All spinach, rocket, and kale produce (Caryophyllales for spinach, Brassicales for rocket, and kale, referred here as species) used in this study were cultivated as described by Culliney and Schmalenberger (2022). A total of 160 samples from L. monocytogenes growth potential experiments were selected: open field and polytunnel spinach (F1 Trumpet; summer harvest), open field and polytunnel rocket (Buzz), polytunnel spinach (F1 Cello), polytunnel rocket (Esmee), open field spinach (F1 Trumpet; winter harvest). Samples were stored for days 0, 2, and 5 at 7°C and for days 7–9 at 12°C for days, where L. monocytogenes and total bacteria counts (TBCs) were enumerated on cultivation media (Culliney and Schmalenberger, 2022).

Growth experiments were executed as described in accordance with the European Union (EU) guidance document’s guidelines for conducting growth potential studies (European Union Reference Laboratory for Listeria monocytogenes (EURL Lm); EURL Lm, 2019). The rationale behind selecting these guidelines is to provide a robust representation of real-life scenarios involving low-level contaminations with the potential to grow under realistic storage conditions. Each sample consisted of 25 g of produce inoculated with 100 cfu g⁻¹ of a three-strain mix of L. monocytogenes, that is, 959 (vegetable isolate), 1,382 (EURL Lm reference strain), and 6,179 (food processing plant isolate). The contents of each were transferred into separate stomacher bags and homogenized in 25 mL of phosphate-buffered saline (PBS) using a stomacher (Seward 400, AGB Scientific, Dublin, Ireland) for 120 s at a high speed (260 rpm). These homogenates were used for all types of microbial analysis.

Growth potentials (log₁₀ cfu g⁻¹) calculated from median values were open field spinach (F1 Trumpet; summer harvest) = 2.59, polytunnel spinach (F1 Trumpet) = 1.40, open field rocket (Buzz) = 1.28, polytunnel rocket (Buzz) = 1.45, polytunnel rocket (Esmee) = 1.23, polytunnel spinach (F1 Cello) = 1.84, polytunnel kale (Nero di Toscana) = 2.56, and open field spinach (F1 Trumpet; winter harvest) = 1.65 as described recently (Culliney and Schmalenberger, 2022).

The associated average L. monocytogenes counts (log₁₀ cfu g⁻¹) across the five time points (± the relative increase or decrease from the previous time point) are displayed in Table 1.

The remaining homogenate suspensions obtained after microbial analysis were transferred into 50 mL conical tubes and centrifuged at 4,500g (15 min at 4°C). Supernatants were discarded, and the derived pellets were stored at −20°C. For DNA extraction, pellets were resuspended in 400 μL of PBS, and 100 μL was used for DNA extraction with the PowerFood DNA Isolation kit (MO BIO Laboratories, Carlsbad, CA, USA) according to the manufacturer’s instructions. The quantity and quality of the extracted DNA were determined with the Take3 plate in an Eon plate reader/incubator (BioTek, Winooski, VT, USA) (Culliney and Schmalenberger, 2024).

All 160 samples were sent to the University of Minnesota Genomics Center (UMGC) for indexing and Illumina MiSeq (San Diego, CA) sequencing. Raw sequencing files were deposited in the Sequence Read Archive (National Center for Biotechnology Information [NCBI]) under the BioProject identification [ID] number: PRJNA117723.4. Bioinformatics analysis was performed using QIIME2 2021.11 (https://qiime2.org/) (Bolyen et al., 2019) as described recently by Culliney and Schmalenberger (2024). The paired-end sequences with quality of each group of 20 samples were demultiplexed and imported with metadata separately via a ManifestPhred33V2 file. This was followed by trimming and truncating (quality filtering at Q20) using the q2-dada2 plugin. Following this, the “qiime feature-table merge” and “qiime feature-table merge-seqs” plug-ins to merge feature tables and the representative amplicon sequence variants (ASV) were conducted, so the following group comparisons could be performed:

Comparison 1 (open field vs. polytunnel vs. plant species) consisted of open field spinach (F1 Trumpet), polytunnel spinach (F1 Trumpet), open field rocket (Buzz), and polytunnel rocket (Buzz).

