Introduction

The plant rhizosphere is an essential zone directly adjacent to the plant root system that hosts a diverse and functionally dynamic microbial consortium (Philippot et al. 2013; Shi et al. 2016; Ling et al. 2022; Mukhtar et al. 2023). These microbes play a vital role in nutrient cycling, organic matter decomposition, and plant defense against different stressors and diseases (Haney et al. 2015; Hu et al. 2018; Gu et al. 2020). Therefore, it is important to comprehend the shift in rhizosphere microbial community composition and stability to changing edaphic environments across diverse plant genotypes. Such knowledge is crucial not only to engineer resilient microbial consortia for varying field conditions but also to model the global microbial distribution in agroecosystems.

Nitrogen (N) fertilization is one of the most important factors influencing the soil edaphic environment and, subsequently, soil microbial community diversity and composition (van der Bom et al. 2018; Mukhtar et al. 2023). Previous research has demonstrated varying impacts of N application on belowground microbial communities, yielding contrasting results and conclusions. These range from substantial positive or negative influences (Yang et al. 2022; Mukhtar et al. 2023) to instances where no significant alterations in community diversity and composition were observed (French et al. 2017; Ouyang and Norton 2020; Duff et al. 2022). For example, N fertilization has been reported to enhance nutrient availability and to support a diverse microbial community (van der Bom et al. 2018; Li et al. 2020), while alleviating N stress can foster stable microbial communities (Hernandez et al. 2021). Conversely, excessive N fertilization (and N deposition) can adversely affect soil microbial diversity (Zhang et al. 2018; Hao et al. 2022; Yang et al. 2022; Wang et al. 2023), thereby negatively impacting ecosystem functioning. Excess N fertilization also led to increased carbon deposition that resulted in differences in the labeled carbon uptake by microbial groups in two grass species (Leptin et al. 2022) highlighting the importance of N additions in influencing plant root carbon release and the subsequent impact on the soil microbiome. Similarly, the complexity and modularity of soil microbial networks have also been shown to vary with N application across various field conditions (Li et al. 2021; Liu et al. 2021; Zhang et al. 2021). While the responses of soil microbial communities have been documented, our current understanding of how rhizosphere microbial communities respond to N fertilization (and N stress) remains inconclusive.

Host plant-microbiome interactions introduce an additional layer of complexity and uncertainty in discerning the microbial response to N levels. On the one hand, the genetic makeup of the host plant profoundly influences the physical and chemical characteristics of the rhizosphere (Sasse et al. 2018; Wang et al. 2022; Lopes et al. 2023), which, in turn, filters and recruits different microbial taxa. Notably, variations in primary and secondary metabolites in root exudates play a pivotal role in shaping the soil microbiota, as these compounds serve as an essential source of energy and a specialized environment that enhances growth as well as influences the composition of microbial communities (Sasse et al. 2018; Zhalnina et al. 2018; Jacoby et al. 2021). On the other hand, host plant genotypes exhibit diverse root system architecture, which may directly or indirectly exert influence on the composition and functionality of microbial communities (Szoboszlay et al. 2015; Saleem et al. 2018) by altering various soil biophysical and edaphic properties. These variations in root exudates and root structure orchestrate the changes in the structure and function of soil microbes and gene expression in the rhizosphere, which varies depending on soil properties and developmental stages (Hao et al. 2022; Lopes et al. 2023).

Plants may also adjust their nutrient acquisition strategies in response to N availability—e.g., changes in resource allocation (e.g., enzyme production) under differing nutrient conditions —impacting the composition of root exudates (Zhu et al. 2016; Hao et al. 2022). Changing N conditions also shifts gene expression profiles and morphological characteristics of plant root systems (Chen et al. 2011; Gaudin et al. 2011; Singh et al. 2022). This could influence the proportion of variance in microbial relative abundance associated with plant genotype. Nevertheless, these evolving plant root- microbe interactions are likely to exert their influence on soil biogeochemical cycling (Mukhtar et al. 2023) and have been shown to impact plant phenology and expression of heterosis (O’Brien et al. 2021; Wagner et al. 2021). These complex interactions and feedback make it essential to understand the shift in rhizosphere microbial community composition, and its complexity and stability in response to changing N levels and how these responses differ between diverse plant genotypes.

