015). Despite this, many fungal lineages have been shown to quickly adapt to warming soils. This is particularly true of phylogenetically younger lineages and those more adept at budding, such as yeasts, unlike filamentous fungi (Treseder et al., 2014; Romero-Olivares et al., 2017; Looby & Treseder, 2018). Although soils may be the reservoir for many microbes in the aerial portions of plants (phyllosphere; hereafter used to specify the leaves), important differences can still accrue between above- and below-ground communities over the development of the plant host (Jannson & Hofmockel, 2020; Koskella, 2020; Blakney et al., 2024b).

In general, microbes that colonise aerial portions of plants survive under difficult conditions of limited nutrients, exposure to UV radiation, and fluctuations in pH, humidity, and temperature (Lindow & Brandl, 2003; Koskella, 2020). The composition and structure of above-ground plant microbial communities has also been shown to be impacted by temperature (Cordier et al., 2012; Koskella, 2020; Zhu et al., 2021; Perreault & Laforest-Lapointe, 2022). Among fungi, for example, increased temperatures have been seen to increase the diversity of yeasts, which are thought to be more stress resistant, as well as ascomycetes, such as Cladosporium sp. (Treseder & Lennon, 2015; Looby & Treseder, 2018; Cureau et al., 2021). However, other studies have highlighted decreases in phyllosphere fungal diversity due to warming (Zhu et al., 2021), as well as across growth stages as communities stabilise (Boutin et al., 2024). Foliar microbes also play important roles for their plant hosts by providing nutrients (Moyes et al., 2016; Laforest-Lapointe et al., 2017), as well as by disrupting the invasion and colonisation of potential phytopathogens (Ritpitakphong et al., 2016; Laforest-Lapointe et al., 2017). This can be particularly true of fungal communities, since they can dispersal aerially, and are not necessarily dispersal limited (McGuire et al., 2011; Abdelfattah et al., 2019; Wang et al., 2021; Johnston-Monje et al., 2023).

In freshly developing foliar plant organs, like leaves and fruits, the order of colonisation is a key driver of microbial community dynamics (Fukami, 2015; Hiscox et al., 2015; Mori et al., 2017; Debray et al., 2021). The priority effects of arrival order are generally described by niche pre-emption or niche modification (Fukami, 2015; Debray et al., 2021). In the former, available niches are filled on a first-come first-served basis, leading to the exclusion or inhibition of late arrivals into the environment (Fukami, 2015; Debray et al., 2021). In the latter, early-arrivals change the local environment to create new niche opportunities for later arrivals (Fukami, 2015; Debray et al., 2021). Several factors can modulate the strength of priority effects, including phylogenetic relatedness, niche overlap, and environmental stability, (Fukami, 2015; Hiscox et al., 2015; Mori et al., 2017; Debray et al., 2021). All of these elements could contribute to the emergence and spread of new phytopathogens, as the climate crisis creates more variable temperatures across geographic regions (Malik et al., 2019; Jannson & Hofmockel, 2020; Koskella, 2020; Zhu et al., 2021; Tiedje et al., 2022; Singh et al., 2023). Therefore, there is a pressing need to document microbial community composition and assembly dynamics through time in different plant-associated environments; e.g., soil, leaves, fruits, flowers, and stems.

To this end, bacterial and fungal communities of grapevines and vineyards have been well surveyed throughout time and space (Perazzolli et al., 2014; Fort et al., 2016; Kecskeméti et al., 2016; Singh et al., 2018; Abdelfattah et al., 2019; Singh et al., 2019; Deyett & Rolshausen, 2020; Perazzolli et al., 2020; Cureau et al., 2021; Liu & Howell, 2021; Steenworth et al., 2021; Behrens & Fischer, 2022; Testempasis et al., 2023; Wicaksono et al., 2023; Cui et al., 2024; Leal et al., 2024; Teixeira et al., 2024). However, to our knowledge, there have been no studies that capture the impact of temperature and phenology on the microbial communities of grapevines. Moreover, in Canada, there have been very few experiments exploring the relationship between grapevines and their microbial communities. Furthermore, to our knowledge, there have been no studies on the microbial communities of the hybrid grapevine cultivars grown in Atlantic Canada, nor of their response to the climate crisis. Therefore, experimentally modelling how fungal communities of grapevine phyllospheres and fruits (carposphere) change in response to increased temperatures at different phenologies will be useful for understanding i) fungal community dynamics over time, and ii) the impact of the climate crisis on the fungal communities.