Comparison 2 (variety vs. species) consisted of polytunnel spinach (F1 Trumpet), polytunnel rocket (Buzz), polytunnel rocket (Esmee), polytunnel spinach (F1 Cello), and polytunnel kale (Nero di Toscana).

Comparison 3 (seasonality) included open field spinach (F1 Trumpet; summer harvest) and open field spinach (F1 Trumpet; winter harvest).

Assigning taxonomic information to the ASV sequences was conducted using a pre-trained Naïve Bayes taxonomic classifier, which was trained on the Silva version 138.99% reference dataset where sequences were trimmed to represent only the region between the 515F/806R primers (V3–V4 region) as described previously (Culliney and Schmalenberger, 2025). Sequences not assigned to a phylum level, chloroplast, and mitochondrial sequences were removed using the filter-table method in the q2-taxa plugin. All subsequent analyses were conducted with both rarefied and unrarefied data. Even sampling depths for use in diversity metrics were for comparison 1: 11,519 → Retained 921,520 (29.48%) features in 80 (100.00%) samples at the specified sampling depth; for comparison 2: 3,117 → Retained 240,009 (9.99%) features in 77 (79.38%) samples at the specified sampling depth; and for comparison 3: 15,015 → Retained 600,600 (40.99%) features in 40 (100.00%) samples at the specified sampling depth. Alpha diversity metrics (observed ASVs, Shannon index, Pielou’s evenness, and Faith’s Phylogenetic Diversity) and beta diversity metrics (weighted unique fraction metric or UniFrac (Lozupone et al., 2007) and Bray–Curtis dissimilarity) using q2-diversity were estimated and viewed on Principal Coordinates Analysis (PCoA) Emperor plots. Analysis of Composition of Microbiomes (ANCOM) test in the q2-composition plugin was used to identify differentially abundant features. ANCOM identified individual taxa whose relative abundances are significantly different across groups. Relative abundance was calculated after conversion of the biome tables from QIIME2 to tsv files (phylum and family levels). Pearson’s correlation coefficient was determined to measure the strength and direction of the linear association between two variables (i.e., between L. monocytogenes populations and the corresponding relative abundance of each of the 20 most abundant families) for all groups over time (Sedgwick, 2012). Pearson’s correlation coefficient from <0.10 is a negligible correlation, 0.10–0.39 indicates weak correlations, 0.40–0.69 represents moderate correlations, while 0.70–0.89 indicates strong correlations, with >0.90 being very strong (Schober et al., 2018). Absolute abundances of bacterial taxa at the family and genus level were estimated by using the total heterotrophic counts published elsewhere (Culliney and Schmalenberger, 2022). Input, filtered, denoised, merged, non-chimeric reads, as well as chloroplast to total DNA content, are reported in Supplementary Tables S1–S3.

RStudio software (Posit, Boston, MA; version 4.1.1) was used for statistical analysis. In situations of normality (Shapiro–Wilk test) and homoscedasticity (Levene’s), a one-way analysis of variance (ANOVA) was conducted to compare input, filtered, denoised, merged, and non-chimeric reads between groups. The remainder of the statistical analysis for alpha and beta diversity metrics was conducted in QIIME2. For alpha diversity (observed ASVs, Shannon index, Pielou’s evenness, and Faith’s Phylogenetic Diversity (Faith, 1992)), comparisons among groups and pairwise comparisons were conducted through Kruskal–Wallis tests. Beta diversity was analyzed through the non-parametric permutation test, permutational multivariate analysis of variance (PERMANOVA) (999 permutations) (Anderson, 2017). Statistical significance was tested at p ≤ 0.05. In situations of normality (Shapiro–Wilk test) and homoscedasticity (Levene’s), a one-way ANOVA Tukey honestly significant difference (HSD) post hoc test applying Benjamini–Hochberg correction for multiple testing was conducted to compare relative abundances for all alpha diversity metrics and relative abundances across subgroups. In situations of non-normality, the Kruskal–Wallis rank sum test, using the kruskal.test function, and Dunn test post hoc analysis for multiple pairwise comparisons between groups were conducted, applying the Benjamini–Hochberg correction for multiple testing (false discovery rate was set at 10%). In situations of unequal variance, the oneway.test function was employed with var = F, and Games–Howell post hoc analysis.