In this study, by using 305 inbreds and 196 F1 hybrids, we investigated the potential impacts of maize genotype on the rhizosphere microbial community composition, microbial heritability, and co-occurrence patterns under two distinct N schemes: nutrient deficient soils (low-N) and fertilized soils (high-N). Maize was chosen as the focal crop because of its economic importance, widespread global cultivation, the remarkable adaptability of hybrids within agroecosystems and considerable difference in microbial community composition between inbred lines and their hybrid rhizospheres (Wagner et al. 2020). Although, soil edaphic and climatic conditions (Bahram et al. 2018; Mukhtar et al. 2020, 2023), as well as the endosphere (Lopes et al. 2022; Wang et al. 2022), significantly influence belowground microbial community structure, this study primarily focused on changes in soil microbial community dynamics within the rhizosphere of a large population of maize inbreds and hybrids in response to different N treatments. We hypothesized that N input will: 1) differentially influence the microbial diversity of rhizospheres across different plant groups (inbred lines and their hybrids), 2) reduce the proportion to which the variation in microbial relative abundance could be attributed to plant genetic factors with a stronger impact in hybrid maize, and 3) increase the modularity and complexity of microbial networks, especially in rhizosphere of inbreds. The outcome of this study contributes to elucidating the combined effects of diverse plant lineages and contrasting N availability that ultimately shapes the structure of rhizosphere microbial communities in maize production ecosystems.

Material and methods

Experimental design and sample collection

A total of more than 500 maize inbred lines and hybrids were planted (in May, 2022) at a University of Nebraska—Lincoln, facility (40.867695, -96.603074). The area was rain-fed and characterized by a mean annual temperature of approximately 12 °C and a mean annual precipitation of around 760 mm. The soil was slightly acidic, primarily silty clay loam, with an organic matter content varying from 2.3% to 3.6% and a nitrate level from 4.1 to 7.0 ppm prior to fertilization. The inbred lines are doubled haploid (DH) lines released from the USDA GEM (Germplasm Enhancement of Maize) project (Vanous et al. 2018). Hybrids were generated in 2020 and 2021 by crossing of BGEM lines with two tester inbred lines B73 and Mo17. Specifically, BGEM lines derived from PHZ51 as the recurrent parent were crossed with B73, while BGEM lines from PHB47 were crossed with Mo17. The selection of B73 or Mo17 lines was due to their high importance as stiff-stalk and non-stiff-stalk varieties. A description about BGEM lines, parental race, and the origin has been previously described (Vanous et al. 2018).

We further grouped the inbred lines into eight subgroups based on stalk stiffness (stiff or non-stiff), flowering time (early or late), and plant height (short or tall). Hybrid groups were similarly categorized into eight subgroups, corresponding to their respective parental BGEM inbred lines and their crosses with either B73 or Mo17. Each subgroup comprised between 14 to 54 different genotypes, and these genotypes were replicated in four blocks (following an incomplete block design): two blocks for unfertilized soils, referred to as the 'low nitrogen treatment' (low-N), and two blocks for fertilized soils with 168 kg/ha of N as urea, referred to as the 'high nitrogen treatment' (high-N). Each row spanned 4.5 m, and 20 seeds were planted per row.

After eight weeks of planting (July 2022), the plant roots were extracted directly from the field soil at a depth of 20 to 30 cm. The roots were shaken to remove loosely adherent bulk soil. To collect rhizosphere samples in the field, 5 – 8 segments of randomly selected roots, including adherent rhizosphere soil was placed into 50 ml tubes containing 35 ml of phosphate buffer and then vortexed for two minutes to remove most of adherent rhizosphere soil. Roots were then removed, and tubes were placed on ice and brought into the laboratory. The rhizosphere suspensions were then filtered through a 100 µM sterile mesh into a fresh 50 ml tube to remove root debris. Rhizosphere samples were then spun down, followed by removal of the supernatant. Then 2 ml of phosphate buffer was added to the rhizosphere soil which was then resuspended and transferred to a 2 ml tube. This was then spun down again to allow for the removal of the supernatant and the rhizosphere soil was then frozen at –20 °C. For a detailed protocol please see (McPherson et al. 2018).