Here, we report a field experiment used to test whether changes in temperature patterns at different stages of fruit development (pre-veraison and post-veraison) or a global rise in temperature (whole season) altered the fungal communities of grapevine phyllosphere and carposphere. Our hypothesis was that increased heat treatments would significantly increase fungal diversity of the phyllosphere and carposphere regardless of development stage. As such, we predicted that i) fungal diversity would increase over the growing season in untreated controls, ii) diversity would be higher among heat treated samples than the untreated controls, and iii) diversity would not be different between treated samples from different growth stages. We used on-the-row mini-greenhouses to increase the temperature at different fruit developmental stages of Vitis sp. cv. L’Acadie blanc; i.e., pre-veraison, post-veraison, and throughout the growing season. Leaves and fruits were sampled at four developmental stages, and fungi were identified via ITS metabarcoding. We measured changes to fungal community composition and α- and β-diversity across different developmental stages to assess the impact of the increased temperature regimes to the grapevine phyllosphere and carposphere communities. These results will allow us to evaluate the adaptability of the fungal community in the face of increasing temperatures in Atlantic Canada. Our data will also generate insights into the potential agroecological roles of grapevine fungal communities, such as pathogenicity and biological control, as well as the fungal components of the terroir.

Materials and methods

1. Site and experimental design

The field experiment, previously described in Campos-Arguedas et al., 2022, was conducted at a commercial vineyard in the Gaspereau Valley in Wolfville, Nova Scotia (45° 4' 19" N 64° 17' 44" W) during the 2020 growing season. The experimental site was planted with an 11-year-old Vitis sp. cv. L’Acadie blanc (Cascade X Seyve-Villard 14-287). The experimental design was a randomised split-plot replicated in five complete blocks. Within each block, plots were split into four treatments, where polycarbonate mini-greenhouses were installed at specific growth stages to simulate increased temperatures. The control treatment lacked greenhouses throughout the season, and was therefore exposed to the ambient environmental conditions without any temperature treatment. The specific growth stages were determined by the modified Eichhorn-Lorenz (EL) system (Coombe, 1995). The whole season heat treatment maintained greenhouses from June the 3rd (developmental stage EL-15) to October the 9th (harvest, EL-38), with an increased growing degree (GDD) days of 197 relative to the control (as previously illustrated by Campos-Arguedas et al., 2022). The pre-veraison heat treatment greenhouses were installed on June the 3rd until September the 10th (onset of veraison, EL-35), with an associated increase of 128 GDD, while in the post-veraison heat treatment the greenhouses were installed on September the 11th until October the 9th (harvest, EL-38), with an increase of 44 GDD (see Campos-Arguedas et al., 2022 for further design details). These periods were selected based on the double sigmoid curve grape berries follow during their development. Note that due to climate variability in field trials, it is not possible to have a fixed temperature increase. Rather, the number of GDD are often reported in agricultural and viticultural field trials. GDD measure the number of days a temperature required for a particular biological process, like grapevine growth, occurred. Therefore, a treatment that received more GDD than the control in the same field trial will have received higher temperatures (Parker et al., 2011; Gregorich et al., 2016; Steenwerth et al., 2021).

2. Crop management and sampling

Grapevines were grown and maintained according to standard management practices, as previously described by Campos-Arguedas et al. (2022). Vitis sp. leaves and berries were sampled at the following growth stages: EL-32 (bunch closure), EL-36 (intermediate fruit sugar levels), EL-37 (fruit not quite ripe), and EL-38 (ripe and harvest). At each of these growth stages, 6-8 berry clusters and their adjacent leaves were randomly picked from all four vines in each treatment per block. Sampled material was divided into leaves and fruits and pooled, generating 160 samples (two compartments (i.e., phyllosphere and carposphere) *4 growth stages *4 treatments *5 replicates). Samples were immediately frozen with liquid nitrogen, transported in dry ice, and stored at –80 °C before being shipped to Université de Montréal’s Biodiversity Centre (Montréal, QC, Canada) on dry ice for further processing (Delavaux et al., 2020; Blakney et al., 2022). As is typical for field experiments, we accounted for the use of the various agricultural management practices (i.e., experimental treatments, fertilisers, pesticides, and management) in the downstream amplicon data by considering each sample and its management history as a unit; i.e., the detection of a given fungal species is considered to be due to the total effect of the management and experimental treatments of the samples.

