The study was conducted at two representative smallholder bell pepper farms located in the Tshwane (farm E), and Germiston (farm G) regions of Gauteng Province, South Africa. The province falls in the summer rainfall area and receives 600–700 mm of rain per annum. It has average annual maximum temperature ranges of 22 and 25°C with an altitude between 1350 and 1800 m above sea level (Dyson, 2015). Production in the selected farms were under shade nets. Asymptomatic bell pepper samples were collected during the September 2022 to April 2023 growing season for four different stages of fruit development. The samples included flowers (FL; n=3 composite samples, each comprising of 15 individual flowers), small fruits (SF; n=3 composite samples of five fruits each), mature fruits at harvest (MF1; n=3 composite samples of three fruits each), and fruits that completed 10 days of storage (ST; n=3 composite samples of three fruits each), resulting in a total of n=24 samples. All samples were randomly collected within the central rows under the shade nets ( Figure 1 ). The samples were transported in cooler boxes containing ice packs to the Plant Pathology laboratory at the University of Pretoria for processing. A selection of fruit was stored in the collection boxes at room temperature for 10 days to simulate postharvest storage. All samples were processed within 24 hours after sampling.

Physiological characteristics were measured at SF, MF1, and ST fruit stages, and included total soluble solids (TSS), pH, and defense-related phytochemicals; total phenolic content (TPC), total flavonoid content (TFC), and ascorbic acid (AA). The TSS content (% Brix) of the fruit was measured using a hand digital refractometer (PR-32, Atago, TSS 0–32%, Palette, Japan) following the method described by Nisansala et al. (2015). The fruit pH was measured as described by Abudi et al. (2020), using a Jenway 3510 pH meter (Bibby Scientific Ltd., Stone, UK) with a flat surface electrode combination (Extech Instruments, Waltham, MA, USA).

Plant material (fresh weight) from each fruit stage was prepared for the assessment of defense-related phytochemicals by cutting the samples into small pieces and snap-freezing them in liquid nitrogen. The frozen samples were then aseptically ground to powder form using a sterile electric coffee blender (Kambrook) packaged in sterile resealable bags and stored at −80°C until use. To test for TPC, the extraction was done as described by Boeing et al. (2014) with some minor modifications: sample powder (1 g) was suspended in 5 mL acetone water (Minema Chemicals, Johannesburg, South Africa) (1:1 v/v). This was followed by vigorous mixing using a vortex before incubation of the mixture for 6 hours at room temperature. The mixture was then centrifuged at 3000 × g for 10 mins, after which the supernatant/plant extract was transferred into clean tubes and used for fruit total phenolic estimation following the Folin-Ciocalteu method (Sellamuthu et al., 2013). In triplicates, the supernatant (9 µL) was mixed with 109 µL of 10% Folin-Ciocalteu reagent (Sigma-Aldrich, Darmstadt, Germany) and incubated at room temperature for 3 mins. This was followed by adding 180 µL of 7.5% sodium bicarbonate solution (NaCO₃) (Minema Chemicals), and the reaction mixture was incubated at 50 °C for 5 mins. The optical density was measured at 760 nm using an ultraviolet (UV) spectrophotometer (ThermoScientific, Multiskan Go, Waltham, MA USA). The absorbance values of the samples were then corrected using the corresponding blank readings prior to concentration calculations. Gallic acid (Sigma-Aldrich) was used as the standard, and results were presented as mg of gallic acid equivalent (GAE)/100 g of bell pepper fruit powder.

The total flavonoid content was measured by a colorimetric assay, as described by Zhishen et al. (1999). Plant extraction was done according to Alipieva et al. (2010) with a few modifications, where 1 g of the sample was suspended in 5 mL of 60% methanol (Sigma-Aldrich). The mixture was then incubated at 50°C in a water bath for 30 mins before centrifuging at 3000 × g for 10 mins. In triplicates, the resultant supernatant (0.15 mL) was added to 1 mL of distilled water, followed by the addition of 0.15 mL of 5% sodium nitrite (NaNO₂) (Minema Chemicals). After 5 min, 0.15 mL of 10% aluminium chloride (AlCl₃) (Sigma-Aldrich) was added to the solution and left to stand for 6 min at room temperature, before adding 0.5 mL of 1 M sodium hydroxide (NaOH) (Minema Chemicals) and vortexing. The absorbance was determined at 510 nm and the sample reading corrected using the respective blanks before calculating the concentrations. Catechin (Sigma-Aldrich) was used as the standard for the calibration curve and TFC was expressed as mg of catechin equivalent (CE)/100 g of bell pepper fruit powder.

