This study reveals the bacterial, seed‐endophytic diversity harboured across a range of European hemp cultivars. To our knowledge, this is the first study to apply a culture‐independent sequencing approach to profile the seed‐endophytic microbiome of hemp. In addition, the 16 cultivars in this study represent the largest diversity panel to which the 16S rDNA amplicon sequencing approach has been applied in any hemp microbiome study. Our results provide novel insights into the diversity and taxonomic composition of hemp seeds and the extent of variation among seed lots.

At the phylum level, our taxonomic profiles of the hemp seed microbiome are similar to those obtained in previous studies. While Cyanobacteria contributed a small number of reads, the communities were otherwise comprised exclusively of Proteobacteria, Bacteroidota, Firmicutes, and Actinobacteriota, four phyla that have been found to dominate the seed microbiomes of other plant species (Simonin et al., 2022; War et al., 2023). The five phyla in our dataset also comprise the five most abundant phyla reported across the hemp root, leaf, and flower endopheres (Wei et al., 2021). The most abundant phylum in hemp seed was Proteobacteria, which was also reported by Willman et al. (2021) to dominate hemp root, leaf and bud endospheres, alongside Bacteroidota and Actinobacteriota in the root. However, Ahmed et al. (2021) reported the hemp root and rhizosphere microbiomes to be strongly dominated by Planctobacteria (Planctomycetota), which was absent from our dataset.

The culturable bacterial community of hemp seed has been previously isolated from the cultivar Futura 75 (Gabriele et al., 2022). All stains isolated in that study were attributed to the phyla Proteobacteria, Firmicutes, and Actinobacteriota but none to Bacteroidota, the second most abundant phylum in our dataset. Of the genera included among the closest BLAST matches to the stains isolated by Gabriele et al. (2022), Bacillus, Cellulomonas, Curtobacterium, Paenibacillus, Sphingomonas, and Stenotrophomonas were also included in our dataset, but three were not: Kocuria, Psychrobacillus, and Staphylococcus. Our dataset included 81 genera that were not isolated by Gabriele et al. (2022), including the 3 most abundant genera in our dataset. This may reflect any combination of plant genetic and environmental effects on seed microbiome assembly (Klaedtke et al., 2016; Wassermann et al., 2022) and differences in the diversity captured by culture‐dependent and ‐independent methods (Cope‐Selby et al., 2017; Jackson et al., 2013).

Alpha diversity was found to highly variable among cultivars by each metric analysed, including the number sequence variants (ASV richness), a diversity index that incorporates the richness and evenness of ASVs (Shannon's diversity index) and a measure of phylogenetic diversity (Faith's PD). The cultivars Eletta Campana, CS, and Earlina 8 FC were among the top five by each alpha diversity metric, while Markant 3 was always the least diverse cultivar. The observed median ASV richness (90) was higher the median calculated (40) from a cross‐species meta‐analysis of similar studies profiling the microbiomes of surface‐sterilized seed (Simonin et al., 2022). It is much lower than the ASV richness figures reported from endosphere samples from the root (1141), leaf (342), and bud (181) of the hemp cultivar Tangerine (Willman et al., 2021). The seed microbiome was found to harbour lower diversity than other endophytic compartments in a culture‐based study in hemp (Scott et al., 2018). This pattern of diversity may be explained by requirements for specialist characteristics required to colonize and survive within seeds (War et al., 2023).

Eighty‐seven genera were included in the dataset, though a relatively small subset dominated most samples. The 10 most abundant genera overall collectively represented the majority of every sample (Figure 1). Furthermore, three taxa (Flavobacterium, Pseudomonas, Pantoea) collectively dominated most samples. Pseudomonas and Pantoea are very commonly observed in seed microbiomes—the most prevalent and abundant ASVs across 50 plant species included taxa assigned to these 2 genera (Simonin et al., 2022). The consistent recruitment of certain taxa into hemp seed microbiomes may be driven by a combination of factors such as ubiquity in the surrounding environment and possession of traits that facilitate colonization, survival, and inheritance within seed, potentially including traits that preferential selection by the host plant. Flavobacterium, Pseudomonas, and Pantoea each include members previously found to confer plant beneficial effects (Gontia‐Mishra et al., 2016; Walterson & Stavrinides, 2015; Weller, 2007).

Association with any one specific taxon does not seem obligate for hemp seed. No taxa are present in every sample, and even the most common taxa are absent or present at low abundance in several samples. The high flexibility of the communities is exemplified by the almost complete dominance of seed of the Markant by Brevibacillus, an uncommon taxon across most of the dataset. At the genus level, no fraction of the hemp seed microbiome was fully stable to be defined as a core microbiota, based on the commonly applied definition of 100% prevalence across all samples (Berg et al., 2020; Neu et al., 2021), though the high prevalence and abundance of some taxa may still suggest key ecological and functional roles within the microbiome such as those hypothesised for core taxa (Neu et al., 2021).