3. DNA extraction from Vitis sp. leaf and fruit samples

Total DNA was extracted from leaves and fruits after being ground separately in liquid nitrogen via sterile mortar and pestles. DNeasy Plant DNA Extraction Kits (Qiagen, Germany) were used following the manufacturer’s instructions with ~160 mg of leaf material and ~225 mg of fruit material (Lay et al., 2018; Blakney et al., 2022). No-template extraction negative controls were included with each kit used in order to assess the influence of the extraction kits on our sequencing results, and the efficacy of our lab preparation. All extracted DNA samples were qualitatively evaluated by mixing ~2 mL of each sample with 1 mL of loading dye containing Gel Red (Biotium) and running it on a 0.7 % agarose gel for 50 minutes at 110 V. The no-template extraction negative controls were confirmed not to contain DNA after extraction.

4. Amplicon generation and sequencing to estimate fungal communities

To estimate the composition of the fungal communities in the phyllosphere and carposphere across the Vitis sp. growth stages, extracted DNA from all samples were used to prepare ITS amplicon libraries following Illumina’s MiSeq protocols (Bell et al., 2016; Lay et al., 2018; Blakney et al., 2022). First, all DNA samples were diluted 1:10 into 96-well plates. To assess potential bias caused by laboratory manipulations, sequencing and downstream bioinformatic processing, we also included the no-template extraction control samples.

The prepared plates of the Vitis sp. leaf and fruit DNA samples were submitted to Génome Québec (Montréal, Québec) for ITS amplicon generation and sequencing (Bell et al., 2016; Lay et al., 2018; Blakney et al., 2023). Diluted DNA samples were used as templates for PCR amplification with the ITS3-KYO2 forward and ITS4-KYO3 reverse primers, which generated a 430 bp fragment from the ITS2 region, between the 5.8S and LSU regions (Toju et al., 2012; Morvan et al., 2020). Amplicons were then prepared for paired-end sequencing using Illumina’s MiSeq platform (Génome Québec, Montréal) (Bell et al., 2016; Lay et al., 2018; Blakney et al., 2022). We estimated this would provide a mean of 40,000 ITS reads per sample, which is in line with previous studies that describe bacterial communities (Bell et al., 2016; Lay et al., 2018; Blakney et al., 2022; Morvan et al., 2020). Raw sequencing data and metadata are publicly available at NCBI Bioproject under accession number: PRA1046454.

5. Estimating ASV’s from amplicon sequencing

The amplicons generated by Illumina MiSeq were used to estimate the diversity and composition of the fungal communities present in the leaves and fruits of Vitis sp. at different growth stages. The integrity and totality of the MiSeq data downloaded from Génome Québec was confirmed using their MD5 checksum protocol (Roy et al., 2018). Subsequently, all data was managed and analysed in R (4.0.3 R Core Team, 2020), and plotted using ggplot2 (Wickham, 2016).

The raw ITS reads were processed to retain the highest quality reads before ASV inference and taxonomic assignment. Due to the variable length of the ITS region, we first used cutadapt (Martin, 2011) to carefully remove primer sequences from all 1,616,465 raw ITS reads generated from the control samples and the Vitis sp. samples, including any primer sequences generated due to read-through (Blakney et al., 2023). The filtered and trimmed reads for the ITS region were then processed through DADA2 for ASV inference (Figure S1). The default settings were kept throughout the pipeline, except the dada inference function, which used the pool =‘pseudo’ argument to increase the likelihood of identifying rare taxa. Consequently, the chimera removal function removeBimeraDenovo included the method =‘pooled’ argument (Callahan et al., 2016b).

Fungal amplicon sequence variants (ASVs) were assigned taxonomy following the default DADA2 pipeline (Callahan et al., 2016b) and using the UNITE fungal database (Abarenkov et al., 2022), as well as the UNITE database for all eukaryotes (Abarenkov et al., 2020), in order to confirm fungal identities (Tedersoo et al., 2018). Although this is the standard pipeline for fungal ITS2 ASV identification, it is worth noting that variability in the amplicon can bias taxonomy assignment (Yang et al., 2018; Nilsson et al., 2019; Sene et al., 2025). The quality of the data was assessed using the included controls, and any off-target eukaryotic ASVs identified in the fungal data were removed. Rarefaction curves confirmed that we obtained sufficient coverage of the fungi present in both the phyllosphere and carposphere (Figure S2; Blakney et al., 2024a).