Ascorbic acid was determined using the indophenol method described by Nielsen (2017) with minor modifications. Fruit powder (4 g) was mixed with 8 mL of metaphosphoric acid (HPO₃) (Sigma-Aldrich), vortexed and incubated at room temperature for 10 mins. The mixture was then centrifuged at 6000 × g for 10 mins, and the resultant supernatant was used for the ascorbic acid assay. In triplicates, 2 mL of the supernatant was mixed with 5 mL HPO₃ in volumetric flasks and titrated against the prepared 2,6-dichloroindophenol dye (Sigma-Aldrich) until a light, but distinct rose-pink color persisted for more than five seconds. The initial and final burette readings were recorded before and after each titration to calculate the amount of the dye used. The same procedure was repeated for standard ascorbic acid and the blank solutions. Ascorbic acid was calculated as described by Nielsen (2017) ( Supplementary Material 1 ).

The data collected was analyzed using R statistical software version 4.3.1 in RStudio (R Core Team, 2023). Means and standard errors for TPC, TFC, AA, TSS, and pH were calculated, and analysis of variance (ANOVA) at a 95% confidence interval was used to test for significant differences of the means. Post hoc analysis was performed using the least significant difference (LSD) test for pairwise comparison.

The collected flower and fruit samples were suspended in sterile double-distilled water containing 0.1% Tween 80 (Sigma-Aldrich) to remove microfloral epiphytes. This was followed by sonication in an ultrasonic water bath (Labotech, Johannesburg) for five minutes, as described by Gomba et al. (2016). These micro-floral washes were filtered through a cellulose nitrate filter paper (0.45 µm pore size; Sartorius, Gottingen, Germany), and the filters containing wash residue (microflora epiphytes) aseptically stored at −20 °C before DNA extraction for the total epiphyte metagenomic DNA.

Total metagenomic DNA extractions of samples at the different pepper developmental stages were done using the Quick-DNA™ Fungal/Bacterial Miniprep kit (Zymo-Research, California, USA), following the manufacturer’s instructions. The DNA concentrations were quantified using a Nanodrop ND2000 Spectrophotometer (Thermo Fischer Scientific, Dreieich, Germany). DNA were stored as aliquots at −20°C, with a separate aliquot prepared for submission to Molecular Research LP (MR DNA; Texas, USA) for paired-end (2×300 bp) high-throughput Illumina MiSeq (Illumina, Inc. USA) sequencing, using the bTEFAP^® method (Dowd et al., 2008). The internal transcribed spacer (ITS) region was targeted, using primers ITS1F (5’-CTTGGTCATTTAGAGGAAGTAA-3’) and ITS2R (5’-GCTGCGTTCTTCATCGATGC-3’), as described by Tedersoo et al. (2015).

Raw sequence data received from MR DNA were processed using the Quantitative Insights Into Microbial Ecology 2 (QIIME 2) platform (Bolyen et al., 2019) on VirtualBox version 6.1.32 (Oracle Corporation, 2022). This involved demultiplexing, quality filtering by adapter trimming, denoising, paired read merging and chimera removal using the DADA2 analysis pipeline (Callahan et al., 2016). Feature tables were then generated from the cleaned amplicon sequence variants (ASVs) and taxonomic classification was done using a self-trained classifier database developed from the UNITE QIIME release version 9.0 (Abarenkov et al., 2023) at a 99% sequence similarity threshold.

