Metataxonomic analysis revealed the lack of long-term effects of microbial soil amendments on apple rhizosphere microbial community structure

Abstract

Apple Replant Disease (ARD) symptoms usually appear shortly after planting when apple trees are replanted in soils where the same or a closely related species has grown previously. In an orchard study, we showed that amendment with AMF and biocontrols resulted in significantly better tree development. In the same experiment, rhizosphere soils were sampled three times and subjected to 16S and ITS amplicon sequencing. In this paper, we report the long-term effects of microbial amendment on rhizosphere community structure and relative abundance of specific microbial taxa. Apple rhizosphere microbial communities varied greatly with spatial locations and sampling time. There was a lack of long-term effects of microbial soil amendment on the overall microbial structure in the apple rhizosphere. Microbial amendment significantly affected the relative abundance of many microbial taxa, most of which had very low abundance. This finding implies that improved girth expansion associated with AMF and biocontrol amendments are likely attributable to changes in specific taxa within the rhizosphere rather than community-wide changes. Predominance of spatial and temporal effects on the rhizosphere microbiome indicates the importance of controlling spatial heterogeneity and temporal dynamics in soil microbial communities when designing microbiome field experiments.

Introduction

Young apple trees in replanted orchards may potentially encounter poor establishment, a well-known problem worldwide (Mazzola and Manici, 2012), commonly known as Apple Replant Disease (ARD). ARD symptoms include stunting and shortened internodes aboveground, root tip necrosis and reduced root biomass. Although the growth rates of surviving trees may be similar to those of unaffected trees, they may start bearing fruit some 2–3 years later and yields can be reduced by up to 60% for the duration of the tree’s commercial life (Willett et al., 1994; Pound et al., 2013). ARD is a complex disease, primarily caused by a group of microorganisms including fungi (Ascomycota and Basidiomycota) and fungus-like eukaryotes (Oomycota), which can be further exacerbated by the presence of root-feeding nematodes (Mazzola and Manici, 2012). Rhizosphere and endosphere bacteria can also be associated with ARD, with disease often linked to bacterial community shifts that may promote or exacerbate the pathogenic complex (Van Horn et al., 2021). Controlling all three ARD components (oomycetes, fungi, and nematodes) leads to the best root development (Tilston et al., 2020). Other factors, such as soil properties and phytoalexin biosynthesis, can also affect ARD development (Mao and Wang, 2019; Winkelmann et al., 2019).

Current ARD management strategies are generally based on the principles of exclusion; crop rotation and tree placement within an orchard, including replanting in the former grass aisles (Rumberger et al., 2004; Deakin et al., 2019), and using genetically different rootstocks (Deakin et al., 2019). However, planting in the alleyway and rotating rootstocks are rarely practised. Planting orchards in virgin sites, in the long term, is not practical (Mazzola and Gu, 2000) on a commercial scale and economically unattractive to both growers and nurseries. To replace traditional broad-spectrum fumigants, alternative management strategies have been proposed, including modification of indigenous soil microbial communities (Yao et al., 2006) and biofumigants (Wang and Mazzola, 2019). In a recent field study (Passey et al., 2026), we showed that amendment with arbuscular mycorrhizal fungi (AMF) and biocontrols at planting resulted in significantly better tree development, leading to a respective 11% and 7% increase in the annual rate of girth expansion over four years following planting.

The apple rhizosphere microbiome is known to affect crop development and varies greatly with locations (McLean et al., 2025) and, to some extent, with rootstock genotypes (Deakin et al., 2019). Amending soils with beneficial microbial strains can improve plant development, increase plant tolerance to biotic and abiotic stresses and, at the same time, may affect soil microbial communities. Inoculation with PGPR and AMF improved the uptake of nutrients and the synthesis of secondary metabolites in Isatis indigotica with associated changes in the structure of the rhizosphere microbiome and root metabolomes (Yang et al., 2025b). Brassicaceae seed meal soil amendments not only provide significant control of ARD but also promote a rhizosphere microbiome that may contribute to long-term orchard soil health (Somera et al., 2021). In contrast, in potted apple trees AMF/PGPR amendment did not significantly improve tree development, although they affected relative abundance of many specific microbial taxa, particularly fungi in the rhizosphere (Robinson-Boyer et al., 2024). The long-term effects (over multiple years) of one-off soil amendment with specific beneficial microbial strains on the apple rhizosphere needs to be further investigated.

In the same experiment in which we showed that amendment with AMF and biocontrol at planting results in significantly better tree development (Passey et al., 2026), rhizosphere soils were sampled three times: 1.5, 2.5 and 3.5 years after amendment. In this paper, we analyse 16S and ITS amplicon sequence data of these rhizosphere soil samples to investigate the long-term effects of microbial amendment on rhizosphere community structure and relative abundance of specific microbial taxa.

Materials and methods

Experimental design and setup

Detailed experiment layout, procedures, and assessments, as well as statistical estimation of temporal girth expansion of individual trees, were previously published (Passey et al., 2026); here only a brief description is given. The orchard was replanted with cultivar Gala (grafted to M9 rootstock) in April 2020 on the site of an old apple orchard (51.17’12.78° N, 0°.27’55.80 E) that had been grubbed in March 2020, to induce ARD development; the soil was pH7 with a sandy loam texture and 2.9% organic matter (tested by NRM, part of Cawood Scientific ltd, England). Trees were planted in two blocks; out of each previous block of 12 x 12 (12 rows of 12 trees), 10 x 10 positions were used for a Latin square design, as there were 10 treatments in total. An additional “guard row” of 10 trees was planted in each block in the old alleyway between the old rows 11 and 12.