6. α-diversity of the Vitis sp. fungal communities in the phyllosphere and carposphere

First, to visualise the taxonomic diversity, ASVs from the phyllosphere and carposphere were plotted separately as taxa cluster maps using the heat_tree function from the metadcoder package (Foster et al., 2017), where nodes represent phyla to genera: node colours represent the abundance of 16S rRNA reads, while node size indicates the number of unique taxa. Taxa cluster maps facilitate the visualisation of abundance, as well as diversity across taxonomic hierarchies (Foster et al., 2017).

Second, we compared species richness using Simpson’s α-diversity index calculated from the phyloseq object (McMurdie & Holmes, 2013). We assessed differences in the mean indices for each heat treatment between growth stages, and their interactions using a multi-factor ANOVA and Tukey’s post-hoc test for significant groups that respected the assumptions of normality (Blakney et al., 2022; Blakney et al., 2023). Normality of the residuals was established with a Shapiro-Wilk test using the shapiro.test function, while the heteroscedascity of residuals was confirmed with using a Bartlett test, bartlett.test function. For significant ANOVAs, a post-hoc Tukey’s Honest Significant Difference test, TukeyHSD, was used to determine which groups were statistically different.

7. Identification of differently abundant ASV’s and specific indicator species

To refine our understanding of the abundance and composition of the Vitis sp. fungal communities, we used two complementary methods to identify taxa specific to growth stages and treatments. First, taxa cluster maps were used to calculate the differential abundance of ASVs between experimental groups using the heat_tree_matrix function from the metadcoder package (Foster et al., 2017). Second, indicator species analysis was used to detect ASVs that were preferentially abundant in pre-defined environmental groups (compartments, growth stages, treatments) using the multiplatt function from the indicspecies package (De Cáceres & Legendre, 2009), with an FDR correction. A significant indicator value is obtained if an ASV has a large mean abundance within a group compared to another group (specificity), and if it has a presence in most samples of that group (fidelity) (De Cáceres & Legendre, 2009; Legendre & Legendre, 2012). The fidelity component complements the differential abundance approach between taxa clusters, which only considers abundance.

8. β-diversity of the Vitis sp. fungal communities in the phyllosphere and carposphere

To test for significant differences between the Vitis sp. fungal communities at different growth stages and heat treatments, we used the non-parametric permutational multivariate ANOVA (PERMANOVA), where any variation in the ordinated data distance matrix is divided among all the pairs of specified experimental factors. The PERMANOVA was calculated using the adonis, adonis function, from the vegan package (Oksanen et al., 2020), with a distance matrix calculated using the Bray-Curtis formula, with 9999 permutations, and the experimental blocks were included as “strata”. We used Bray-Curtis distances because it is appropriately sensitive for microbial community data, and the frequent “zero” counts of ASVs common in these datasets (Jeganathan & Holmes, 2021; Kers & Saccenti, 2022). We confirmed that our data met the assumption of homogeneity using the betadispr function, and applied an ANOVA and Tukey’s Honest Significant Difference post-hoc test to determine that none of the groups were statistically different.

Similarity between communities was also tested and visualised using principal co-ordinate analysis (PCoA, Legendre & Legendre, 2012) using the Bray-Curtis distance matrix. Singleton ASVs were removed before the phyloseq data were transformed using Hellinger’s transformation, such that ASVs with high abundances and few zeros are treated equivalently to those with low abundances and many zeros (Legendre & De Cáceres, 2013).

To further characterise the ecological mechanisms that may be responsible for changes to β-diversity, we partitioned β-diversity into two components: turnover (i.e., species replacement) and nestedness (i.e., loss/gains that result in poor species richness being a subset of richer sites; Blakney et al., 2024b). We compared the β-diversity components for each heat-treated plot at each growth stage with its cognate plot at the following growth stage (e.g., control phyllosphere at EL-32 was compared to control phyllosphere at EL-36), using the betapart.core.abund function from the betapart package (Baselga et al., 2023). Within the phyllosphere and the carposhere, we separately identified any significant differences among the means of each component between growth stages for each heat treatment and their interactions, with a multi-factor ANOVA, as described above for α-diversity, as normality was respected.

