Specifically, a principal coordinate analysis of weighted UniFrac distances indicated significant clustering of individual microbial communities by phenological stage . In a pairwise comparison of community compositional differences between each phenological stage, the leaf bacteriome had the greatest number of significant adjusted P values, with 21 of the 21 pairwise comparisons being significantly different, followed by the leaf mycobiome . Rainfall, fertilizer applications, temperature, and irrigation hours fluctuated across our sampling period . Rainfall was sparse in this sample location , ranging from 0.00 to 2.55 inches each month , and total rainfall was a minor determinant of community structure across all four communities, explaining only 0.8 to 2.0% of the variation . Similarly, fertilizer application describes a small percentage of the variation in the data for all four communities examined. We evaluated temperature based on the average temperature and interactive effects it might have with water availability in order to capture the full range of conditions that could affect microbial community composition. Temperature had a minor impact on communities, as this factor describes only 0.6 to 2.8% of the variation in the data that include temperature as an interaction factor. In addition to phenology, interactions between phenology and sample year were a driving factor of leaf bacterial community composition, explaining 8.2% of the changes across the data . Taken together, these beta diversity analyses indicate that plant phenological stage was the major driving factor in community composition for bacterial and fungal communities associated with leaves and roots.

Significant compositional shifts were also visible at the phylum level, container growing raspberries particularly in the leaf bacterial community . Other covariates tested were minor or insignificant contributors to citrus-associated leaf and root microbiome composition.We identified core microbial taxa for each of our seven phenological stages. Our core bacterial and fungal leaf and fungal root microbiomes include genera that were greater than 0.01%, and core root bacteriome included genera greater than 0.1%, relative abundance in at least 75% of the samples within a phenological stage. All of our downstream analyses used genera that met our core taxa cutoffs in at least one phenophase. We assessed our core taxa and separated them into three categories: high stability, defined as core member of six or more phenophases, medium stability, core member of three, four, or five phenophases, or low stability, core member of two or fewer phenophases. We determined that of the identified core there were 3 leaf bacterial, 8 leaf fungal, 62 root bacterial, and 22 root fungal core genera that had high stability across phenophases . This suggests that both bacterial and fungal root communities have a substantially greater number of consistent or stable microbial features across the developmental cycle. However, our experimental design did not differentiate between endophytes versus epiphytes and, thus, may have missed some fine resolution microbial community shifts occurring between the endosphere and episphere. There were two bacterial and one fungal genera that were highly stable in both roots and leaves . A phylogenetic analysis of the core genera indicates that both bacterial and fungal root communities were rich in highly stable and phylogenetically diverse core taxa . Root core genera from the bacterial clade Alphaproteobacteria and the fungal family Pleosporomycetidae were all or nearly all binned as highly stable, indicating that genera in these clades were consistently high in relative abundance across all phenological stages.

Medium- and low-stability core genera appear randomly dispersed across the root community phylogeny, with no obvious patterns.However, leaf bacterial and fungal core community phylogenetic trees contained high, medium, and low stability patterns at the class and phylum levels . All core genera in the fungal class Tremellomycetes had medium to high stability. In contrast, all core genera in the fungal class Sordariomycetes had low stability across phenophases and met the defined core cutoffs only during fruit set or mature fruit stages. The leaf taxa within the bacterial class Gammaproteobacteria consisted of genera with high, medium, and low stability across the phenophases. Interestingly, all theGammaproteobacteria were core members of the full flowering or floral bud development microbiomes regardless of their stability in other phenophases. Another distinct phylogenetic pattern observed in the leaf community was genera in the bacterial phylum Actinobacteria that had low or medium stability across all phenophases. However, 95.0% of core genera in the Actinobacteria clade were core during fruit set and/or fruit development. The only exception to this within the Actinobacteria clade was Bifidobacterium, which was associated only with full flowering and was not a core member of fruit set or fruit development microbiomes . Lastly, the leaf bacterial class Betaproteobacteria contains low- to medium-stability core genera with the most dispersed stage associations. Overall, these data indicate that root bacterial and fungal communities have greater stability across phenophases than those of leaves . Additionally, core taxa had phylogenetically related trends within the high-, medium-, and low-stability classifications, indicating that conserved, vertically descended microbial traits may play a role in determining bacterial and fungal associations across phenophases, particularly in above ground leaf tissue.We completed a genus-level differential relative abundance analysis on our list of core taxa that were $0.01% relative abundance and $75% prevalence in one or more phenophases.