Eight of the 10 treatments were from a full factorial design of three factors, each with two levels (yes/no): (1) a single-strain AMF of Diversispora sp. isolated at Niab East Malling (England); (2) two biocontrols (formulated Trichoderma and Bacillus commercial products: T. harzianum T-22 [Trianum P, Koppert]; and B. subtilis [later classified as B. amyloliquefaciens, Serenade, Bayer]); (3) two PGPR strains (Azospirillum brassilense sp245 and Gluconacetobacter diazotrophicus Pal5). In addition, two more treatments were included: an AMF product (Rootgrow™) with a five strain (species) mix, and biofumigant Brassica carinata pellets (Biofence; applied to planting position two weeks prior to planting). All AMF and PGPR products were commercially available and supplied directly from the manufacturer “PlantWorks Ltd.” (Sittingbourne, England). The two commercial biocontrol products (Trianum P and Serenade) were purchased commercially from distributors in the UK. Biofence was supplied by “Tozer Seeds” (Cobham, England). All amendments were applied via liquid application, except for the vermiculite-based AMF products, at the time of planting. For treatments including liquid products, the biocontrols and/or PGPR were reapplied to those trees allocated to the relevant treatments in every other row of both blocks in early May 2022 and again in late April 2023. Thus, half the trees for these treatments were inoculated just once (at planting) and the other half were reinoculated twice.

The experimental orchard followed standard UK commercial management practices, including routine foliar nutrient and disease control (primarily for scab and powdery mildew) sprays. No products were applied directly into soil to control plant pathogens except the two biocontrol products used for amendment. Weeds within the herbicidal strip (1.5 m wide)?? under the trees were suppressed annually using a 2,4-D and glyphosate mix (2020–2023) and propyzamide (2024). Herbicides were applied at least one month prior to or post any soil amendments: the initial microbial products in 2020 and subsequent liquid re-inoculations (PGPR/biocontrols) in 2022 and 2023. Tree bases were occasionally cleared by hand weeding or strimming when necessary.

Sampling rhizosphere soil

Rhizosphere soils were sampled on three occasions: late September 2021 (all 100 experimental trees in one block), mid-October 2022 (all 100 trees in the other block), and mid-October 2023 (6 rows from each of the two blocks for 120 trees total). Fine roots (< 2 mm diameter, non-woody, feeder roots) from the zone where the propagules of soil-borne plant pathogens are present at greatest density (i.e. ~10–20 cm depth and 20–50 cm from the trunk) were exposed and excised around each tree. Old and dying roots were not sampled to reduce the confounding incidence of secondary saprotrophic organisms (Tewoldemedhin et al., 2011). Excised roots were shaken free of loosely attached soil, and the remaining adhering soil was considered ‘rhizosphere’ soil, which was removed and collected in sealable polythene bags. Bags were immediately transferred to a 5 °C cold store on returning from the orchard and within 48 hours aliquots of ~1 g of rhizosphere soil were weighed and stored in a -20 °C freezer prior to DNA extraction.

DNA extraction and amplicon sequencing

DNA was extracted from ~1g of rhizosphere soil samples following protocols for 250-1000mg of soil from the E.Z.N.A Soil DNA Kit (Omega Bio-tek), with the modification of using a Geno/Grinder 2010 (SPEX CertiPrep) for 2 minutes at 1500 rpm to lyse samples. The optional isopropanol precipitation and extra cHTR reagent steps were included in this protocol. Extracted DNA was finally eluted in 100 μl 10mM Tris-HCl (pH 8.5). The quantity and purity of extracted DNA were measured by NanoDrop1000 (Thermo Fisher Scientific). Subsequent PCR amplification, library preparation and amplicon sequencing was carried out by Novogene UK (Cambridge, UK). For fungi, the target primers were ITS1-1F [ITS1-1F-F: 5’-CTTGGTCATTTAGAGGAAGTAA-3’ (Gardes and Bruns, 1993), and ITS1-1F-R: 5’-GCTGCGTTCTTCATCGATGC-3’ (White et al., 1990); for bacteria, the target primers were for 16S region V4 [515Fmod: 5’-GTGYCAGCMGCCGCGGTAA-3’ (Parada et al., 2016) and 806R: 5’-GGACTACNVGGGTWTCTAAT-3’ (Caporaso et al., 2011). All samples were sequenced on the Illumina NovaSeq platform in the 250 nt paired end mode.

Bioinformatic analysis

Raw FASTQ reads were filtered to remove sequences containing ambiguous bases within primer regions or incorrect forward–reverse primer combinations. Primer and barcode sequences were subsequently trimmed from retained reads. Amplicon sequence variants (ASVs) were inferred using the DADA2 pipeline (v1.34) implemented in R (Callahan et al., 2016). Separate processing workflows were applied for ASV inference and frequency table construction. Reads containing unknown bases (N) were discarded, and quality filtering was applied with a maximum expected error threshold of two for both forward and reverse reads to generate the ASV inference dataset. Error models were learned from dereplicated reads, and sequence inference was performed using both the filtered reads and an unfiltered copy of the dataset (used for frequency table generation), with the pooling option set to pseudo. Forward and reverse reads were merged, ASVs inferred, and chimeric sequences removed from both datasets using the removeBimeraDenovo function. Final frequency tables were produced by filtering and relabelling ASVs from the unfiltered tables to retain only those present in the filtered ASV set.

Taxonomic assignment was carried out using the SINTAX algorithm implemented in USEARCH (Edgar, 2016). Fungal sequences were classified using a curated version of the Warcup V2 database, updated to align with NCBI taxonomy as of 2 September 2022. Bacterial taxonomy was assigned using the RDP training set version 18 for 16S rRNA gene sequences. A confidence threshold of 80% was applied for all taxonomic assignments.

Statistical analysis

This study aimed to quantify the effect of soil amendments (primarily microbial products) on microbiome composition in the rhizosphere of apple trees relative to other non-treatment factors, namely locations (orchard blocks, row and tree [column] within a block), sampling time and residual errors. It should be noted that time effects were partially confounded with block effects since samples in 2021 and 2022 were obtained from separate blocks. However, this should not affect the total contribution from treatment and non-treatment sources to the observed variability in microbiome composition (measured as the total sum of squares). Preliminary analysis showed that the contribution of re-inoculation of liquid products was insignificant, accounting for < 1% of the total variability, and hence this factor was ignored in all statistical analyses.