16S library preparation and Sequencing data analysis

DNA was extracted from the rhizosphere soil using a MagAttract PowerSoil Pro DNA Kit Soil DNA Isolation kit (Qiagen LLC, Germantown, MD). The V4 region of the 16S rRNA gene was then amplified from the rhizosphere DNA using the primers 515F and 806R (Caporaso et al. 2012). To amplify this region PCR was used with initial denaturing at 95 °C for 5 min, 30 cycles of 95 °C for 40 s, 65 °C for 30 s and 72 °C for 1 min, followed by final extension at 72 °C for 10 min. The sequencing libraries were prepared using a dual-index primer system (Kozich et al. 2013). Amplicons were purified using an NGS 96-well normalization kit (Norgen Biotek Corp, ON, Canada) and then normalized based on the DNA concentrations which were quantified using QuantiFluor ONE dsDNA Dye (Promega Corporation, Madison WI) and then pooled in an equimolar mixture, and further purified using the SPRIselect beads (Beckman Coulter, Brea, CA). Sequencing was then performed on an Illumina MiSeq platform using the MiSeq 600 cycles v3 kit. Sequences generated in this study are found on NCBI (accession number: PRJNA1092665).

Raw sequence data of forward and reverse reads (FASTQ) was processed within the QIIME2 environment (Bolyen et al. 2019) (release 2023.5) and denoised sequences within QIIME2 using the DADA2 pipeline. We clustered the remaining sequences into amplicon sequence variants (ASVs) against the SILVA 138.1 database (Quast et al. 2013) using a pre-trained Naïve Bayes feature classifier. Raw ASV reads were subjected to a series of filters to produce a final ASV table with biologically relevant and reproducible 16S sequences. Samples with 1) sequencing depth < 3000, 2) ASVs that were not observed in at least 25 samples were removed. Chloroplast and mitochondrial sequences were eliminated from all samples. A total of 7,007 ASVs—with a total frequency of 41,912,761 and a median frequency per sample of 20,153—were obtained and classified into the 293 genera and 24 known phyla.

Microbial abundance and diversity analyses

The relative abundance (%) of microbial taxa was individually quantified at both the phylum and genus levels for each sample, and subsequently averaged across inbred and hybrid plants, including their subgroups, as well as across various nutrient treatments by using base functions in R. Alpha and beta diversity were assessed using the ‘vegan’ package in R (Oksanen et al. 2022). Alpha diversity was measured using the inverse Simpson index, which provides insights into species richness and evenness within individual samples. ANOVA, followed by a post-hoc test (Tukey's test) that was used to determine whether there were significant differences between groups (inbred and hybrids) and subgroups based on stalk stiffness (stiff or non-stiff), flowering time (early or late), and plant height (short or tall) under low- and high-N treatments (n = 14 – 54 for alpha diversity).

Beta diversity was evaluated using the Bray–Curtis dissimilarity index, which quantified the variation in species composition between different samples. The data were rarefied for alpha diversity, ensuring that all samples had the same number of sequences, while non-rarefied data were used for beta diversity and network analysis, preserving the original sequencing depth for a comprehensive assessment of community composition and interactions. Additionally, statistical tests such as Analysis of Similarities (ANOSIM, method = Bray) and Multivariate Analysis of Variance Using Distance Matrices (Adonis) were applied to determine the significance of dissimilarity between sample groups (n = 14 – 54), allowing for the comprehensive analysis of diversity patterns in the dataset.

Plant genotype driven microbial variation

We estimated the proportion to which the variation in microbial relative abundance could be attributed to plant genetic factors (hereafter referred to as VPM) for microbial groups from the rhizosphere of all genotypes in two replicated blocks at the genus and phylum levels. The VPM was estimated across the experiment including both inbred and hybrid rhizospheres, as well as separately for each group under low-N and high-N treatments. We characterized both higher and lower taxonomic levels to obtain a more comprehensive understanding of the plant genotype-associated microbial taxa and to investigate whether, instead of the entire community, only a subset of the microbiome is responsive to host variation. VPM was estimated by using proportion of variance explained by the genotype (Vg) relative to the total variance, which includes both (Vg) and error (Ve) and N is the number of replicates (2) (Eq-1).