Our differential relative abundance analysis can determine finer-scale phenophase associations beyond just classification as a core microbiome member by looking for increases in relative abundance, proportionate to other members of the microbial community , during specific phenophases. Ecologically dominant taxa are predicted to have a proportionately larger contribution to community function. Among all the phenophases, those associated with flowering had striking microbial enrichments, particularly among the leaf bacteria. Acinetobacter was a core member of five phenophases but was significantly enriched during full flowering compared to other phenophases . Acinetobacter had a gradual enrichment from flush and floral bud development to full flowering. This gradual enrichment signature indicates that Acinetobacter was present throughout the year but has a high temporal turnover rate that is in sync with the transitions from flush to floral bud development and then to full flowering. We also observed bacteria that were sharply enriched during full flowering rather than undergoing gradual enrichments over the phenophases that lead up to full flowering . These include Snodgrassella, Frischella, Gilliamella, and Bifidobacterium . The sharp enrichment patterns during full flowering suggest that these taxa were introduced into the community via a dispersal event.We also identified bacterial leaf genera that had significant depletions during floral bud development and/or full flowering . Four Actinobacter genera, Corynebacterium, Dietzia, Georgenia, and Ornithinimicrobium, were significantly depleted during floral bud development and full flowering . Bacillus, Methylobacterium, Romboutsia, and Sphingomonas also significantly decreased in relative abundance during floral bud development and/or full flowering . For all differentially abundant genera, including bacteria and fungi, across all phenophases, see Fig. S5 and Table S4.We performed a network analysis on the foliar bacterial communities from all samples. We focused on significantly enriched and/or depleted populations and populations with direct connections or putative first-degree interactions . The goal of this approach was to give a broad overview of bacterial interactions across phenophases and identify taxa that potentially interact with specific phenophase-enriched taxa and potentially play a role in observed seasonal community compositional shifts. Rhizobium, Sphingomonas, an unknown bacterium, an unknown Bacillaceae , Acinetobacter,and Romboutsia have the highest normalized betweenness centrality scores ranging from 0.110 to 0.187. Betweenness centrality is a proxy for influence within a network because it measures how often a particular node is the shortest connection or bridge between two other nodes. These high betweenness centrality scores and placement within the network indicate that these genera are potentially keystone taxa that may perform a stabilizing role in the microbial communities across phenological transitions and events . Groups of taxa connected by putative positive interactions cluster together to form distinct modules. These modules are separated by putative negative interactions. Our analysis organized bacterial taxa that were enriched in fruit set and fruit development into a single highly connective community module .

This suggests that fruit set- and fruit development-associated microbiomes are compositionally similar and few microbe-microbe interactions change during the transition from fruit set to fruit development. Leaf bacteria associated with flowering also formed a module within the network . Specific bacteria within the fruit set/development and flowering modules also interact with taxa that were enriched in the other four phenophases, blueberries in pots which cluster together into a third module . Overall, these predicted positive interactions represent inter- or codependent microbe-microbe relationships, and the putative negative interactions indicate potential direct or indirect competition. These predicted microbe-microbe interactions within the microbiome likely affect community compositionin addition to the exogenous influences of abiotic environmental conditions and biotic host physiological factors .The majority of studies examining how plant developmental stage affects the plant’s microbiome have focused on bacteria associated with the rhizosphere of annuals or herbaceous perennials such as maize , rice , sorghum , wheat , Arabidopsis , and Boechera . These important studies indicate that rhizosphere-associated microbiomes can shift in association with plant developmental stages in both domesticated and wild plants that have short-lived aboveground tissues. Studies of the endophytic xylem sap microbiome in grapevine, a deciduous perennial, also showed that microbial shifts were linked to changes in phenological stage . However, much less is known about how overall plant phenology affects above- and belowground microbiomes of evergreen woody perennials that have lifespans that can be decades long and can retain their leaves for multiple years, compared to annuals or deciduous perennials that produce and shed all their leaves each season. Here, we investigated microbiome compositional dynamics in above and below ground tissues of mature 20-year-old Citrus sinensis trees to determine if temporal microbiome fluctuations were associated with host phenological events. The unique contribution of our research was the separation of leaf development from tree phenology. We did this by analyzing the changes in the foliar microbiome on fully mature leaves, which developed in the leaf cohort from the previous year, in relation to the phenological stages of the current year. Thus, the leaves were exposed to the same starting inoculum, minimizing the bias of any potential priority effects . Our results indicate that the phyllosphere microbiome has an active and dynamic relationship with host phenology. More specifically, microbial shifts occurred as trees transitioned from the spring leaf flushing stage and entered flowering. The transition from spring flush to floral bud development and full flowering aligns with important transitions in source-to-sink transport of photosynthate in the tree . During foliar flushing periods, young leaves are a primary carbohydrate sink as they rapidly expand and mature. This source-to-sink transport of photosynthate shifts during floral bud break and development, when mature leaves transition to serve as source tissues to developing floral tissues that are also primary sink tissues. In addition to changes in source-to-sink transport, there are also significant changes in water dynamics within the canopy of the tree associated with full flowering. Flowers have the highest transpiration rate of the tree even compared to the leaves, which drastically increases the amount of water being transported into the overall canopy of the tree . Interestingly, the significant shift in overall foliar community composition from flushing to full flowering was not coupled with a change in species richness, indicating that the same taxa were present, just in different relative abundances in relation to one another. This demonstrates that foliar microbiome assemblage is changing in sync with tree physiology and development. Empirical data, including presence/absence and relative abundance, can also be used to infer patterns or microbial enrichments and/or depletions relative to othertaxa in the community, as well as ecological mechanisms that contribute to plant microbiome assembly, such as microbial species turnover and dispersal . Interestingly, microbial enrichment and depletion patterns of specific taxa suggest that microbial species turnover and dispersal events within the citrus microbiome occur in sync with phenological stage transitions. These enrichment/depletion patterns for specific taxa were more apparent in leaves than in the root compartment. Specifically, the bacterial genus Acinetobacter was enriched in leaves as trees transitioned from spring flush to floral bud development and peaked in relative abundance during full flowering, which is when mature leaves shift toward becoming source tissues for developing flowers and fruit. This may create a microenvironment that selects for an increase in relative abundance of these taxa when carbohydrate is translocating out of the leaves.