Treatment effects on alpha (α – within sample) and beta (β – between sample) diversity indices were assessed using three statistical models: (i) ‘y ~ Block/(Row + Tree) + Time*Treatment’ for all 10 treatments, (ii) ‘y ~ Block/(Row + Tree) + Time*AMF*PGPR*Biocontrol’ for the 2x2x2 factorial design data, and (iii) ‘y ~ Block/Row + Time’ for those untreated samples. In all cases, y is the dependent variable of interest.

The R package vegan version 2.5-6 (Oksanen et al., 2019) was used to calculate α diversity indices from the original count data: species richness (ACE) and two specific diversity indices (Shannon and Simpson). Treatment effects on α diversity indices were assessed by permutational ANOVA of the index ranks with the R package lmPerm v2.1.0 (Wheeler, 2016). Non-metric multidimensional scaling (NMDS) of the Bray-Curtis (β diversity) indices (of the DESeq2 normalised ASV count data) was carried out with vegan. Permutational multivariate analysis of variance (PERMANOVA) through the “adonis2” function in the vegan package was used to quantify treatment effects based on the sequential permutation test (Type I Sum of Squares). Because of the confounded effects between Block and Time factors, the contribution by the Block factor was overestimated at the expense of the Time factor. In addition, Principal Component Analysis (PCA) were applied to the normalised ASV counts data with the DESeq2 variance stabilising transformation method to ensure data was approximately homoscedastic (Anders and Huber, 2010). As for diversity indices, ANOVA was then performed to quantify the treatment effects to the observed variability in the PC scores.

DESeq2 analyses were applied to compare the relative abundance of individual ASVs for the 2x2x2 factorial design data for AMF, PGPR and Biocontrol effects within the framework of generalised linear modelling, assuming a negative binomial distribution for residuals (Love et al., 2014). The significance of differentially abundant ASVs was based on a Wald test with the p-value adjusted by the Benjamini-Hochberg (BH) method (Benjamini and Hochberg, 1995). DeSeq2 analysis was applied to all taxonomy ranks for fungi (phylum, class, order, family, genus, species and ASV). For bacteria species rank was not available. When aggregating ASVs into higher taxonomy ranks, ASVs with < 80% confidence in assigning a taxonomy identification were excluded.

Results

General amplicon sequencing and taxonomy

All trees established and survived following planting until 2023, when 13 trees had died, but the number of dead trees showed no relationship with AMF, PGPR or biocontrol treatment. Only a few samples failed amplicon sequencing. There were 98, 98 and 115 samples with reads of sufficient quality for 2021, 2022 and 2023, respectively, including corresponding 10, 9 and 12 (11 for 16S) untreated samples from 2021, 2022 and 2023.

The number of ITS raw reads per sample ranged from 11,340 to 139,174 (median = 82,342); and there were 3015 fungal amplicon sequence variants (ASVs), with the number of raw reads per ASV ranging from 43 to 1,361,453 (median = 233). About 62%, 51%, 40%, 32%, 28% and 17% of these ASVs were assigned to the taxonomy rank of phylum, class, order, family, genus and species, respectively, at the 80% confidence level; the corresponding values for the percentage of normalised sequence reads were 94%, 89%, 84%, 72%, 68% and 33%. Irrespective of sampling time, most fungal ASV (or reads) belong to Ascomycota, with Mucoromycota and Basidiomycota having similar prevalence (Supplementary Figure S1A). Reads from 2023 samples had a much higher prevalence of Ascomycota at the expense of Mucoromycota. At the class level, there was a much higher prevalence of Sordariomycetes in the 2023 samples at the expense of Mortierellomycetes and Tremellomycetes. Two Fusarium ASVs were among the most abundant fungal ASVs – accounting for 9.4% (first) and 3.4% (third) of the total number of reads. The second most abundant ASV was from Mortierella alpina, accounting for 4.1% of the total number of reads.

The number of 16S raw reads per sample ranged from 12,569 to 94,932 (median = 39,411). There were 18,163 bacterial ASVs, with the number of raw reads per ASV ranging from 22 to 101,477 (median = 122). About 83%, 68%, 44%, 32% and 32% of these bacterial ASVs were assigned to the taxonomy rank of phylum, class, order, family, and genus, respectively, at the 80% confidence level; the corresponding values for the percentage of sequence reads were 94%, 86%, 56%, 48% and 48%. Irrespective of sampling time, most of the bacterial ASVs (or reads) belonged to Proteobacteria with Actinobacteria and Acidobacteria as the distinct second and third (Supplementary Figure S1B). Reads from 2023 samples had a much higher prevalence of Acidobacteria and Verrucomicrobia at the expense of Proteobacteria. The most abundant bacterial ASV is from the Bradyrhizobium genus, accounting for 0.8% of reads, whereas the next three most abundant ASVs all accounted for ca. 0.7% of reads (Spartobacteria genera incertae sedis, Mycetohabitans and Sphingomonas).

Within-sample (α diversity)

A summary of statistical analyses of α diversity indices is given in Table 1 for three model formulations: all treatments, only those in the 2x2x2 factorial design, and only the untreated samples. There are several results standing out from these summaries. First, variation in the rhizosphere microbiome was driven primarily by local temporal and spatial heterogeneity (between and within blocks) and sampling time, with effects typically at least tenfold greater than those of amendment treatments (Table 1). Second, fungal species richness is more variable across the three time points compared to bacterial species richness, opposite to the within-sample compositional indices (Shannon and Simpson indices). Third, for fungi, there were large reductions in the within-sample diversity indices in 2023 (Figure 1), which is opposite to the bacterial communities (Figure 2). This large differences in species richness among three years, particularly between 2023 and the other two years, are illustrated in Figure 3 where the number of taxa in the core microbiome for each year and for common core taxa across all three years are plotted against occupancy (%) threshold values used to define core microbiome members. Figure 3 also shows that bacterial and fungal species richness varied with years in the opposite direction. There were significant changes in the within-sample diversities of apple rhizosphere microbiomes over time for those untreated samples, particularly for bacteria (Table 1).