$${V}_{PM}= \frac{{V}_{g}}{{V}_{g}+{V}_{e}/N}$$
(1)

Taxa that appeared in fewer than 200 samples, representing roughly 10% of the total samples, were omitted, and the remaining data were log-transformed before conducting the VPM analysis. The 'lmer' function from the R package 'lme4' was used to construct linear mixed-effects models with all random effects. Finally, a permutation test with 5000 iterations was conducted to assess whether the observed VPM is attributable to random variability or was statistically significant. Subsequently, rhizosphere microbial taxa that were significantly associated with plant genetic factors (p < 0.05) are referred to as plant genotype-associated microbial taxa (HT), while microbial taxa that were not significant were removed from analysis.

Network analysis

Co-occurrence networks were constructed to investigate microbial interactions within subgroups of both inbreds and hybrids under low-N and high-N conditions. We created distinct networks for each of the eight inbred subgroups, as opposed to forming two large networks encompassing both inbred and hybrid individuals by utilizing all genotypes within the subgroups and their two replications. This approach allowed for a more nuanced analysis of network structures while avoiding the potential confounding effects of amalgamating all inbreds into a single network. We constructed co-occurrence networks using function sparcc (iter = 100, and other parameters with default values) from R package “SpiecEasi” with ASVs present in > 10% samples. To enhance network clarity and reduce the occurrence of false positives, we applied a correlation cutoff of 0.3 (|correlation| \(\ge\) 0.3), accompanied by a significance threshold of 0.05 (p < 0.05). Significance of correlations was calculated using 5000 bootstraps. Group used for the network statistical analysis contained between n = 14 – 54 genotypes.

The co-occurrence network graphs were generated using the 'igraph' package in R (Csardi and Nepusz 2006). Comprehensive insights into network properties were obtained through the computation of key parameters, including node count, edge count, modularity (M) (from absolute weight), and clustering coefficients (CC, type = ’global’). All statistical analyses were conducted using the R programming software, unless stated otherwise.

Results

Rhizosphere microbial diversity and community structure

We first assessed the relative abundance of microbial groups in the rhizosphere at differing taxonomic levels between maize inbreds and hybrids under low- and high-N conditions. A total of 152 and 142 out of 293 microbial groups (at the genus level) differed between inbred and hybrid under low-N and high-N treatments, respectively. Pseudomonas was the dominant bacterial taxon at the genus level and was significantly more abundant in the rhizosphere of hybrids (ANOVA, p < 0.05, Supplementary Table 1). The mean relative abundance of Pseudomonas in hybrids was 28.3% for low-N and 41.5% for high-N, compared to inbreds with 22.4% for low-N and 20.1% for high-N. The second most abundant taxon at the genus level in the rhizosphere of hybrids was an unclassified taxon of the phylum Acidobacteriota whose mean relative abundance ranged from 6.9% in low-N to 5.6% for high-N (Fig. 1a). In contrast Ralstonia was the second most abundant in the rhizosphere of inbreds ranging from 4.8% in low-N and 7.6% in high-N. Moreover, the BCP (Burkholderia-Caballeronia-Paraburkholderia), commonly found in bulk soil and the rhizosphere soil was in lower abundance (p < 0.05) in the rhizosphere of hybrids compared to inbred lines. The relative abundance of BCP in the rhizosphere of inbreds was on average 2.9% in low-N and 4.2% in high-N conditions, while its abundance was less than 1% in the hybrid samples under low-N conditions. At the phylum level, a total of 15 phyla under low-N and 12 under high-N conditions (out of 24) significantly differed (p < 0.05) in relative abundance between inbreds and hybrids. Among them, Proteobacteria, Acidobacteriota, and Actinobacteriota were the dominant bacterial taxa, with mean relative abundances of 49.6%, 11.5%, and 8.3%, respectively, across samples.