Comprehensive fungal community and statistical analysis, including visualization of the generated datasets, were performed in R statistical software version 4.3.1 in RStudio (R Core Team, 2023). Fungal community structure and alpha diversity were calculated from non-rarefied sequences, while sequences rarefied to a minimum library size of 38–231 reads were used to determine beta diversity as described by Cameron et al. (2021). The alpha diversity indices measured at genus level included the Chao1 richness index, Shannon, and Gini-Simpson diversity indices. These were visualized using boxplots generated using the ggplot2 function in R.

The Shapiro–Wilk normality test was used to determine the distribution of the dataset for downstream statistical testing based on fruit developmental stage and sample origin (González-Estrada et al., 2022). Where normal distribution was observed, an analysis of variance (ANOVA) was carried out, followed by a post hoc Tukey’s Honestly Significant Difference (HSD) test for pairwise comparison (Sawyer, 2009). Alternatively, the Kruskal–Wallis test was used for samples not normally distributed, followed by a post hoc Dunn’s test (MacFarland and Yates, 2016). Beta diversity was determined using the Bray-Curtis dissimilarity matrix and visualized through principal coordinates analysis (PCoA) ordination (Xia and Sun, 2023). Statistical differences in beta diversity between fruit developmental stages and origin were inferred with permutational multivariate analysis of variance (PERMANOVA, 999 permutations) using the adonis function (Xia and Sun, 2023).

To assess ecological importance, common fungal taxa were determined at a minimum detection threshold of 1% relative abundance at a minimum sample prevalence of 0.9 (de Souza et al., 2016). Interaction of common mycobial organisms occurring in at least 90% of the total samples was determined using Spearman rank correlation by carrying out a sparse correlation for compositional data (SPARCC) using the phyloseq package in R (Kurtz et al., 2015; Neu et al., 2021). The fungal co-occurrence network generated was visualized using the igraph package (Csardi and Nepusz, 2006). To assess the postharvest pathogen profile, ITS amplicon sequence data of taxa with putative pathogenic nature based on literature were extracted from the processed files and their identity confirmed using NCBI BLASTn (Altschul et al., 1990). The abundance trends of potential pathogen and biological control taxa across the fruit developmental stages were evaluated using ggplot package in R. A redundancy analysis (RDA) was carried out on Hellinger transformed data to determine the influence of fruit physiological parameters on common fungal composition across the fruit developmental stages (Legendre and Gallagher, 2001). The RDA plot was generated using the ggplot.

Fruit physiological parameters (TPC, TFC, AA and TSS) increased with fruit maturity (SF-ST) in farm E ( Table 1 ), while fruit pH, decreased with maturity. The changes in physiological parameters were significantly different between the small fruits and mature fruits at harvest and storage (p < 0.05, Table 1 ). Farm G’s fruits also depicted a general increase in TPC and TFC concentrations between the small fruits and mature fruits at storage, despite a notable drop at MF1 stage ( Table 1 ). Ascorbic acid concentration and % brix (TSS) significantly increased with fruit maturity, while fruit pH declined (p < 0.05, Table 1 ).

A total of 1,552,622 fungal reads (max. number of reads =101,632, min. number of reads = 38,231, and avg. number of reads = 64,692.58) were retrieved from the 24 bell pepper samples. Data processing yielded a total of 1,778 ASVs comprising of five phyla, 24 classes, 64 orders, 158 families and 346 genera. Alpha diversity measures at genus level ASVs across all the developmental stages ( Figure 2a ; Table 2 ) for farm E showed that small fruits had a significantly lower mean Chao1 ASV richness compared to the flowering stage (ANOVA, p = 0.00153). The ASV richness, however, increased with fruit maturity (SF-MF1). Similarly, for farm G, the ASV richness dropped from the FL to SF stage and again peaked with fruit maturation. Despite the variance in ASV richness for farm G, the values were not significantly different across the fruit developmental stages. In both farms, there was a drop in ASV richness at storage ( Figure 2a ; Table 2 ).