Results

1. Illumina MiSeq yielded similar numbers of fungal ASVs from both Vitis sp. phyllosphere and carposphere

Illumina’s MiSeq produced 1,616,465 raw reads for the whole fungal ITS dataset, which were processed through cutadapt and DADA2 (Callahan et al., 2016a; Callahan et al., 2016b), where 1,160,088 reads were retained from all the experimental samples (Figure S1). The number of ITS reads were similar between both leaf and fruit samples, and across growth stages, with a mean of 8,044 ± 2,328 reads among leaves, and 7,865 ± 2,281 reads among fruit (Table 1). From this, a total of 655 distinct fungal ASVs were inferred. The majority of reads from across the dataset were assigned to class Dothideomycetes (phylum Ascomycota), where the percentage of Dothideomycetes reads ranged from 43 % to 99 %, with an average of 93 %, across all 146 samples (Figures 1 and S3). We also observed a consistent trend concerning the number of unique ASVs across the dataset, with a mean of 29 ± 8 unique ASVs among leaf samples and a mean of 22 ± 6 ASVs among fruit samples (Table 1).

2. Fungal communities were significantly different between phyllosphere and carposphere

Overall, the PERMANOVA supported that the fungal communities from the phyllosphere and carposphere of Vitis sp. were significantly different, though the effect size was small (PERM R² = 0.01461, p = 0.044; Table 2). The interaction between growth stage and compartments (i.e., phyllosphere and carposphere) was also significant (Table 2). Several ASVs were identified as specific to the phyllosphere, though weakly significant (Table 3); the ascomycetes Nigrospora spp. (Trichosphaeriaceae), Sphaerulina spp. (Mycosphaerellaceae), Alternaria spp. (Pleosporaceae), an unknown Didymellaceae (Pleosporales), and a sole basidiomycete, Piptoporus spp. (Fomitopsidaceae, Table 3). Only ASVs identified as Cladosporium spp. (Cladosporiaceae) were enriched (i.e., more reads) in the phyllosphere compared to the carposphere (p < 0.05; Figure S4). In the carposphere, only the Polyporales and Hymenochaetales orders were enriched compared to the phyllosphere (p < 0.05; Figure S4), but no family or genus was found to be significant. Two ASVs were identified as specific to the carposphere, the basidiomycetes Naganishia spp. (Filobasidiaceae), and Tilletiopsis spp. (Entylomatales; Table 3).

Contrary to our initial hypothesis, where we predicted that fungal diversity would increase over the growing season, we observed little influence of the different growth stages on α- (Figure 2) and β-diversity (Figure 3). For example, although we did find that different growth stages had a significant effect on the fungal communities in the phyllosphere and carposphere, the effect size was small (PERM R² = 0.03329, p = 0.027; Table 2). We also found that species turnover across growth stages was more significant in explaining the β-diversity of phyllosphere and carposphere communities than species nestedness (p. adj < 0.001; Figure 3D). We did not detect any taxonomic enrichments or depletions according to growth stage (Figure 4), and the fungal communities remained largely taxonomically similar through time (Figure 3A).

3. Increased heat significantly altered fungal communities in the phyllosphere

The different heat treatments did significantly influence the fungal communities of the phyllosphere and carposphere of Vitis sp, and with a larger effect size than compartment or phenology (PERM R² = 0.05794, p < 0.001; Table 2). The interaction between treatment and growth stage was also significant (Table 2), though likely driven by the significant changes between the phyllosphere and carposphere communities, and the significant impact of the heat treatments. Although the structure and composition of the phyllosphere and carposphere communities across all treatments remained largely similar (Figure 3A) among the fungal phyllosphere communities, the heat treatments had significant effects on the α-diversity. Simpson’s index was higher for phyllosphere communities treated with increased heat pre-veraison at growth stages EL-32, 36, and 38, when compared to communities treated with increased heat post-veraison (p. adj = 0.04631; Figure 2A), in opposition to our initial prediction. Moreover, it is worth noting that the untreated control phyllosphere samples remained stable in terms of composition, relative abundance and diversity throughout the experiment (Figures 2A, 3B and 3D), contrary to our prediction. This further supports the relatively weak influence of the different growth stages on phyllosphere community structure (Figure 2A). In the β-diversity analyses of the phyllosphere, the PCoA captured 45.9 % of the variation and illustrated a shift in fungal community similarity, where communities from the control and post-veraison heated communities were more similar in composition versus communities that were heated all season and pre-veraison (Figure 3B). The increased heat treatments also increased species turnover relative to the controls in the phyllopshere communities (p. adj < 0.05; Figure 3D).