Figure 1

Figure 2

Figure 3

None of the treatment effects on the three fungal within-sample diversity indices were statistically significant (Table 1). For all three bacterial within-sample diversity indices, the overall differences among the 10 treatments and the main effect of AMF amendment (Table 1) were statistically significant. Applying AMF reduced bacterial within-sample indices. In addition, the interaction between AMF and PGPR amendment on the bacterial Shannon and Simpson indices was also statistically significant but the effect was limited (Table 1).

Between sample (β diversity)

Similar results were obtained for between-sample diversity indices as for within-sample diversity indices, namely, a much larger contribution to the observed variability in the Bray-Curtis indices by spatial and temporal factors than by amendment treatments (Table 1). Figure 4 illustrates the temporal differences in the rhizosphere microbial communities based on the NMDS analysis: fungal samples fell into three distinct clusters delineated by sampling time. When all 10 treatments were considered together, there was evidence for treatment effects on between-sample diversity indices for both fungi (P < 0.01) and bacteria (P < 0.001) (Table 1). However, when only the 2x2x2 factorial design data were analysed, none of the terms involving treatment(s) were statistically significant (Table 1). There were distinct changes in microbial communities over time for untreated samples, particularly for fungi (Table 1). Similar results were obtained when principal component scores were used.

Figure 4

Amendment treatment effects on the relative abundance of individual ASVs

Differential abundance analysis was only applied to the 2x2x2 factorial design focusing on the main effects of AMF, PGPR and biocontrol amendments. Table 2 summarises differential abundance analysis at different ranks for each of the main treatment effects (namely, the comparison between those trees with or without a specific amendment).

Overall, only a small proportion of microbial groups at a given taxonomy rank were affected by a specific amendment treatment (Table 2), especially for bacteria. For instance, out of the 18,159 bacterial ASVs subjected to DESeq2 analysis, AMF, biocontrol, and PGPR treatments significantly affected the abundance of only 51, 45 and 25 ASVs, respectively. For fungi, the relative abundance of 73, 104 and 59 of the 2,999 ASVs were affected by the respective AMF, biocontrol, and PGPR treatments. Biocontrol amendment appeared to affect fungal ASVs more than AMF or PGPR amendment, whereas PGPR appeared to affect abundance for the least number of bacterial ASVs. Only a few of those ASVs whose abundance was significantly affected by amendment had both large average abundance (> 2.3 at the ln-scale) and absolute LFC values (> 5.0) (Figure 5). We focus on those affected ASVs with a mean abundance of at least 10 for bacteria (2.3 on the ln scale) or 20 for fungi (3.0 on the ln scale), and with the taxonomy assigned to at the least the genus rank. Different threshold values of abundance were used for bacterial and fungal ASVs because the median abundance was 122 and 233 for bacterial and fungal ASVs, respectively. The summary of all individual ASVs with their abundance significantly affected by amendment treatments is given in Supplementary Tables S1-S6, showing that most of these ASVs with their relative abundance affected by amendments had a very low mean abundance, often < 1.0.

Figure 5

Of those bacterial ASVs significantly affected by AMF amendment, only two can be assigned to the genus rank and have a mean abundance of > 10.0 (Supplementary Table S1). AMF amendment increased the abundance of an ASV from Chujaibacter (LFC = 6.48) and decreased the abundance of an ASV from Spartobacteria genera incertae sedis (LFC = -0.33). AMF amendment increased the abundance of many fungal ASVs, including several ASVs that can be assigned to the genus rank with a mean abundance of > 20: ASVs from Saitozyma podzolica (LFC = 0.71), Apiospora arundinis (LFC = 0.72), Penicillium (LFC = 0.57), and Goffeauzyma (LFC = 0.54). AMF amendment reduced the abundance of many other ASVs, including Paraglomus laccatum (LFC = -1.18) and Talaromyces rugulosus (LFC = -1.65) (Supplementary Table S2).

Abundances of 25 bacterial ASVs were affected by PGPR amendment, including increased abundance of an ASV from Devosia (LFC = 0.31) and an ASV from Chujaibacter (LFC = 7.54) (Supplementary Table S3). PGPR amendment increased the abundance of three ASVs with an abundance of > 10.0 from Agaricomycetes (Supplementary Table S4), but also decreased the abundance of many fungal ASVs, including Alnicola (LFC = -1.35) and Candolleomyces candolleanus (LFC = -1.57) (Supplementary Table S4).

Biocontrol amendment did not affect the abundance of any bacterial ASV with a mean abundance of > 10.0 (Supplementary Table S5). In contrast, it affected many fungal ASVs with a mean abundance of > 20.0 (Supplementary Table S6), including increased abundance of ASVs from Fusarium (LFC = 0.42), Alnicola (LFC = 1.67), Linnemannia (one ASV with LFC = 0.98, and another with LFC = 1.30), Alternaria (LFC = 0.75), Tomentella (LFC = 2.62) and Hymenoscyphus (LFC = 9.84) (Supplementary Table S6). The abundance of a Trichoderma ASV (mean abundance = 10.99) was also increased (LFC = 3.36) by biocontrol (Trichoderma and Bacillus) amendment. Abundances of several fungal ASVs with a mean abundance of > 20.0 were decreased by biocontrol amendment, including Pseudogymnoascus pannorum (LFC = -1.00) and Candolleomyces candolleanus (LFC = -1.37).

Discussion

We amended soil at the planting hole with microbial products (AMF, PGPR and biocontrols) at the time of apple planting and previously demonstrated that there is a significant improvement in tree growth following AMF and biocontrol amendments (Passey et al., 2026), consistent with recent research studies (Jiang et al., 2025; Yang et al., 2025a). However, this study demonstrates that such amendment treatments had a very limited impact on the overall microbial communities in the apple rhizosphere within the four years following the amendment, despite it affecting the relative abundances of specific microbial taxa. Instead, a considerable amount of the observed variability in rhizosphere microbial communities resulted from differences across the three sampling times and spatial heterogeneity.