Farm E had a similar trend in Shannon and Simpson diversity as the ASV richness, where the floral stage had a significantly higher diversity compared to the other fruit developmental stages (ANOVA, p = 0.0023, and p = 0.0307 respectively). According to the two indices, the diversity decreased at the small fruit stage, increased with fruit maturation, and decreased at fruit storage ( Figure 2a ; Table 2 ). In contrast, farm G had a linear increase in Shannon diversity from floral stage to fruits at storage. Diversity according to the Simpson index increased from FL to SF with a slight decrease towards maturity and subsequent increase in storage. Despite the observed variations, there was no statistical difference in diversity across the developmental stages in farm G (ANOVA, p > 0.05, and p >0.05) for Shannon and Simpson diversity indices respectively ( Figure 2a ; Table 2 ).

Beta diversity analysis using PCoA ordination showed clear clustering of the fungal community based on fruit development stage ( Figure 2b ). Floral and small fruit fungal communities from both farms clustered together except for EFL, which was separate and sparsely distributed from the rest. Fungal communities from mature fruit stages also clustered together, except for those from EMF1, which formed a slightly separate group ( Figure 2b ). A total of 47% of the variation in the fungal community was accounted for within the first two PC’s in the dataset, with a strong and significant developmental stage influence (PERMANOVA: R² = 0.3760, F=4.0164, P=0.001) on the fungal microbiota between early and late fruit maturity (28%).

Further analysis at genus level showed that 57 (19.1%) of the total taxa in farm E were shared across the four fruit stages. A total of 91 (30.5%), 9 (3.0%), 28 (9.4%), and 23 (7.7%) of the taxa identified were exclusive to EFL, ESF, EMF, and EST stages, respectively. Farm G samples indicated that 54 (25.7%) of the total taxa were shared across the four fruit stages. The GFL, GSF, GMF1 and GST samples exclusively accounted for 16 (7.6%), 7 (3.3%), 51 (24.3%), and 19 (9.0%) of the taxa respectively.

Ascomycota and Basidiomycota were the most abundant phyla across all samples, accounting for 83.9% and 15.3% of the total fungal community, respectively. The remaining 0.8% of the total fungal community population was accounted for by Chytridiomycota (0.08%), Mortierellomycota (0.02%), Incertae sedis/uncultured fungi (0.4%), and unclassified fungi (0.3%). Cladosporium (37%) and Alternaria (16%) were ranked as the most abundant genera across all four different fruit developmental stages. These were distantly followed by Fusarium, accounting for 5% of the total relative abundance ( Figure 3 ). Other common and abundant genera included Sampaiozyma, Psedopithomyces, Naganishia, Filobasidium, Plectospharella, Aureobasidium, Hohenbuehella, Sporobolomyces, Exserohilum, Acremonium, Coprinellius, Rhodotorula and Xenodidymella.

Among the common genera across all the fruit developmental stages were Alternaria, Cladosporium and Fusarium, previously reported as pathogens of bell pepper ( Supplementary Material 2 ). The abundance and distribution of Alternaria were almost similar in both farms while Fusarium was more abundant in farm G ( Figure 3 ). Colletotrichum was also present at all fruit developmental stages in farm G and only present on mature fruits (MF1 and ST) in farm E. According to the NCBI BLASTn taxonomy database, the species-level identity for Alternaria ASVs were mainly A. alternata, A. arborescens, A. cinerarie, A. japonica, A. solani, and A. sonchi. Fusariam comprised of F. oxysporum, F. graminearum, F. equiseti, F. domesticum, F. dimerum, F. delphinoides, F. citri, F. solani, F. tricinctum, F. proliferatum, F. penzigii, and F. chlamydosporum. Colletotrichum ASVs included C. gloeosporioides, C. truncatum and C. cirnanams.

Upon analysis of the trends at genus level across the fruit developmental stages, Cladosporium, Filobasidium, Acremonium, and Plectosphaerella had increased abundance from flowering to storage. Alternaria, Sampaiozyma, Fusarium, Naganishia, Aerobasidium, Hohenbuehelia, Rhodotorula, and Xenodidymella had fluctuating abundance over the fruit developmental stages, which differed by farm. Alternaria abundance in farm E, for instance, was highest at the small fruit stage, reduced with fruit maturity, and increased at storage. Alternaria abundance in farm G was highest at the floral stage and reduced as the fruit matured. Fusarium in both farms had the highest abundance at the small fruit stage, reduced towards harvest and increased at storage. In both farms, the abundance of Sampaiozyma increased as Alternaria decreased ( Table 3 ) with a Pearson correlation analysis showing significant negative correlation between the two taxa across the fruit developmental stages (r = −0.63, p< 0.05, Supplementary Material 3 ). The abundance of Hohenbuehelia and Exserohilum were inversely proportional to that of Sampaiozyma across the fruit developmental stages (r = −0.6, p <0.05). Sporobolomyces and Exserohilum decreased in abundance from flowering to storage ( Table 3 ). Sporobolomyces in farm E peaked at the small fruit stage, while flowers recorded the highest abundance in farm G. Exserohilum in both farms had reduced abundance from flowering to stored stages.