A number of fungal taxa were also significantly enriched among the phyllosphere communities due to specific treatments (p < 0.05; Figure 4). ASVs belonging Cladosporium spp. and the Mycosphaerellaceae were over-represented in samples treated for the whole season with increased heat compared to samples from the control and post-veraison heat treatment (p < 0.05; Figure 4A). Samples treated with increased heat pre-veraison were also over-abundant in Cladosporium spp. compared to control samples or those treated with increased heat all post-veraison (p < 0.05; Figure 4A). The pre-veraison samples were also enriched in Filobasidium spp. compared to the samples treated with increased heat post-veraison (p < 0.05; Figure 4C).

In the Vitis sp. carposphere communities, the α- (Figure 2B) and β-diversities (Figure 3) remained stable across the growing season regardless of heat treatments, contrary to our predictions. However, there were slight variations among taxa in the carposphere communities: the samples treated with increased heat all season were enriched in Cladosporium and Fusarium spp., relative to both the control and post-veraison samples (p < 0.05; Figure 4B), while the pre-veraison fruit samples were enriched in the Mycosphaerellaceae compared to the controls (p < 0.05; Figure 4B). Finally, the control samples and those treated with increased heat pre-veraison were enriched in Filobasidium spp. compared to samples treat all season (p < 0.05; Figure 4D).

Discussion

Microbial communities are influenced by temperature and are expected to be impacted globally due to increased temperature variation induced by the climate crisis (Malik et al., 2019; Jannson & Hofmockel, 2020; Koskella, 2020; Zhu et al., 2021; Perreault & Laforest-Lapointe, 2022; Tiedje et al., 2022). This will have significant consequences on productivity and the quality of most agricultural crops, including grapevines, as microbial communities are tightly related to the terroir characteristics of wine (Bokulich et al., 2014; Bokulich et al., 2016; Belda et al., 2017; Liu et al., 2020; Zhou et al., 2021; Gobbi et al., 2022; Johnston-Monje et al., 2023). Therefore, changes in the fungal community of the phyllosphere or carposphere could have negative effects by favouring pathogen populations. Changes in these fungal communities could also yield positive effects by benefitting fungi with a role in biological control or the regulation of pathogen populations. However, despite their important role in grapevine health and disease, the impact of temperature and phenology on the fungal community in the grapevine phyllosphere has been understudied (Perazzolli et al., 2014; Fort et al., 2016; Kecskeméti et al., 2016; Singh et al., 2018; Abdelfattah et al., 2019; Singh et al., 2019; Deyett & Rolshausen, 2020; Perazzolli et al., 2020; Cureau et al., 2021; Liu & Howell, 2021; Steenworth et al., 2021; Behrens & Fischer, 2022; Testempasis et al., 2023; Wicaksono et al., 2023; Cui et al., 2024; Leal et al., 2024; Teixeira et al., 2024), even more so for the carposphere (Kecskeméti et al., 2016; Deyett & Rolshausen, 2020; Testempasis et al., 2023). Here, our field experiment used on-the-row mini-greenhouses to increase the temperature of Vitis sp. cv. L’Acadie blanc across four different growth stages. We then tested how increased temperatures significantly altered the fungal communities of the phyllosphere and carposphere of Vitis sp. throughout the growing season. We identified the fungal communities of the phyllosphere and carposphere at each growth stage using ITS metabarcoding and measured fungal community composition and α- and β-diversity to assess the impact of the increased temperature treatments to the phyllosphere and carposphere communities.

Our results showed typical fungal communities for the phyllosphere, dominated by Ascomycota, with high abundance of Dothideomycetes (Figures 1 and S3), primarily Alternaria spp. and Aureobasidium spp. (Kraus et al., 2019; Molnár et al., 2023). Common phyllosphere fungi, including Alternaria and Nigrospora spp. were also identified as indicator species (Table 3). We observed that the increased temperature treatments had more significant effect on fungal communities than the different growth stages (Table 2, Figures 2, 3 and 4). Although the PERMANOVA supported the significant differences between the fungal communities across growth stages (Table 2), we did not observe any clear causes in terms of composition (Figures 1, 3 and 4). This could be attributed to fungal communities being more stable through time, unlike bacterial communities (Blakney et al., 2022; Blakney et al., 2023; Blakney et al., 2024b). Indeed, previous studies have also reported limited temporal effects on fungal communities (Duchicela et al., 2013; Gschwend et al., 2021). Unlike bacterial communities, which grow and shift relatively rapidly, fungal communities tend to remain more stable over time and are less affected by their hosts (Hannula et al., 2019; Hannula et al., 2021; Zhang et al., 2023; Li et al., 2024). This may be due to fungi having slower growth rates and being more mobile than bacteria. Thus, in our experiment it would be reasonable not to detect vastly different fungal communities across different developmental stages.