In the present study, temporal differences in rhizosphere microbial communities also contained spatial effects, since the differences between 2021 and 2022 were confounded by block effects. Despite this, the temporal differences are expected to be large and far greater than the amendment effects. Large differences are evident in the rhizosphere microbial communities (both alpha and beta diversities) in 2023 compared with those in 2021 and 2022. Seasonality is unlikely to be the reason for the observed year-to-year differences since all rhizosphere samples were taken in autumn (post-harvest). Many of the temporal differences were between 2023 and the other two years, even though the 2021 and 2022 samples were from different orchard blocks. For fungal communities, within-sample diversity (including species richness) was much reduced in 2023, which is opposite to the bacterial communities. These differences most likely resulted from a record-breaking hot and dry June and a September heatwave: a seven-day run of temperatures exceeding 30 °C in early September, a UK record for the month, contributing to the joint-warmest September on record. Soil moisture might have also played a role: less dispersal and mixing of bacterial communities in relatively dry conditions. The present results may suggest that fungi suffered more from the hot conditions experienced in 2023 than bacteria, consistent with published results (Pietikäinen et al., 2005). Temporal change in soil fungal communities was shown to be slightly more rapid (namely, higher temporal turn-over) than bacterial communities (Martinović et al., 2021); however, the opposite trend was found in other studies (Stevenson et al., 2014; Joos et al., 2025).

In addition to seasonal changes in climatic conditions and in soil physicochemical properties (moisture, temperature, pH, and nutrient levels), agricultural management (including cropping systems, plant species, specific genotype, and soil management) can drive temporal changes in soil microbial communities. Microbial communities in agricultural soils are more variable over time than in non-agricultural land, with the changes correlated with soil moisture and temperature (Lauber et al., 2013). Soil bacterial communities are more shaped by time, whereas fungal communities are shaped by specific plant species (Hannula et al., 2019); for example, cultivation of runner bean (Phaseolus coccineus) affects the structure of the soil microbial community over time (Stavridou et al., 2022). Temporal changes in soil microbial communities may be better explained by their functional responses to environmental factors (Martinović et al., 2021). Inherent temporal dynamics in microbial communities as well as associated plant development dynamics make it difficult to draw reliable conclusions on the effects of long-term agricultural management practices on the soil microbiota (Fernandez-Gnecco et al., 2022). In this study, the large temporal variation in the soil microbial community raises important questions regarding how to assess treatment effects on microbial communities in perennial plant species and how to define the core microbiome. Sampling time needs to be carefully considered in future studies for interpreting field data and for comparative synthesis or meta-analysis.

We found that spatial location was the primary factor affecting both fungal and bacterial communities (α-diversity and β-diversity). Site is well known to be the dominant factor in apple-associated microbial communities, including in bulk soil (Lauber et al., 2009), rhizosphere (Lauber et al., 2009), and endophytes in the above-ground parts of apple trees (Liu et al., 2020), consistently dwarfing the contribution by rootstock and/or scion genotype. Differences in soil (specifically in the rhizosphere microbiome) among sites over a large distance may arise from multiple associated factors, including local variation in climate, soil characteristics, and vegetation. The present study demonstrated large variation in apple rhizosphere microbial communities between neighbouring orchard blocks, as well as between neighbouring tree rows within an orchard block, consistent with previous observations in a UK cider orchard (Deakin et al., 2019). This fine-scale variability illustrates the importance of experimental design as well as sufficient biological replications both within and between sites for drawing conclusions that may be extrapolated with confidence.

Whilst amendment treatment with microbial products had limited impacts on apple rhizosphere communities, changes in the abundance of specific microbial taxa could be linked to the amendment. This overall finding is consistent with our two other small-scale amendment studies: one an orchard study (Cook, 2022) and the other a potted tree study (Robinson–Boyer et al., 2024). This may be because this amendment, containing only a few specific strains and applied directly around the roots at planting, only results in transient effects on rhizosphere microbial communities, or does not affect the overall community structure at all. The limited long-term impact of microbial amendments on the rhizosphere microbiome may have resulted from the application frequency in the present study: liquid biocontrol and PGPR products were applied annually, while AMF was applied only once during planting. This protocol was adopted to align with the most practical and cost-effective method currently favoured by the apple industry in the UK. The present finding implies that the improved girth expansion associated with AMF and biocontrol amendment (Passey et al., 2026) are probably attributable to changes in some specific taxa within the rhizosphere rather than community-wide changes. Similarly, the reduction in ARD severity associated with Brassicaceae seed meal amendment is attributed to changes in a few specific taxa: reduced abundance in putative ARD pathogens and increased abundance in several specific beneficial taxa (Somera et al., 2021).

The relative abundance of many taxa was affected by amendment with AMF, PGPR or biocontrols. However, care must be taken when interpreting these affected taxa in relation to the observed benefits on tree development. Firstly, many ASVs cannot be assigned to a sufficiently low taxonomic rank for functional interpretation with confidence. Thus, much of the interpretation had to be selective – based only on those ASVs that can be assigned to low taxonomy rank(s). Secondly, the presence of an organism alone may not necessarily lead to the predicted effect (Somera and Mazzola, 2022). Finally, predicted functions of specific ASVs may not necessarily correlate quantitatively with their relative abundance.