Fig. 1
figure 1

Dominant microbial taxa and alpha-diversity in the rhizosphere across maize inbreds and hybrids under different N treatments. (a) The relative abundance of the dominant microbial taxa (top 10 dominant taxa) across inbred lines and hybrids at the genus level under low-N (LN) and high-N (HN). Microbial alpha-diversity (measured as the Inverse Simpson index) (b) across inbred and hybrid lines and (c) across different subgroups of inbreds—classified based on stalk stiffness, flowering time, and plant height (SE, TE, SL and TL)—and corresponding hybrids (H1 to H4) under LN and HN treatments. S: Stiff stalk, N: Non-stiff stalk. SE: Short-early, TE: Tall-early SL: Short-late, and TL: Tall-late. WD2101 (soil type). * p < 0.05 and ** p < 0.01

Full size image

The alpha diversity was significantly influenced by both plant groups (inbreds and hybrids) and N treatments (low- and high-N) (ANOVA, F: 160.0 and 8.9, respectively, p < 0.01) (Fig. 1b). The alpha diversity in the rhizosphere of inbreds was significantly higher than their hybrid counterparts, particularly under high-N conditions (Fig. 1b). Moreover, the alpha diversity for the rhizosphere microbial communities of the hybrids under low-N (ranging from 1.5 to 41.5, mean: 13.4) was, on average, more than two-fold higher than those values observed under high-N treatments (ranging from 1.3 to 35.4, mean: 7.7). However, the N treatment significantly increased (p < 0.05) the alpha diversity in inbred rhizosphere (Fig. 1b) suggesting a contrasting influence of N treatment on rhizosphere microbial alpha diversity across different plant groups.

We also assessed the change in alpha diversity between subgroups of inbred lines and their corresponding hybrids. The subgroups SE (short-early), SL (short-late), TE (tall-early), and TL (tall-late) represent plant height and flowering time, with S and N indicating stiff stalk and non-stiff stalk, respectively. The corresponding hybrids are represented by H1, H2, H3, and H4, respectively (Fig. 1c). Within inbred lines, the highest alpha diversity was observed for TL (20.3 for S and 22.4 for N), followed by SE (18.2 for S stalks) under high-N conditions. The rhizosphere communities of the maize hybrids, on the other hand, exhibited a significant reduction in the alpha diversity for almost all groups compared to their parental inbreds (BGEM). The most significant difference in alpha diversity between inbred and hybrid rhizosphere microbial communities was observed between inbred subgroup TL (20.3 for S stalk and 22.4 for N stalk) and their corresponding hybrid subgroup H4 (8.6 for S stalk and 5.9 for N stalk) under high-N treatments. Hybrid subgroups (H1 to H4) were statistically indistinguishable (p > 0.05) from each other under high-N treatments.

The plant groups (inbred and hybrid) and N treatments (low- and high-N) also significantly influenced the microbial community composition across the maize rhizosphere (Fig. 2). Adonis (permutational multivariate analysis of variance using distance matrices) showed that microbial community composition differed significantly across both plant groups (F = 105.6, p < 0.01) and N treatments (F = 41.5, p < 0.01), as well as the interaction between plant groups and N treatments (F = 23.6, p < 0.01). We further conducted ANOSIM (Analysis of Similarities, distance matrix: Bray–Curtis) to assess whether dissimilarity within the plant groups was different (i.e., smaller) than those between groups. The bacterial and archaeal communities in the rhizosphere of hybrids were distinct from those observed in the rhizosphere of inbreds under high-N treatments (R = 0.1, p < 0.001), while no significant difference was observed under low-N treatments (Fig. 2a).

Fig. 2
figure 2

Microbial beta-diversity and clustering across maize inbred and hybrid rhizosphere communities under different N conditions. Changes in rhizosphere microbial community composition (Bray–Curtis) between maize inbreds and hybrids under (a) low-N and (b) high-N treatments. The ellipses are drawn at the 95% confidence level