Our most significant finding was that the increased temperature during the pre-veraison stage significantly increased the α-diversity of the phyllosphere fungal communities, and that this effect was maintained throughout the growing season (Figure 2A). This suggests that the phyllosphere fungal communities may be more prone to changes earlier in development, nearer the time microbial communities begin colonisation. For instance, the increased temperature treatment could promote the establishment of primary colonisers or pioneer species, or create other kinds of niche space as a priority effect, which continues to structure the community throughout the season (Chase et al., 2010; Hiscox et al., 2015; Kraus et al., 2019; Smets et al., 2022). Interestingly, increasing the temperature all-season did not yield a similar increase in α-diversity of the phyllosphere communities as the pre-veraison treatment did (Figure 2A). Both treatments shared a similar temperature increase at a similar developmental stage for the leaf, but we did not observe a consistent impact on the diversity of the fungal communities. This could suggest that any priority effect or additional niche space created in the phyllosphere by the pre-veraison temperature treatment may not be sustained by the prolonged all-season heat treatment. Instead, the longer all-season heat treatment may provide more time for the fungal communities in the phyllosphere to recalibrate.

The potential pre-veraison priority effect could also be related to the significant enrichment of ASVs belonging to the Cladosporium, or Filobasidium spp. in the phyllosphere communities (Figure 4). Both groups have the potential to be antagonistic toward the plant: Filobasidium spp. are common basidiomycetes and tend to be poor saprotrophs, favouring environments where they can parasitize other fungi, or plants (Weiss et al., 2014; Detheridge et al., 2020), while Cladosporium spp. are also well-known in the phyllosphere and diverse plant pathogens (Kraus et al., 2019; Cosseboom & Hu, 2023). We also identified a number of ASVs specific to the phyllosphere as indicator species (Table 3) from among the Dothideomycetes class, which dominated our communities by relative abundance (Figure 1), including Sphaerulina and Alternaria spp. (Table 3). A number of Sphaerulina species are known as causative agents of leaf spot disease among other plants (Ali et al., 2021). Equally diverse among grapevine phyllosphere communities are Alternaria spp., which range from saprotrophs to pathogens (Molnár et al., 2023). Alternaria spp. produce a wide range of metabolites and depending on their host and environmental factors can contribute to anti-microbial activities or virulence (Molnár et al., 2023). Any of these taxa could contribute to the development of the community through priority effects (Fukami, 2015; Debray et al., 2021).

The pre-veraison priority effect showed increased fungal diversity after increased heat early in fruit development and that higher diversity was maintained at each subsequent growth stage. Fungi arriving later during leaf development could depend on the earlier microbes i) occupying all available niche space (niche pre-emption), or ii) altering the niches that exist (niche modification; Fukami, 2015; Debray et al., 2021). Soil fungi have been shown to extensively partition available niche space, allowing for increased diversity (Cho et al., 2017). Other studies have also suggested that early-arrivals are less specialised and less able to manage microbe-microbe interactions, which encourages species turnover (Debray et al., 2021; Wang et al., 2023; Bai et al., 2024). We also observed significantly increased species turnover in the pre-veraison heat treated communities compared to the untreated controls (Figure 3C). Fungal turnover in the phyllosphere has been observed previously, suggesting dispersal-limited communities, subject to strong environmental filtering occurring on the leaf and fruits (Wang et al., 2023; Bai et al., 2024).

Our data suggests that primary colonisers could be used in biological control or even in agroecological protection, specifically by applying a cocktail of organisms early in the season composed of good colonisers or pioneer species in order to limit possible colonisation by phytopathogenic organisms (Kraus et al., 2019). Biological disease control can be achieved by applying registered biological control agents, but also by promoting populations of biological control agents naturally present in the vineyard; an approach widely used in agroecology (Kraus et al., 2019; IPPC, 2024). Microbes – essentially bacteria, fungi and nematodes (Lindow & Brandl, 2003; Perazzolli et al., 2014; Sapkota et al., 2015; Laforest-Lapointe et al., 2017; Koskella, 2020) – that colonise the phyllosphere or carposphere (Vorholt, 2012) can act as biocontrol agents insofar as they compete for space and nutrients, i.e., niche pre-emption (Fukami, 2015; Debray et al., 2021). For example, Aureobasidium pullulans and Trichoderma spp. compete with the grapevine pathogen Botrytis cinerea (Fedele et al., 2020); ASVs for both Aureobasidium and Trichoderma spp. were detected in our data (Figure 1).