Co-occurrence network analysis based on statistical correlation showed the positive (green arrows) and negative (red arrows) correlations between the fungal ASVs in the community ( Figure 4 ). Strong positive correlations were observed between Alternaria and the genera; Exserohilum, Hohenbuehelia, Pseudopithomyces, Sporobolomyces, Xenodidymella and Aureobasidium. On the other hand, Alternaria had a strong negative correlation with Sampaiozyma and moderate negative correlation with Coprinellus and Filobasidium ( Figure 4 ). Fusarium was depicted to have a strong positive correlation with Acremonium, Plectosphaerella and a moderate positive correlation with Sporobolomyces. It had a strong negative correlation with Aureobasidium, Coprinellus and Filobasidium ( Figure 4 ).

Acremonium had a strong negative correlation with Aureobasidium, Corprinellus. Plectosphaerella had a strong positive correlation with Acremonium and a strong negative correlation with Aureobasidium. The network also depicted that Cladosporium positively correlated with almost all taxa around it with Pseudopithomyces interaction being the only negative, albeit weak, correlations. Naganishia also had a strong positive correlation with all the surrounding taxa; Aureobasidium, Coprinellus, Filobasidium and Sampaiozyma.

Aureobasidium, Coprinellus, Filobasidium, and Sampaiozyma were also included in this positive correlation hub. However, they additionally showed negative correlations with selected ASVs. For instance, Coprinellus negatively correlated towards Pseudopithomyces and Sporobolomyces. It also had a positive correlation towards Aureobasidium, Cladosporium, Filobasidium, Naganishia and Sampaiozyma. Sampaiozyma on the other hand strongly correlated negatively towards Alternaria, Exserohilum, Hohenbuehelia, Pseudopithomyces and Sporobolomyces but had strong positive correlations with Cladosporium, Filobasidium and Naganishia ( Figure 4 ). Filobasidium was strongly negatively correlated towards Alternaria, Exserohilum, Fusarium, Hohenbuehelia, Pseudopithomyces, Sporobolomyces and Xenodidymella, and positively correlated with Cladosporium and Naganishia. Despite Auerobasidium having a strong negative correlation towards Acremonium, Fusarium and Plectosphaerella, it interestingly had a strong positive correlation with Alternaria, Cladosporium, Exserohilum, Hohenbuehelia and Xenodidymella.

The redundancy analysis showed that fruit physiological attributes (AA, TPC, TFC and TSS) had a significant influence on fungal composition, contributing to 8.53% of the total variance (ANOVA, Variance = 0.085335, F = 4.2068, p = 0.001, Table 4 ). Total flavonoid content (TFC) had the highest influence on fungal composition (ANOVA, Variance = 0.021337, F= 5.2594, p = 0.0002) while TPC had the lowest effect (ANOVA, Variance = 0.009587, F= 2.3631, p = 0.066, Table 4 ).

The ordination ( Figure 5 ) indicated that 58.61% of the observed variance was captured in RDA1 and RDA2. The taxa Acremonium and Plestospharella clustered closer to TFC, TPC and TSS. Fusarium was shown to strongly correlate with the changes in fruit pH. Cladosporium and Filobasidium depicted a strong positive correlation to the increase in ascorbic acid and was also negatively correlated to TFC, and TSS increase. Alternaria, Aureobasidium, Exserohilum, Naganishia, Sporobolomyces, Rhodoturula and Xenodidymella had moderate correlation with all the tested physiological attributes (AA, TFC, TPC, TSS and pH), as shown by their central positioning ( Figure 5 ).