This section describes how alpha and beta diversities were shaped by the selection of different leafy vegetable plants (spinach and rocket) as well as other cultivation methods (open field and polytunnel) and how diversities evolved with storage. The primary assumption was that both plant species and environment affect alpha and beta diversities. In turn, differently developing phyllosphere communities were expected to affect L. monocytogenes growth over time, as well as the succession of the phyllosphere community over time.

This section describes how their plant hosts shaped alpha and beta diversities at different taxonomic levels, that is, order (Caryophyllales for spinach, Brassicales for rocket and kale, referred here as species) vs. variety (Trumpet and Cello for spinach, Buzz and Esmee for rocket) and how diversity evolved during storage. The primary assumption was that both plant species and variety affect alpha and beta diversities. In turn, differently developing phyllosphere communities were expected to affect L. monocytogenes growth over time, as well as the succession of the phyllosphere community over time.

This section describes how alpha and beta diversities were shaped by seasonality (winter and summer) in spinach Trumpet and how diversities evolved during storage. The primary assumption was that winter vs. summer production affects alpha and beta diversities. In turn, differently developing phyllosphere communities were expected to affect L. monocytogenes growth over time, as well as the succession of the phyllosphere community over time.

Previous research assessing the effect of nitrogen fertilizer and leaf mineral content revealed that plant species alone, like spinach and rocket, influence the development of the phyllosphere (Darlison et al., 2019). However, the current study further revealed that the vegetable cultivation method had the strongest influence on the bacterial phyllosphere community structure. At the same time, plant species had a more pronounced effect on the overall abundance of phyllosphere bacteria. Here, polytunnel and open field cultivation of rocket and spinach displayed more similar phyllosphere bacterial communities compared to plant species alone. Additionally, the phyllosphere bacterial communities of various rocket and spinach cultivars were found to be significantly different in the present study. Previous research has identified the presence of microbe-plant variety interactions in field-grown lettuce. Dominated by Pseudomonadaceae and Enterobacteriaceae families, a clone library of three lettuce cultivars revealed significant differences between the relative abundances of genera belonging to the Enterobacteriaceae family, including Erwinia and Enterobacter (Hunter et al., 2010). While another study of the microbial diversity and structure of the phyllosphere of Alfalfa (Medicago sativa L.) identified significant effects of the season and the site where the plants were cultivated in open fields, no significant differences were detected between the two tested varieties (Zhang et al., 2022). While lettuce and alfalfa are bred for cultivation, and both plants are cultivated through a broad range of varieties, only lettuce is bred with the aim of human consumption of the leaves, as is the case for spinach and rocket. One may speculate that the breeding focus of lettuce, spinach, and rocket is primarily on the consumer experience of eating the leaves; thus, different varieties may differ more substantially in their leaf structure than this is the case for other plant varieties that are bred for livestock feeding.