Full size image

Plant genotype-driven rhizosphere microbial traits

We investigated the proportion of variation in microbial relative abundance associated with plant genotype (abbreviated as VPM) at both the genus and phylum levels under different N treatments. A high VPM value would suggest that maize genetic diversity plays a significant role in shaping the microbial communities, while a low value would indicate that environmental factors may have a stronger influence. When measuring VPM across the samples including both inbreds and hybrids, a total of 38 microbial taxa under low-N and 21 under high-N, at the genus level, were detected as significantly plant genotype-associated microbial taxa (HT) with a mean VPM of 0.21 and 0.25 (p < 0.05) (Fig. 3). Among HT, Ralstonia exhibited the highest proportion of variance associated with plant genotype (VPM = 0.41, p < 0.05), while the BCP was the second most plant genotype-associated taxon (VPM: 0.34, p < 0.05) and Niastella was the third most plant genotype-associated taxon (VPM = 0.33, p < 0.05) under low-N treatments (Fig. 3a). In contrast, Amycolatopsis showed the highest VPM (0.39, p < 0.05), followed by Sphingobium in second place and Ralstonia in third (VPM: 0.36, p < 0.05) under high-N treatments. Moreover, Pseudomonas, the most dominant taxon in our samples, displayed a moderate VPM (0.26, p < 0.05) under high-N. We also noted that VPM was positively correlated with microbial relative abundance under low-N treatments. The correlation between microbial relative abundance and VPM was r = 0.48 (Pearson’s correlation test, p = 0.002) for low-N and r = 0.26 (Pearson’s correlation test, p = 0.25) for high-N.

Fig. 3
figure 3

The proportion of variance in microbial relative abundance associated with plant genotype (VPM) in the rhizosphere at the genus level. a the top 10 plant genotype-associated microbial taxa (HT) (ranked based on VPM values) along with their relative abundance under both low-N (LN) and high-N (HN) conditions (* p < 0.05). b the magnitude of microbial-plant genotype associations (HT × VPM), and (c) VPM for rhizobiome traits under low-N and high-N treatments for both inbred and hybrid lines. The colors indicate whether VPM are significant (p < 0.05) among either inbred (pink), hybrid (pink), or both (green), while gray represents taxa with p > 0.05. NA* Unclassified taxa

Full size image

We further quantified the VPM for taxa in rhizospheres of inbreds and hybrids separately and identified those that exhibited significant VPM in either inbred or hybrid samples, or in both groups. When measuring VPM across the rhizosphere of inbreds, the number of HT taxa was 31 in low-N treatments and 17 in high-N treatments, each with a mean VPM > 0.2 (p < 0.05). However, the HT was reduced in the rhizospheres of hybrids, particularly under high-N treatment, with only 7 microbial taxa with a mean VPM of 0.25 (Fig. 3b). Moreover, Pseudomonas, the most dominant taxon in our samples, displayed a nonsignificant association with both the rhizosphere of inbreds and hybrids (p > 0.05) under both low-N and high-N (Fig. 3a). When identifying plant genotype-associated microbial taxa present in both inbred and hybrid rhizospheres, a total of 12 microbial groups (9 under low-N and 3 under high-N) were found to be in common (Fig. 3c). Among them, all plant genotype-associated microbial groups under high-N conditions and 8/9 microbial groups under low-N were associated with Proteobacteria, while one microbial group (under low-N) was associated with Actinobacteriota.

At the phylum level, Desulfobacterota showed significant VPM (p < 0.05) under high-N treatments, while Bacteroidota, Actinobacteriota, and Crenarchaeota taxa showed significant VPM (p < 0.05) under low-N treatments (Fig. 4) when measuring VPM across the rhizosphere samples of both inbreds and hybrids. On the other hand, in the rhizosphere of inbreds, there were four microbial taxa at the phylum level in low-N treatments with a mean VPM > 0.15 (p < 0.05). There was one taxon under low-N and one taxon under high-N treatment with a mean VPM > 0.15 (p < 0.05) in the rhizosphere of hybrids.

Fig. 4
figure 4

The proportion of variance in microbial relative abundance associated with plant genotype (VPM) in the rhizosphere at the phylum level. (a) the magnitude of microbial-plant genotype associations (no of genotype-associated microbial taxa (HT) × VPM), (b) the plant genotype-associated microbial taxa (HT) (ranked based on VPM values across samples) under both low-N (LN) and high-N (HN) conditions (* p < 0.05) and (c) VPM for rhizobiome traits under low-N and high-N treatments for both inbred and hybrid lines. The colors indicate whether VPM are significant (p < 0.05) among either inbred (pink), hybrid (pink), or both (green), while gray represents taxa with p > 0.05. UC: Unclassified taxa

Full size image

Microbial network topology and complexity

Co-occurrence networks were established to better understand change in interaction between microbial taxa at the genus level across inbred and hybrid rhizospheres in response to N treatment. Microbial networks were constructed for each subgroup of maize inbred lines (classified based on stalk stiffness, plant height, and flowering time) and their hybrids (with Mo17 and B73 inbred lines, H1-H4) under both low- and high-N treatments (Fig. 5 and Supplementary material Fig. S1). The differences between networks were measured by using different parameters including the network size (total number of nodes, N), network connectivity (total number of links, E), modularity (M), and clustering coefficient (the extent to which nodes are clustered, CC).