The main objective of the present study was to characterize the bell pepper mycobiome across different fruit developmental and storage stages. This understanding can help identify possible pathogen entry points guiding timely intervention to reduce postharvest decay and spoilage. The study showed Ascomycota and Basidiomycota as the most abundant phyla across all bell pepper fruit developmental stages, accounting for a cumulative 99% of the total surface fungi. This agrees with the study findings by Khwantongyim et al. (2021), which showed that these two phyla were the most abundant on apple fruits, accounting for 77.6% and 22.1% of the total counts, respectively. The dynamics of Ascomycota are specifically important as this dominant phylum includes the pre- and postharvest pathogenic genera, namely, Alternaria, Fusarium and Colletotrichum, as identified in the current study. The dominance of Ascomycota could also be associated with their intrinsically higher genomic potential for resource utilization, competition, and tolerance to stress when compared to saprobic Basidiomycota (Egidi et al., 2019). The consistent presence of Basidiomycota such as Sampaiozyma, Filobasidium, Naganishia, Hohenbuehelia suggests a key, yet lesser-understood role of this phylum in fungal dynamics, potentially influencing rot development during storage.

Genus-level analysis revealed further insights into the dynamic shifts in fungal communities across developmental stages. For instance, less than 26% of the observed genera on each farm were shared across all fruit developmental stages, indicating that most of the observed fungal genera were unique to specific stages of development. During the early stages, the bell pepper community was dominated by fungal genera such as Cladosporium, Alternaria, and Fusarium, which are typically associated with leaf blight diseases on several hosts (Ghoneem et al., 2025; Manathunga et al., 2024). At maturity and during storage, Cladosporium became more prominent. The increase in Cladosporium counts on mature fruits supports the finding by Swett et al. (2019) that fruit susceptibility increases with maturity. In their study, Cladosporium infection of raspberry increased from 2.5% at preharvest to over 50% after harvest, compared to the present study where the relative abundance increased from 26.7% at preharvest to 47.7% after harvest. At postharvest, Alternaria and Fusarium also slightly increased, while populations of some taxa, for instance Aureobasidium, Naganishia and Rhodoturula, declined. This reinforces a potential link between microbial competition, the role of these fungi in temporal community dynamics and their contribution to the onset of decay during storage (Verma et al., 2022; Fink and Manhart, 2023). This can also be associated with the observed decline in community diversity at storage as the pathogenic taxa possibly outcompeted other fungal genera in the community (Zhang et al., 2021). Zhang et al. (2021) reported that fungal communities are more sensitive to rot, resulting in a significant decline in abundance during fruit decay. Fungal community structure in terms of genus types was similar in the two farms but only varied in abundance across the developmental stages. For instance, Fusarium were slightly more in farm G compared to farm E and this could be attributed to farmer agronomic practices ranging from seed sources, planting substrate and routine management practices.

PCoA ordinations support the observed alpha diversity trends by clearly distinguishing fungal communities between early and later bell pepper development stages. The distinction could be a result of the physiological changes in fruit over time, as indicated by the tested physiological parameters and the RDA analysis in the current study. Previous studies have shown that physiological changes during fruit development result in biochemical shifts, which in turn influence microbial community assembly or colonization patterns and may further affect the entire plant holobiont (Lone et al., 2024; Zhimo et al., 2022). According to Baglioni et al. (2024) and Pang et al. (2021), plant secondary metabolites shape the endophyte and epiphyte microbial composition by acting as defense compounds against some microbes and as signaling molecules or source of energy for other microbes.

The RDA analysis showed that plant microbes associated differently with the changes in fruit physiological attributes. These differences in fungal association with specific fruit physiological stages and metabolic compounds suggest that fungal taxa exhibit selective physiological preferences, which can in turn strongly impact their colonization and fruit quality during bell pepper development and postharvest (Lievens et al., 2015; Palmieri et al., 2023). However, RDA explains only 8.53% of fungal variation, with the remaining 91.47% possibly being accounted for by other variables not tested in this study. This includes abiotic factors, such as environmental conditions (Verma et al., 2022), and handling practices (Abdelfattah et al., 2016; Bill et al., 2021).