Both AMF and PGPR amendments increased the abundance of the same ASV from Chujaibacter, which may be associated with soil disease-suppression in banana (Fan et al., 2023). PGPR amendment increased the abundance of an ASV from Devosia. Several bacterial genera, including Devosia, are consistently enriched on AMF hyphae, and one specific Devosia strain synergistically acts with AMF on plant roots to promote plant growth, nitrogen uptake, and mycorrhization (Zhang et al., 2024). All amendments, particularly PGPR, increased several ASVs from Agaricomycetes; this group contains both pathogens and beneficial (ectomycorrhizal) microbes. AMF amendment decreased the abundance of an ASV from Paraglomus laccatum, which, though an AMF species of a phylogenetically ancient family, tends to be less beneficial in terms of stimulation of plant growth and nutrient uptake than species of phylogenetically more recent AMF families such as Glomeraceae, Entrophosporaceae, and Diversisporaceae (Säle et al., 2021). The abundance of one ASV from Alnicola sp. is increased by biocontrol amendment but decreased by PGPR amendment. Alnicola has a widespread distribution and contains 60 species that typically form mycorrhizal relationships with species of Alder (Moreau, 2005). Amendment with biocontrol species increased abundance of several other ASVS: (1) an ASV from Tomentella, a widespread genus of ectomycorrhizal fungi forming inconspicuous, often woolly or web-like crusts on wood and soil, crucial for forest nutrient cycling (Zhu et al., 2024); (2) two ASVs from Linnemannia, of which one particular species (L. elongata - formerly Mortierella elongata) is known to stimulate plant growth, such as Arabidopsis thaliana (De Tender et al., 2024); and (3) one ASV from Cladophialophora chaetospira, a dark septate endophytic fungus known for its beneficial symbiotic relationships with crops like strawberries (Fragaria x ananassa) and Chinese cabbage (Brassica rapa pekinensis) (Harsonowati et al., 2020). Interestingly, biocontrol amendment decreased the abundance of an ASV from Pseudogymnoascus pannorum, a common soil fungus that may become an opportunistic human pathogen (Lueschow et al., 2019). Biocontrol treatment also increased the abundance of one Trichoderma ASV, though its mean abundance was low.

Some soil amendments also resulted in significant changes in the abundance of several other ASVs, which are hard to interpret. For instance, biocontrol amendment increased the abundance of one ASV from Fusarium and another from Hymenoscyphus. The former genus contains numerous known plant pathogens, including one causing ARD in China (though not reported in Europe) (Duan et al., 2022), whereas the latter genus contains an ash dieback pathogen (Hietala et al., 2022). AMF amendment increased the abundance of one ASV from Apiospora arundinis, a potential plant pathogen on Polygonatum cyrtonema (Gong et al., 2024).

The present study demonstrates that spatiotemporal dynamics predominates over tested management practices in shaping apple rhizosphere microbial communities and the lack of long-term effects of microbial soil amendments on the overall microbial structure in the apple rhizosphere. This finding emphasises the importance of controlling spatial heterogeneity and temporal dynamics in soil microbial communities when designing microbiome field experiments. Amendments influence the abundance of specific ASVs in the rhizosphere and some of these ASVs may have putative biological interpretations in relation to the observed tree development. However, much of the observed variability in the rhizosphere microbiome remains unexplained, and there was also large variability in tree growth within each individual treatment (Passey et al., 2026). Further research is thus needed to investigate the relationship between tree development and the abundance of individual ASVs within individual trees.

References

• 1

  AndersS.HuberW. (2010). Differential expression analysis for sequence count data. Genome Biol.11, R106–R106. doi: 10.1038/npre.2010.4282.1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2

  BenjaminiY.HochbergY. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. J. R. Stat. Society Ser. B. Stat. Method.57, 289–300. doi: 10.1111/j.2517-6161.1995.tb02031.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3

  CallahanB. J.McMurdieP. J.RosenM. J.HanA. W.JohnsonA. J. A.HolmesS. P.et al. (2016). DADA2: High-resolution sample inference from Illumina amplicon data. Nat. Methods13, 581–583. doi: 10.1038/nmeth.3869

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4

  CaporasoJ. G.LauberC. L.WaltersW. A.Berg-LyonsD.LozuponeC. A.TurnbaughP. J.et al. (2011). Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. PNAS108, 4516–4522. doi: 10.1073/pnas.1000080107

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 5

  CookC. (2022). Understanding the Resilience of Soil Beneficials to Combat Apple Replant Disease (ARD) (Cranfield, England: Cranfield University). PhD.

  • Google Scholar

• 6

  DeakinG.Fernández-FernándezF.BennettJ.PasseyT.HarrisonN.TilstonE. L. E. L.et al. (2019). The effect of rotating apple rootstock genotypes on apple replant disease and rhizosphere microbiome. Phytobiomes J.3, 273–285. doi: 10.1094/PBIOMES-03-19-0018-R

  • CrossRef
  • Google Scholar

• 7

  De TenderC.VandecasteeleM.OmmeslagS.De ZutterN.VandenbusscheE.HaegemanA.et al. (2024). Linnemannia elongata: A key species in chitin-based plant growth promotion. Phytobiomes J.8, 366–377. doi: 10.1094/pbiomes-05-23-0031-r

  • CrossRef
  • Google Scholar

• 8

  DuanY. N.JiangW. T.ZhangR.ChenR.ChenX. S.YinC. M.et al. (2022). Discovery of Fusarium proliferatum f. sp. malus domestica causing apple peplant disease in China. Plant Dis.106, 2958–2966. doi: 10.1094/pdis-12-21-2802-re

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9

  EdgarR. C.2016SINTAX: a simple non-Bayesian taxonomy classifier for 16S and ITS sequences. bioRxiv, 074161–074161. doi: 10.1101/074161

  • CrossRef
  • Google Scholar

• 10

  FanH.HeP.XuS.LiS.WangY.ZhangW.et al. (2023). Banana disease-suppressive soil drives Bacillus assembled to defense Fusarium wilt of banana. Front. Microbiol.14. doi: 10.3389/fmicb.2023.1211301

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11

  Fernandez-GneccoG.CovacevichF.ConsoloV. F.BehrJ. H.SommermannL.MoradtalabN.et al. (2022). Effect of long-term agricultural management on the soil microbiota influenced by the time of soil sampling. Front. Soil Sci.2. doi: 10.3389/fsoil.2022.837508

  • CrossRef
  • Google Scholar

• 12

  GardesM.BrunsT. D. (1993). ITS primers with enhanced specificity for basidiomycetes - application to the identification of mycorrhizae and rusts. Mol. Ecol.2, 113–118. doi: 10.1111/J.1365-294X.1993.TB00005.X