Fig. 5
figure 5

Changes in microbial network topological properties across plant groups and N treatments. Boxplots illustrate topological properties of microbial networks associated with subgroups of non-stiff stalk (NSS) maize inbred lines and their hybrid rhizospheres under low- and high-N treatments. Network topological properties include (a) no of nodes (N), (b) network connectivity (total number of links, E), (c) modularity (M), and (d) clustering coefficient (the extent to which nodes are clustered, CC). Note: Microbial networks representing weak and nonsignificant interactions with a correlation < 0.3 and p > 0.05 were excluded from the analysis. * p < 0.05 and ** p < 0.01

Full size image

Microbes in the rhizosphere of non-stiff stalk inbreds had a significantly lower number of nodes (N: 173 and 159) compared to their hybrids (N: 206.5 and 196.5) under both low-N and high-N conditions (p < 0.05). Moreover, stiff stalk inbreds under high-N conditions had significantly fewer nodes than their hybrids under low-N conditions (p < 0.05). We also found that hybrids had a relatively higher number (p = 0.05) of edges (E: 1361.5) compared to inbreds under low-N conditions (E: 811.75).

Among other network parameters, the clustering coefficient was significantly higher for inbreds compared to hybrids under high-N conditions, a trend observed in both stiff and non-stiff stalk maize (Fig. 5 and Figure S1). Additionally, stiff stalk inbreds had a significantly higher clustering coefficient (CC: 0.40) under high-N conditions than hybrids under low-N conditions (CC: 0.29). Finally, different genotype groups (inbred and hybrid) and N treatments had minimal impact on network modularity for both stiff and non-stiff stalk maize.

Discussion

By leveraging large-scale maize field experiments, we measured the root rhizosphere microbial community responses to N treatment and assessed how the responses of these communities vary depending on the host which differed between inbred and their hybrids. The focus of this study was on the rhizosphere because it is a hotspot for diverse microbial activity and biogeochemical cycling (Zhang et al. 2019; Meier et al. 2022; Chai et al. 2023). We found a strong influence of both plant genetic factors and environmental filtering in shaping rhizosphere microbial community structure, variation in microbial relative abundance associated with the host, and network interactions. Our results contribute new insights into how host plant-microbial interactions are shaped across the rhizosphere of diverse maize lines in nutrient-deficient and fertilized soils.

The N input significantly increased the diversity of the rhizosphere microbial communities in inbred but decreased it in hybrid maize. This observation was true for all the different subgroups of inbreds lines (classified based on stalk stiffness, plant height and flowering time) and their hybrids. There is evidence suggesting contrasting influence of N fertilization on soil microbial diversity across varying field conditions (Zhang et al. 2018; Li et al. 2020; Mukhtar et al. 2023) including cropping systems (Kong et al. 2023). Our results support these findings and further suggest that the magnitude of N impact on microbial diversity differs between inbreds and their hybrid rhizospheres in agroecosystems. In this study, N treatment increased the average alpha diversity by 1.2 times in the rhizosphere of inbreds while decreasing alpha diversity by more than twofold in hybrids (Fig. 1). This underscore the significant role of host genotype (Chai et al. 2023) resulting in maize heterosis and overall plant vigor that impacts the composition and dynamics of the microbial community (Picard and Bosco 2005; Brisson et al. 2019; Wagner et al. 2020). As shown previously, hybrids exhibit differences in root traits and exudate production compared to inbred lines, and generally support more auxin and antibiotic producing rhizobacteria and genetically diverse Pseudomonas populations (Picard and Bosco 2005; Picard et al. 2008; Schmidt et al. 2016). Along with differences in genotype between plant groups, N application may also alter the maize root exudation production (Zhu et al. 2016), further influencing the rhizosphere microbial diversity. Consequently, soil microbes in the rhizosphere of hybrid lines under high-N treatment may undergo enhanced environmental filtering (compared to inbred lines), supporting higher relative abundance of a specific microbial taxa, such as Pseudomonas (Fig. 1). Taken together, the rhizosphere of hybrids appears to be exhibit heightened selectivity for microbial taxa compared to inbreds especially in N fertilized soils accentuating the compound impact of host plant genetic factors and edaphic conditions in shaping rhizosphere microbial community structure.