Among the taxa consistently present across all bell pepper developmental stages were Alternaria, and Fusarium. These fungal phylotypes have a wide distribution in agricultural environments and exhibit variable abundance (Egidi et al., 2019; Yadav et al., 2022), providing a potential explanation for their high prevalence across the fruit developmental stages. The phylotypes have also been previously reported as pre- and postharvest pathogens of bell peppers (Tiamiyu et al., 2023). For instance, Alternaria species, particularly A. alternata, A. solani and A. tenuissima, are considered among the most aggressive fungal species affecting several horticultural crops including bell peppers globally (Balamurugan and Kumar, 2023). The significant crop losses associated with Alternaria is largely due to its high prevalence, which can be attributed to its ubiquitous nature and diverse composition, as well as its broad host range (Dang et al., 2015).

Fusarium species have also been linked with bell pepper internal fruit decay (Tiamiyu et al., 2023). The presence of Fusarium across all fruit developmental stages and at storage is consistent with previous studies on bell pepper crops at pre- and postharvest stages (Engalycheva et al., 2024; Liu et al., 2016). This can be attributed to its versatile nature with the ability to survive in almost every environment, including air, water, soil, plants, and other organic substrates (Ekwomadu and Mwanza, 2023). It also has efficient dispersal mechanisms hence its high abundance and distribution in nature (Tupaki-Sreepurna and Kindo, 2018). Some of the Fusarium species that have been reported on bell peppers include F. oxysporum, F. solani, F. verticillioides, F. commune, F. torulosum, and F. sporotrichioides (Engalycheva et al., 2024; Gilardi et al., 2019). Colletotrichum was also among the taxa identified from the retrieved ASVs across all fruit developmental stages in one of the farms, highlighting its potential to infect bell peppers both at pre and postharvest stages (Ali et al., 2016). Pathogenic Colletotrichum spp. that have been reported on Capsicum annuum globally include, C. gloeosporoides, C. truncatum, C. fruticola, C. scovillei, C. endophyticum, C. karsti, C. plurivorum, C. siamense and C. tropicale (de Silva et al., 2019; Liu et al., 2016).

The identification of pathogenic taxa Alternaria, and Fusarium from floral stage through to fruit at storage provides insights potentially linking deterioration during storage with quiescent infections at the early fruit developmental stage. This highlights the need to strengthen preharvest management, for instance timely spray cycles ideally before or at anthesis to reduce postharvest losses and maintain fruit quality. Delayed spray cycles beyond anthesis stage may result in reduced fungicide efficacy particularly against pathogens causing internal fruit rot (Paul et al., 2018). It is also essential to use synthetic fungicides judiciously, as their improper application can suppress potential biocontrol agents (BCAs), possibly enabling pathogenic taxa to proliferate (Szymanski et al., 2023), ultimately impacting fruit quality and postharvest storage outcomes. Also, among prevalent microbial taxa were Pseudopithomyces, Plectosphaerella, Bipolaris, Exserohilum, Lectera, Xenodidymella, Curvularia, and Nigrospora. Despite their existence across the fruit developmental stages, their role in the bell pepper fungal community is yet to be fully understood. These taxa, however, have been reported as pathogenic in crops such as barley, maize, rice and wheat but not in bell peppers (McDonald et al., 2018; Perelló et al., 2017; Wang et al., 2022).