The most common diseases present in northern climates are downy mildew (Plasmopara viticola), powdery mildew (Erysiphe necator), anthracnose (Elsinoe ampelina), and botrytis bunch rot (Botrytis cinerea), but there are more than 50 diseases listed in the American Compendium of Grape Disease (Carisse et al., 2006; Wilcox et al., 2015). Among the common phytopathogens identified in our data Botrytis, Cladosporium, and Fusarium spp., were all detected in the phyllosphere, and in higher abundance in the carposphere (Figure 1; Lorenzini & Zapparoli, 2015; Cosseboom & Hu, 2023; Bustamante et al., 2024). These diseases not only affect yields, but also the quality of the grapes and, therefore, the wine they produce (Lorenzini & Zapparoli, 2015). Currently, these diseases can be controlled by the application of synthetic fungicides, which are still a key input in viticulture, because most traditional wine grape varieties (V. vinifera varieties) are highly sensitive to them (Pedneault & Provost, 2016; Provost & Pedneault, 2016). Synthetic fungicides can also change or damage the resident fungal community (Fournier et al., 2020). As such, more and more winegrowers are seeking to reduce the use of these products in favour of biological disease control along with the introduction of resistant grape varieties (Pedneault & Provost, 2016; Provost & Pedneault, 2016; Zhang et al., 2020; Candel et al., 2023; Oliver et al., 2024).

The advantage of using organisms naturally occurring in the vineyard is their redundancy, as several can either act in concert or individually if conditions favour one species over another. Fungal communities can therefore play an important role in the resilience of agrosystems – including viticulture – to climate change, and be a tool for the agroecological control of grapevine diseases. Our study provides a robust baseline for the succession of fungal communities across the development of the phyllosphere and carposphere that future experiments can use, particularly in determining the ecological roles of the fungal ASVs we report.

Conclusion

Temperature variations induced by the climate crisis are a significant threat to global agriculture, including viticulture. Canadian viticulture has been reported to be the least prepared to adapt to future temperature variation (Jobin Poirier et al., 2020). Moreover, climate changes also have important consequences for the resident above-ground fungal communities. To date, there have been very few experiments exploring the relationship between grapevines and their microbial communities. To our knowledge, there have been no observations concerning the microbial communities of the hybrid grapevine cultivars grown in Atlantic Canada, nor of their microbial response to the climate crisis. We hypothesised that increased heat treatments would significantly increase fungal diversity of the phyllosphere and carposphere regardless of development stage of the host, Vitis sp. cv. L’Acadie blanc. We found specific phyllosphere and carposphere communities that were statistically different, with each compartment having unique indicator taxa, despite both compartments having similar composition overall. Both phyllosphere and carposphere fungal communities tended to remain stable across grapevine development. However, we did observe significant changes in the phyllosphere communities pre-veraison due to increased temperatures. These findings suggest that early-season warming fosters fungal diversity, which could influence microbial interactions relevant to disease suppression. The on-the-row mini-greenhouses or tunnels could be deployed early in the growing season to foster more diverse fungal communities, as more diverse communities tend to be more resilient to phytopathogens. This could allow producers to avoid having to resort to more costly options, such as additional fungicides or external heaters. Further research is needed to assess the potential role in phytopathogen control of this approach. This knowledge may also be integrated into decision-making processes for selecting sites for new vineyards, where historic meteorological data could be used to identify sites with warmer conditions prevalent early in the season. Finally, our study provides a standard reference for future work to determine the precise ecological roles of the fungal ASVs that were enriched due to the increased temperature, and more broadly for the adaptation of Atlantic Canadian grapevines and their fungal communities to future climates.

Acknowledgements

AJCB performed the DNA extraction and sequencing prep, analysed the data, and drafted the manuscript with input from all co-authors. This work was supported by a Mitacs Acceleration Grant to AJCB & KP (Grant Number IT31632), which we gratefully acknowledge. We thank graduate students Francisco Campos Arguedas and Guillaume Sarrailhé who performed the greenhouse experiment as part of their projects, and Paméla Nicolle at Université du Québec en Outaouais for her technical help.