The number of plant genotype-associated microbial taxa (HT) under low-N was significantly higher than under high-N treatments, with inbred rhizospheres supporting a greater number of HT than hybrid rhizospheres. In fact,  HT (across the samples) under low-N were almost double (HT = 38) than those observed under high-N (HT = 21), supporting the hypothesis that key microbial taxa selected by the host may be a benefit to plant adaptation under nutrient-limited conditions (Zhang et al. 2019; Meier et al. 2022; Chai et al. 2023). Moreover, the proportion of variance in microbial relative abundance associated with plant genotype (refer as VPM) in this study (0.14 to 0.41) were within the same range as heritability values determined in other studies on rhizosphere communities of maize regardless of different environmental conditions (background variations) and the relative abundance of different microbial groups (Walters et al. 2018; Deng et al. 2021; Meier et al. 2022). In our study, the maize rhizosphere of inbred and hybrid was dominated by Pseudomonas which could be linked to the timing of our sampling, conducted at 8 weeks after planting. For instance, a bloom of Pseudomonas was observed in the maize rhizosphere between weeks 7 and 8 of development (Walters et al. 2018), which may not always occur in similar development stages (i.e., seven weeks after planting) (Wagner et al. 2020). Interestingly, the relative abundance of Pseudomonas, on average, was 1.3 (under low-N) to 2.1 times (under high-N) higher in the rhizosphere of hybrid as compared to inbred lines (Fig. 1). This may explain the fewer heritable taxa in hybrids compared to inbreds under both low- and high-N treatments. Despite the significant differences in the relative abundance of more than half of the microbial groups at the genus level between low-N and high-N, we observed minimal difference in VPM for the rhizosphere microbial communities from hybrids at phylum level across different N treatments. This implies that a restricted subset of microbial groups at lower taxonomic levels are responsive to plant genetic factors underscoring the importance of taxonomic resolution when analyzing microbial heritability.

The rhizosphere microbial networks showed a significantly higher number of nodes in hybrids under low-N compared to inbreds under high-N. This is likely due to the fact that: 1) hybrids may experience intermediate or lower stress under different N levels, which is not surprising given their higher overall vigor and greater adaptability than inbreds (Ruiz et al. 2019), and 2) the high selectivity for specific microbial taxa in the rhizosphere of hybrids (e.g., Pseudomonas) (Fig. 1). On the other hand, the clustering coefficient was higher in the microbial communities in the rhizosphere of inbreds as compared to hybrids only under high-N conditions suggesting changes in the cooperative interactions in the rhizosphere microbial communities in response to changing N availability. Taken together, microbial network topology across inbred and hybrid rhizospheres highlights a fundamental caveat in their functioning which depends on the variation in species response to soil N conditions. These intraspecies differences in maize should be considered when developing effective plant-specific soil inoculants and biofertilizers, especially under nutrient-deficient soils for low-input agriculture.

Conclusions

This study focused on the rhizosphere microbial communities, a region of the soil in agroecosystems that is a hotspot for taxonomical and functional diverse microbes, influenced by both soil conditions and plant root functional characteristics. We showed that the response of the microbial community structure and network topology to N application were based on plant genetic architecture. This may have also been influenced by sampling time with a proliferation of Pseudomonas taxa in the rhizosphere. Therefore, ignoring the influence of plant genotype on soil rhizosphere microbiota may result in either underestimation or overestimation of microbial responses to N availability and thereby agroecosystem functioning. However, it remains to be confirmed whether microbial responses to N levels in the rhizosphere of maize inbreds and hybrids will exhibit similar patterns under different climate and soil conditions, and how these patterns vary across different sampling times.