Bell pepper fungal community across the different fruit developmental stages also comprised of potential BCAs. Among them were Acremonium, Aureobasidium, Filobasidium, Hohenbuehelia, Rhodotorula, Sampaiozyma, Sporobolomyces, and Wickerhamomyces. Some of these taxa exhibited an inverse relationship with pathogenic taxa. Specifically, Sampaiozyma and Filobasidium were more abundant in samples where Alternaria and Fusarium were less prevalent, whereas their abundance decreased in samples where Alternaria and Fusarium were more dominant. A similar trend was seen in the cases of Rhodoturula and Xenodidymella, highlighting their potential inhibitory effect against pathogenic taxa. Some of these taxa, for instance Aureobasidium pullulans, have been explored for biocontrol activity against B. cinerea, R. solani, Penicillium expansum and P. digitatum (Agirman and Erten, 2020; Di Francesco et al., 2021; Galli et al., 2021). These studies associated inhibition by A. pullulans to several action mechanisms including competition for space and nutrients, production of volatile organic compounds (VOCs) and secretion of extracellular lytic enzymes. Filobasidium oeirense and F. wieringae have also been reported to inhibit growth of B. cinerea and A. porri mainly by production of antifungal VOCs (Abo-Elyousr et al., 2024; Ruiz-Moyano et al., 2020). Production of VOCs by Rhodoturula species and Wickerhamomyces anomalus have also been linked to inhibition of Aspergillus flavus spore germination, mycelial growth, and toxin production (Hua et al., 2014). Apart from antibiosis, resource competition and mycoparasitism, some BCAs induce plant systemic resistance against pathogenic species (Chaudhary et al., 2024; Prasad et al., 2023). The observed inverse relationship between the potential BCAs and pathogenic taxa in this study could be due to one of these mechanisms, particularly antibiosis or resource competition. Future studies should focus on their functional level of interaction.

Dynamic changes in fungal interaction on bell peppers during development and storage were noted in the current study through co-occurrence network analysis. Potential synergistic and antagonistic interactions within the bell pepper fungal community were revealed, further explaining the temporal changes within the mycobiome genera. This supports the observations of Wright et al. (2019), that microbial communities are characterized by rapid, dynamic succession that maintains ecological stability. The negative correlation of the genera Aureobasidium, Sampaiozyma, Filobasidium and Coprinellus to Alternaria and Fusarium may depict their inhibitory effect as majority of them have been reported as potential BCAs (Abo-Elyousr et al., 2024; Di Francesco et al., 2021; Elkhairy et al., 2023). Cases of positive correlation between potential pathogenic taxa e.g., Alternaria, Pseudopithomyces, Exserohilum, Xenodidymella and Hohenbuehelia, as well as that of Fusarium, Acremoniaum, Plectosphaerella and Sporobolomyces was also noted. This kind of interaction provides a possible glimpse into dual or multi-pathogen infection. This type of host interaction is less studied despite its potential to alter the course of disease and severity expression (Abdullah et al., 2017). For instance, co-infection by F. oxysporum f. sp. medicaginis and R. solani exacerbated disease severity in commercial alfalfa fields in China (Fang et al., 2021). A study by Lerch-Olson and Robertson (2020) also showed that co-inoculation with Pythium and Fusarium species greatly enhanced disease severity in soybeans compared to single pathogen inoculation.

Positive correlation was also depicted among the potential BCAs, Aureobasidium, Coprinellus, Sampaiozyma and Filobasidium, which may be interpreted as mutualistic or complementary interaction within their microbial community and can impact their inhibitory effect against pathogenic taxa. This supports the findings by Ramudingana et al. (2024), which shows that some endophytic fungi, such as Preussia africana and Coprinellus micaceus, have mutualistic associations and can be explored for combined control against fungal pathogens. Other genera such as Cladosporium positively correlated with all other fungi in the bell pepper mycobiome, affirming its ecological adaptability contributing to high abundance throughout the fruit developmental stages (Salvatore et al., 2021). Although considered a weak pathogen, its high adaptability and prevalence enable it to influence the surface microbial communities of fresh produce, potentially causing spoilage under favorable conditions. Therefore, effective postharvest management practices are essential to mitigate its impact (Bento de Carvalho et al., 2024). Also worth noting in the current study, was the negative correlation of Aureobasidium to Fusarium and a positive correlation to Alternaria. This highlights the intricate nature of microbial interactions in fruit maturation, where a single organism inhibits one pathogen while simultaneously promoting the growth of another (Knudsen and Dandurand, 2014). This complexity highlights the absence of a “silver bullet” biocontrol agent, as the efficacy of biocontrol agents can vary based on environmental conditions, target pathogens and the overall dynamic interactions on the host plant (Knudsen and Spurr, 2023).