A novel aspect of the present study was the identification of the presence or absence of bacteria, and their shifts in relative abundance, which may be of potential importance to the L. monocytogenes growth. For example, Pseudomonadaceae, which are of high abundance and are associated with the hydrolysis of proteins into amino acids, can induce the stimulation of L. monocytogenes growth (Marshall et al., 1992; Zilelidou and Skandamis, 2018). Contrariwise, Lactobacillales that were present in low abundance are commonly associated with decreased L. monocytogenes survival due to their competitive growth abilities (Østergaard et al., 2014). Indeed, the L. monocytogenes growth-enhancing Pseudomonas species has previously been associated with spinach leaves of neutral pH (Babic et al., 1996). Additionally, as Pseudomonas species are pectolytic, their presence is positively correlated with the degradation and spoilage of such leafy vegetables, which increases during storage, as observed in the present study. Exposure to solar active radiation influenced the relative abundance of the Betaproteobacteria and Gammaproteobacteria, which is the class level of the Pseudomonadales order (Truchado et al., 2017). Relative abundances of Gammaproteobacteria were not significantly different with reductions in cumulative photosynthetically active radiation (PAR) from 4,889 to 3,602 μmol m⁻² s⁻¹, but were substantially higher when cumulative PAR was 3,115 μmol m⁻² s⁻¹. In the present study, the protection of spinach and rocket produce from PAR by cultivating in a polytunnel setting, compared to an open field, did not lead to significantly higher Pseudomonadaceae content.

In the present study, L. monocytogenes populations of all groups were positively correlated with Pseudomonadaceae content. In particular, Pseudomonadaceae content appeared to be most important for L. monocytogenes growth on spinach F1 Trumpet produce, especially from day 7 to 9. Relative increases from days 7–9 for open field spinach produce were associated with L. monocytogenes’ most significant increase during the same period. Conversely, when Pseudomonadaceae content decreased from days 7–9 for polytunnel spinach, the L. monocytogenes populations remained stationary. Indeed, amino acids hydrolyzed from proteins by Pseudomonadaceae are localized within the cellular tissue of leafy vegetables (Koseki and Isobe, 2005; Vacher et al., 2016). Open field spinach produce is likely exposed to more liquids on leaf surfaces due to wetter outdoor climatic conditions, potentially causing higher leaching of those nutrients for L. monocytogenes utilization compared to polytunnel produce (Tukey, 1970; Comte et al., 2012; Vacher et al., 2016; Kyere et al., 2019; Zhu et al., 2022). Overall, spinach contained higher total abundances of Pseudomonadaceae than rocket. The leaf physiology of rocket, that is, less surface area and fewer stomata (Maylani et al., 2020) might have prevented the release of some nutrients, that is, amino acids (hydrolyzed protein) for L. monocytogenes utilization (Culliney and Schmalenberger, 2022), thus limiting the growth of bacteria more than this is the case for spinach. However, at the genus level, higher numbers of Pseudomonas sp. on Esmee than on Buzz do not seem to influence the growth potentials of L. monocytogenes.