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13

  GongZ.YangY.ZhangL.LuoQ.LuoJ.ZhangY. (2024). First report of Apiospora arundinis causing leaf spot on Polygonatum cyrtonema in China. Plant Dis.108, 524. doi: 10.1094/pdis-08-23-1580-pdn

  • CrossRef
  • Google Scholar

• 14

  HannulaS. E.KielakA. M.SteinauerK.HubertyM.JongenR.LongJ. R. D.et al. (2019). Time after time: Temporal variation in the effects of grass and forb species on soil bacterial and fungal communities. mBio10, 10.1128/mbio.02635–02619. doi: 10.1128/mbio.02635-19

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15

  HarsonowatiW.MarianM.SuronoNarisawaK. (2020). The effectiveness of a dark septate endophytic fungus, Cladophialophora chaetospira SK51, to mitigate strawberry fusarium wilt disease and with growth promotion activities. Front. Microbiol.11, 585. doi: 10.3389/fmicb.2020.00585

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16

  HietalaA. M.AganA.NagyN. E.BørjaI.TimmermannV.DrenkhanR.et al. (2022). The native Hymenoscyphus albidus and the invasive Hymenoscyphus fraxineus are similar in their necrotrophic growth phase in ash leaves. Front. Microbiol.13. doi: 10.3389/fmicb.2022.892051

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17

  JiangW.ChenR.SongL.QinL.XuX.LiX.et al. (2025). From metabolic fingerprints to field solutions: Engineering the apple rhizosphere microbiome via host-directed Bacillus recruitment for sustainable apple replant disease control. Microbiome14, 43. doi: 10.1186/s40168-025-02301-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 18

  JoosL.OmmeslagS.BaeyenS.AsselbergW.LooK. V.ClementL.et al. (2025). Year-long, multiple-timepoint field studies show the importance of spatiotemporal dynamics and microbial functions in agricultural soil microbiomes. mSystems10, e00112-00125. doi: 10.1128/msystems.00112-25

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19

  LauberC. L.HamadyM.KnightR.FiererN. (2009). Pyrosequencing-based assessment of soil ph as a predictor of soil bacterial community structure at the continental scale. Appl. Environ. Microbiol.75, 5111–5120. doi: 10.1128/AEM.00335-09

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20

  LauberC. L.RamirezK. S.AanderudZ.LennonJ.FiererN. (2013). Temporal variability in soil microbial communities across land-use types. ISME J.7, 1641–1650. doi: 10.1038/ismej.2013.50

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21

  LiuJ.RidgwayH. J.JonesE. E. (2020). Apple endophyte community is shaped by tissue type, cultivar and site and has members with biocontrol potential against Neonectria ditissima. J. Appl. Microbiol.128, 1735–1753. doi: 10.1111/JAM.14587

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 22

  LoveM. I.HuberW.AndersS. (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol.15, 550–550. doi: 10.1186/s13059-014-0550-8

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23

  LueschowS. R.JohnsonL. J. A. N.WilliamsT. F.McCleeryR.Porras-AlfaroA. (2019). Opportunistic Pseudogymnoascus and Geomyces species isolated from humans and bats. Sydowia71, 25–33.

  • Google Scholar

• 24

  MaoZ.WangY. (2019). “ Apple replant disease: causes and management,” in Integrated Management of Diseases and Insect Pests of Tree Fruit. Eds. XuX.FountainM. ( Burleigh Dodds Science Publishing, Cambridge), 39–57.

  • Google Scholar

• 25

  MartinovićT.OdriozolaI.MašínováT.Doreen BahnmannB.KohoutP.SedlákP.et al. (2021). Temporal turnover of the soil microbiome composition is guild-specific. Ecol. Lett.24, 2726–2738. doi: 10.1111/ele.13896

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26

  MazzolaM.GuY. H. (2000). Impact of wheat cultivation on microbial communities from replant soils and apple growth in greenhouse trials. Phytopathology90, 114–119. doi: 10.1094/phyto.2000.90.2.114

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 27

  MazzolaM.ManiciL. M. (2012). Apple replant disease: Role of microbial ecology in cause and control. Annu. Rev. Phytopathol.50, 45–65. doi: 10.1146/annurev-phyto-081211-173005

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 28

  McLeanH.MikaberidzeA.DeakinG.XuX.Papp-RuparM. (2025). The interplay between scion genotype, root microbiome, and Neonectria ditissima apple canker. FEMS Microbiol. Ecol.101. doi: 10.1093/femsec/fiaf014

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29

  MoreauP. A. (2005). A nomenclatural revision of the genus Alnicola (Cortinariaceae). Fungal Diversity20, 121–155.

  • Google Scholar

• 30

  OksanenJ.BlanchetG. B.FriendlyM.KindtR.LegendreP.McGlinnD.et al. 2019vegan: community ecology package. doi: 10.32614/cran.package.vegan

  • CrossRef
  • Google Scholar

• 31

  ParadaA. E.NeedhamD. M.FuhrmanJ. A. (2016). Every base matters: Assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples. Environ. Microbiol.18, 1403–1414. doi: 10.1111/1462-2920.13023

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32

  PasseyT.Robinson-BoyerL.JacksonR.ZhouJ.XuX. (2026). Microbial soil amendment to improve apple tree establishment following planting. Technol. Horticutlure. 6, e004. doi: 10.48130/tihort-0025-0039

  • CrossRef
  • Google Scholar

• 33

  PietikäinenJ.PetterssonM.BååthE. (2005). Comparison of temperature effects on soil respiration and bacterial and fungal growth rates. FEMS Microbiol. Ecol.52, 49–58. doi: 10.1016/j.femsec.2004.10.002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 34

  PoundM. P.FrenchA. P.AtkinsonJ.WellsD. M.BennettM. J.PridmoreT. P. (2013). RootNav: Navigating images of complex root architectures. Plant Physiol.162, 1802–1814. doi: 10.1104/pp.113.221531

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35

  Robinson-BoyerL.Papp-RuparM.DeakinG.FaggG.PasseyT.XuX.The effects of microbial amendment and water stress on apple rhizosphere and root microbiome. Acta Hortic. 1427, 141–150. doi: 10.17660/ActaHortic.2025.1427.18