In all cases there was high similarity in community composition across replicates—for samples from a given cultivar, across which plant genotype and likely environmental conditions were consistent, similar assemblages of bacteria were recruited into seeds in similar proportions. This supports findings that deterministic processes are more important than stochastic factors in microbiome assembly, especially in seed (Guo et al., 2021). There was much greater variation in community composition across cultivars. These differences cannot be confidently attributed to host genotype as the cultivars were sourced from across several suppliers—some cultivars are therefore likely to differ substantially in the environmental conditions experienced during cultivation and post‐harvest, including those related to agronomic and storage practices. While some endophytes are likely inherited from seed to seed (Johnston‐Monje & Raizada, 2011; Kim et al., 2022), external factors are expected to affect seed microbiome composition as the majority of the seed microbiome seems to originate from environment, especially the soil (Rodríguez et al., 2020). Soil‐derived bacteria can enter from the rhizosphere and spread throughout the plant endosphere, eventually colonizing the developing seed (Compant et al., 2011). Bacteria harboured by pollen, flowers, or fruits may also be able to colonize seed (Frank et al., 2017; Shade et al., 2017). Seed microbiome assembly has been shown to be influenced by geographic location, year‐to‐year variation in environmental conditions, and storage conditions (Chandel et al., 2021; Dutta et al., 2022; Klaedtke et al., 2016; Morales Moreira et al., 2021; Rochefort et al., 2019). For example, Markant 3, as the only cultivar in this study to be sourced from a Dutch supplier, it is not possible to attribute its unusual Brevibacillus‐dominated microbiome to either genotypic or environmental effects, as one confounds the other.

On the other hand, cultivars sourced from the same supplier often displayed marked differences. This may indicate a genetic effect on community assembly, which has also been reported for the seed microbiome of other crops (Adam et al., 2018; Chen et al., 2022; Johnston‐Monje & Raizada, 2011; Kim et al., 2020; Rybakova et al., 2017; Wassermann et al., 2022) and the belowground microbiome of cannabis (Comeau et al., 2020). However, environmental effects may still explain the differences observed in the study at hand, as cultivars sourced from the same supplier may still have differed in their year and location of cultivation. An experimental study that controls for environmental variation is required to reveal the influences of host genetic factors or vertical inheritance on hemp seed microbiome assembly.

Regardless of the source of variation, this study has revealed that hemp seed lots can harbour distinct seed‐endophytic bacterial communities. The observed differences in seed microbiome composition may have consequences for plant performance. Some taxa detected within the hemp seed include seed‐transmitted strains demonstrated to have plant beneficial effects, such as Bacillus (Mukherjee et al., 2020), Kosakonia (Jeong et al., 2021), Pantoea (Rahman et al., 2018), and Pseudomonas (White et al., 2018). Bacillus and Pseudomonas species have also been found to promote growth and reduce disease severity in hemp (Balthazar et al., 2021; Comeau et al., 2021). Most of the hemp seed‐endophytic microbiome remains unexplored in terms of functional characterization. Gabriele et al. (2022) found that some endophytes cultured from hemp seed were capable of producing indole‐3‐acetic acid, which can promote plant growth (Duca et al., 2014), though germination tests conducted with one such strain yielded no evidence of biostimulation (Gabriele et al., 2022).

This dataset represents a key starting point for further work on the hemp seed microbiome. A next step should be the profiling of its fungal fraction, which has received little attention beyond the isolation of taxa in culture (Scott et al., 2018). Seed‐endophytic fungi can contribute to host performance (Rétif et al., 2023) and may influence the composition of the bacterial fraction (Tannenbaum et al., 2020). Future studies should also aim to supplement taxonomic profiles with functional characterization of these seed‐endophytic communities. Culturable strains could be evaluated for plant‐growth‐promoting characteristics using in vitro tests and inoculation experiments, while community‐level function could be assessed using culture‐independent approaches, such as qPCR to quantify the abundance of bacterial genes of known functional importance to the plant (Wassermann et al., 2023), or the experimental transplant of entire microbiomes (Tosi et al., 2020). Elucidating whether members of the hemp seed microbiome contribute to plant growth and health, either at the level of individual strains or community‐level interactions, will determine the potential viability and efficacy of microbiome engineering in hemp seed. Also vital to this would be an understanding of the dynamics of the hemp microbiome beyond germination—host‐beneficial seed‐endophytes that persist within the maturing plant would represent promising candidates for further study and utilization.