Moreover, a higher Lactobacillales content was associated with the lower L. monocytogenes growth potential compared to both rocket Buzz and spinach F1 Trumpet produce. A recent study of a mixed spinach salad containing chicken meat identified low levels of Lactobacillales content, consisting of only Carnobacteriaceae and Enterococcaceae, which increased from 0 to 1% at day 7 of storage at 15°C (Söderqvist et al., 2017a). There, the authors did not detect any Lactobacillales on plain baby spinach. In another study, storage of romaine lettuce over 14 days revealed a significant increase in Carnobacteriaceae’s relative abundance from 1.93 to 52.26% and a non-significant increase in Pseudomonadaceae content from 13.38 to 21.20% (Dharmarha et al., 2019). Both bacteriocin-producing, for example, Divercin AS7 and non-bacteriocin-producing species of Carnobacteria, C. divergens and C. maltaromaticum, have been demonstrated to be effective in vitro at minimizing epithelial cell invasion caused by L. monocytogenes Scott A (Pilchová et al., 2016) and Listeria spp. (Marković et al., 2022). Carnobacteria piscicola LK5 and 2762 strains suppressed the maximum population density reached by L. monocytogenes in brain heart infusion broth (Buchanan and Bagi, 1997). However, little of the L. monocytogenes maximum population density suppression was due to the strain’s bacteriocin production. Those authors suggested that the suppression potential of the strain C. piscicola 2762 was not caused by peroxide, pH depression, or oxygen depletion, but was caused by induced nutrient depletion. In the present study, Carnobacteriaceae were absent from open field spinach produce but present in significantly higher quantities on polytunnel spinach produce, particularly at days 7 and 9. This may have also inhibited the growth L. monocytogenes, leading to its lower growth potential. Moreover, spinach F1 Cello variety had no Carnobacteriaceae present, but significantly higher Pseudomonadaceae content (+9.43%) compared to spinach F1 Trumpet from the polytunnel setting. Thus, potentially explaining the higher growth potential of the spinach F1 Cello variety. However, albeit a higher Pseudomonadaceae content (+5.58%), polytunnel spinach F1 Cello may have caused less leaching of nutrients (hydrolyzed amino acids) due to being less exposed to rain and liquid on surface of the leaf, thus resulting in lower L. monocytogenes growth potential for spinach F1 Cello (1.84 log₁₀ cfu g⁻¹) compared to open field spinach F1 Trumpet (2.59 log₁₀ cfu g⁻¹).

In addition to Carnobacteriaceae, polytunnel spinach F1 Trumpet, which displayed a lower growth potential of L. monocytogenes, showed an increasing trend in Pectobacteriaceae content (genus Dickeya). In contrast, open field spinach, which was associated with a larger L. monocytogenes growth potential, exhibited a decreasing trend in Pectobacteriaceae content (genus Dickeya). Additionally, polytunnel spinach F1 Cello, which had a decreasing trend of Pectobacteriaceae content (genus Dickeya), was associated with a higher L. monocytogenes growth potential than polytunnel spinach F1 Trumpet. Similarly, rocket Esmee had an increasing trend in Pectobacteriaceae content (genus Dickeya), whereas rocket Buzz, with a consistently lower Pectobacteriaceae content (genus Dickeya), was associated with higher L. monocytogenes colonization. Pectobacteriaceae spp., in particular the genus Dickeya, is a necrotroph that is known to cause soft rot, where deterioration of vegetables occurs from the secretion of plant cell wall-degrading enzymes (Bellieny-Rabelo et al., 2019; Wasendorf et al., 2022). Additionally, Pectobacterium spp. are associated with a type VI secretion system, which also targets plant pathogens lacking cognate immunity proteins by secreting bactericidal effectors and further releasing low molecular weight bacteriocins, that is, carocin, pectocin, and carotovoricin (Shyntum et al., 2019). Moreover, Pectobacterium, Dickeya, and Serratia spp. produce the β-lactam antibiotic carbapenem (1-carbapen-2-em-3-carboxylic acid). However, leafy vegetable isolates of Pseudomonas sp., which putatively influenced L. monocytogenes growth on spinach in this study, have been found to possess antibiotic resistance genes toward β-lactam antibiotics such as meropenem and colistin (Yin et al., 2022).

In 2016, Pectobacteriaceae was added to the Enterobacterales order. Prior to this, only a single Enterobacteraceae family existed for that order (Adeolu et al., 2016). In the present study, an unknown family from the Enterobacterales order was identified, with a relative abundance ranging from 0.00 to 33.46%. While the current study has no particular information on this new taxonomic bacterial group, the Enterobacteraceae of the same order possess the ability to produce colicins and microcins (Rebuffat, 2011). Microcins have proven ineffective against L. monocytogenes, but colicins produced with the help of the ColE1 gene are highly effective as an anti-listerial agent (Marković et al., 2022). Enterobacter spp., particularly Enterobacter cloacae, isolated from shredded iceberg lettuce, significantly reduced L. innocua colonization due to its nutritional competitiveness (Francis and O’Beirne, 2002).

Darlison et al. (2019) suggested that the influence of phyllosphere diversity on the proliferation of foodborne pathogens such as L. monocytogenes should be determined. Indeed, the significantly higher alpha diversity (Shannon index) of produce essentially appears to be correlated with lower L. monocytogenes growth potentials in the current study. However, the more diverse polytunnel rocket Buzz variety had more L. monocytogenes growth than the rocket Esmee variety. Indeed, higher and increasing Pectobacteriaceae content of Esmee, compared to the consistently low Pectobacteriaceae content, may be responsible for the 0.22 log₁₀ cfu g⁻¹ difference between those two growth potentials.

In the current study, seasonality was a significant driver of phyllosphere development in spinach. Bacterial diversity of the phyllosphere of Typha latifolia plants was not meaningfully influenced by short-term perturbations in weather conditions, such as rain events, but somewhat affected by seasonal climatic conditions and leaf-associated changes (Stone and Jackson, 2020). Darlison et al. (2019) suggested that annual variations resulting from varying weather conditions influenced phyllosphere communities of rocket and spinach. Although they could not rule out the effect of site-specific factors, as the produce was sampled in different parts of the same field over the 2 years. The present study accounted for site-specific factors by cultivating from the same location within both field and polytunnel settings, and also observed that weather parameters significantly influenced the spinach phyllosphere. Recently, the spinach phyllosphere has also been shown to be substantially influenced by seasonality (PERMANOVA, p < 0.003) (Ibekwe et al., 2021). An additional study revealed that the bacterial colonization of lettuce and rocket phyllosphere is also driven, at least in part, by seasonality (Dees et al., 2015).

To date, no previous studies have described the kale phyllosphere. Nevertheless, the kale endosphere has been recently studied (McNees et al., 2020). Across three different brands of store-purchased kale, Illumina sequencing of their endospheres revealed two common dominating Operational Taxonomic Units (OTUs) present were Pseudomonas and Enterobacteriaceae. In the present study, for kale, these, along with Micrococcaceae were also dominating families. Kale Nero di Toscana had the most similar content of Pseudomonadaceae as spinach F1 Cello. Although it demonstrated higher L. monocytogenes growth, due to the lower TBCs of kale (i.e., 2.80–4.74 log₁₀ cfu g⁻¹) and lower diversity, compared to rocket and spinach, less inhibition of the L. monocytogenes growth potentially occurs due to less competition for resources required for growth. Utilization of chloroplast-excluding protocols at the polymerase chain reaction (PCR) stage COMPETE (RInvT primer) (McManamon et al., 2019) or BLOCK (pPNA clamp) (Fitzpatrick et al., 2018; Culliney and Schmalenberger, 2024) as employed for open field spinach produce in a recent study, and would have been appropriate for rocket Esmee and kale Nero di Toscana. Their chloroplast-to-total DNA content was high, ranging from 58.00 to 97.30% (rocket Esmee) and 92.27 to 99.75% (kale Nero di Toscana), and thus could have prevented the exclusion of the 7 and 16 samples, respectively. Using the NGS approach, Listeria content was regularly detected on kale, but rarely occurred for rocket and spinach produce. The high TBCs of spinach and rocket may have contributed to this observation. Moreover, cultivation methods may not have detected cells that were at their viable but not-culturable (VBNC) stage (Müller and Ruppel, 2014). Thus, TBCs for all produce, including kale, may have been underestimated and, therefore, their total DNA content may have been associated with higher actual abundances. For example, a previous study used quantitative PCR (qPCR) and culturable techniques (TSA) to analyze lettuce samples from the same field and revealed that only 0.1–8.4% of TBCs were culturable bacteria (Rastogi et al., 2010). qPCR methods may be used in the future to enumerate TBC for this reason. However, qPCR-based quantifications may potentially overestimate bacterial population densities due to chloroplast co-amplification (Culliney and Schmalenberger, 2022) and multiple 16S ribosomal RNA (rRNA) gene copies per bacterial cell (Schmalenberger et al., 2001); hence, cultivation-dependent and independent approaches have biases. Furthermore, primer selection for 16S rRNA gene-based amplicon sequencing may also be responsible for an additional due to primer mismatch, which appears to be the case for L. monocytogenes 16S with the popular V3–V4 primers.

Ibekwe et al. (2021) revealed that the four common dominating phyla Proteobacteria, Firmicutes, Bacteroidetes, and Actinobacteria, which comprised 66.35% of their phyllosphere, were significantly different from the overall abundance, ranging from 88.21 to 99.92% of those four main phyla for spinach in the present study. Additionally, their Pseudomonadaceae content (0.49–11.5%) was, on average, lower than the Pseudomonadaceae content observed on spinach, rocket, and kale produce in this study. With an overall relative abundance of 35–53%, Pseudomonas has been referred to as the most commonly occurring genus in the spinach and rocket phyllospheres, even after being harvested in different seasons (spring and autumn) (Rosberg et al., 2021). Upon closer inspection of the Pseudomonadaceae family’s relative abundances, potential seasonal effects exist, especially for spinach. However, in the current study, the relative abundances of winter and summer open field spinach did not show significant differences. However, there was still a large difference of 0.94 log₁₀ cfu g⁻¹ between their L. monocytogenes growth potentials. LAB are more commonly detected on leafy produce cultivated in spring and summer compared to autumn and winter (Caponigro et al., 2010); however, the opposite was true in the current study. With a relative abundance of less than 1%, Lactobacillales may have been responsible for the significant growth potential difference. The Lactobacillales decreased from 0.52 to 0.22 to 0.03% from days 5–9 for summer produce, which was correlated with large increments in L. monocytogenes growth, which did not occur when Lactobacillales remained constant and on average in higher relative abundance for winter produce. More specifically, winter produce with lower L. monocytogenes growth had a significantly higher content of the Lactococcus genus (Streptococcaceae family; Lactobacillales order). Indeed, L. lactis subsp. lactis has been previously isolated from rocket leaves and is known as a bacteriocinogenic strain due to its ability to produce lantibiotic, which is an antimicrobial nisin variant that is highly effective as an anti-listerial agent on food products, including iceberg lettuce (Franz et al., 1997; Kruger et al., 2013; Ho et al., 2018; McManamon et al., 2019; Ho et al., 2021).

The remainder of phyllosphere-associated bacteria, which showed positive and negative correlations with L. monocytogenes populations identified in this study, did not appear to be potentially responsible for the conflicting epiphytic L. monocytogenes growth on spinach or rocket leaves. Correlations were determined using Pearson’s correlation, which is primarily used for linear relationships between two continuous variables, due to the normal distribution and increasing L. monocytogenes populations over time. However, a recent study revealed that Pearson’s can also be more efficient in testing a monotonic nonlinear relation compared to Spearman’s (van den Heuvel and Zhan, 2022). Future studies may use Spearman’s correlation as it evaluates the monotonic relationship between two continuous variables (Schober et al., 2018). Indeed, this approach is most often used for bacterial growth curves, which reach the stationary phase. In either case, such correlations must be interpreted with caution. For example, Zhao et al. (2021) revealed that association means that one variable provides information about another, whereas correlation means that two variables show an increasing or decreasing trend. Therefore, correlation implies an association, but not causation. Additionally, due to the absence of absolute numbers upon sequencing (Gloor et al., 2017), comparing relative abundances could lead to inaccurate conclusions when comparing phyllosphere microbiome over time or when comparing different phyllosphere communities, for example, kale or spinach, which have considerably different absolute cfu data, as relative data reflect a different amount of absolute numbers. Future studies should conduct correlations between absolute cfu data, that is, total bacterial populations and relative abundances from NGS datasets that are converted into absolute values via an additional qPCR step.