  • CrossRef
  • Google Scholar

• 36

  RumbergerA.YaoS.MerwinI. A.NelsonE. B.ThiesJ. E. (2004). Rootstock genotype and orchard replant position rather than soil fumigation or compost amendment determine tree growth and rhizosphere bacterial community composition in an apple replant soil. Plant Soil264, 247–260. doi: 10.1023/B:PLSO.0000047760.13004.94

  • CrossRef
  • Google Scholar

• 37

  SäleV.PalenzuelaJ.Azcón-AguilarC.Sánchez-CastroI.da SilvaG. A.SeitzB.et al. (2021). Ancient lineages of arbuscular mycorrhizal fungi provide little plant benefit. Mycorrhiza31, 559–576. doi: 10.1007/s00572-021-01042-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 38

  SomeraT. S.FreilichS.MazzolaM. (2021). Comprehensive analysis of the apple rhizobiome as influenced by different Brassica seed meals and rootstocks in the same soil/plant system. Appl. Soil Ecol.157, 103766. doi: 10.1016/j.apsoil.2020.103766

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 39

  SomeraT. S.MazzolaM. (2022). Toward a holistic view of orchard ecosystem dynamics: A comprehensive review of the multiple factors governing development or suppression of apple replant disease. Front. Microbiol.13. doi: 10.3389/fmicb.2022.949404

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 40

  StavridouE.KaramichaliI.LagiotisG.PatseaE.OsathanunkulM.MadesisP. (2022). Seasonal shifts in soil microbiome structure are associated with the cultivation of the local runner bean variety around the Lake Mikri Prespa. Biology11, 1595. doi: 10.3390/biology11111595

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41

  StevensonB. A.HunterD. W. F.RhodesP. L. (2014). Temporal and seasonal change in microbial community structure of an undisturbed, disturbed, and carbon-amended pasture soil. Soil Biol. Biochem.75, 175–186. doi: 10.1016/j.soilbio.2014.04.010

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 42

  TewoldemedhinY. T.MazzolaM.MostertL.McLeodA. (2011). Cylindrocarpon species associated with apple tree roots in South Africa and their quantification using real-time PCR. Eur. J. Plant Pathol.129, 637–651. doi: 10.1007/s10658-010-9728-4

  • CrossRef
  • Google Scholar

• 43

  TilstonE. L.DeakinG.BennettJ.PasseyT.HarrisonN.FernándezF.et al. (2020). Effect of fungal, oomycete and nematode interactions on apple root development in replant soil. CABI Agric. Biosci.1, 14. doi: 10.1186/s43170-020-00014-7

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 44

  Van HornC.SomeraT. S.MazzolaM. (2021). Comparative analysis of the rhizosphere and endophytic microbiomes across apple rootstock genotypes in replant orchard soils. Phytobiomes J.5, 231–243. doi: 10.1094/pbiomes-08-20-0058-r

  • CrossRef
  • Google Scholar

• 45

  WangL.MazzolaM. (2019). Field evaluation of reduced rate brassicaceae seed meal amendment and rootstock genotype on the microbiome and control of apple replant disease. Phytopathology109, 1378–1391. doi: 10.1094/PHYTO-02-19-0045-R

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 46

  WheelerR. E. (2016). Permutation Tests for Linear Models in R. Available online at: https://cran.r-project.org/web/packages/lmPerm/vignettes/lmPerm.pdf (Accessed December 20, 2025).

  • Google Scholar

• 47

  WhiteT. J.BrunsT. L.TaylorJ. W. (1990). “ Amplification and direct sequencing of fungal ribosomal RNA genes for phylogentics,” in A Guide to Molecular Methods and Applications. Eds. InnisM. A.GelfandD. H.SniskyJ. J.WhiteT. J. ( Academic Press, New York), 315–322.

  • Google Scholar

• 48

  WillettM.SmithT. J.PetersonA. B.HinmanH.StevensR. G.LeyT.et al. (1994). Growing profitable apple orchards in replant sites: An interdisciplinary team approach in Washington State. HortTechnology4, 175–181. doi: 10.21273/horttech.4.2.175

  • CrossRef
  • Google Scholar

• 49

  WinkelmannT.SmallaK.AmelungW.BaabG.Grunewaldt-StöckerG.KanfraX.et al. (2019). Apple replant disease: Causes and mitigation strategies. Curr. Issues Mol. Biol.30, 89–106. doi: 10.21775/cimb.030.089

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 50

  YangM.LiuY.XiaY.LiM.HuangC.HouF.et al. (2025a). A synthetic microbial community derived from healthy apple rhizosphere alleviates apple replant disease. Hortic. Res.12. doi: 10.1093/hr/uhaf217

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51

  YangS.YuanT.DuanT.ZhuH.ZhangX.ZhangH.et al. (2025b). Integrated omics analysis of PGPR and AMF effects on soil microbiota and root metabolites in Isatis indigotica. Front. Microbiomes4. doi: 10.3389/frmbi.2025.1709335

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52

  YaoS.MerwinI. A.AbawiG. S.ThiesJ. E. (2006). Soil fumigation and compost amendment alter soil microbial community composition but do not improve tree growth or yield in an apple replant site. Soil Biol. Biochem.38, 587–599. doi: 10.1016/j.soilbio.2005.06.026

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 53

  ZhangC.van der HeijdenM. G. A.DoddsB. K.NguyenT. B.SpoorenJ.Valzano-HeldA.et al. (2024). A tripartite bacterial-fungal-plant symbiosis in the mycorrhiza-shaped microbiome drives plant growth and mycorrhization. Microbiome12, 13. doi: 10.1186/s40168-023-01726-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 54

  ZhuY.-Q.LiX.-L.ZhaoD.-X.WeiY.-L.YuanH.-S. (2024). Four new species of Tomentella (Thelephorales, Basidiomycota) from subtropical forests in Southwestern China. J. Fungi10, 440. doi: 10.1111/j.1756-1051.2002.tb01912.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar