The experiment was conducted in a paddy field in Zhanjiang City, Guangdong Province, China. The paddy field had been used for rice production for several years using chemical fertilizers before the experiment. We aimed to understand the effect of organic fertilizer and biochar applications (soil improvements) on the soil cyanobacterial community in paddy fields, and how the resulting changes in soil property mediated the effect. The experiment was a 2 × 2 full factorial design with four treatments: Ctr (control, no organic fertilization or biochar addition), +OF (organic fertilizer application only), +BC (biochar application only), and +Mix (organic fertilizer and biochar applications). The four treatments, each with three replicates, were randomly assigned to 12 rectangular plots (7 m × 8 m), separated by 40 cm wide ditches.

The quantity and frequency of organic fertilizer and biochar applications followed local conventional approachs, which involved application once every 3 years at a rate of 0.25 tons of biochar/0.5 tons organic fertilizer per 667 m² each time. Specifically, organic fertilizer and biochar were applied by rotary tillage into 0–20 cm soil depth before rice planting in March 2018. Since then, the field was used for cultivating rice, with two crops planted annually. During the crop growth, all treatments received chemical fertilizers at regular doses and frequencies. The organic fertilizer was derived from fermented cow manure, while the biochar was pyrolyzed from discarded eucalyptus tree branches and leaves at temperatures between 550°C and 600°C. The basic characteristics of these two materials are shown in Supplementary Table S1.

Soil sampling was conducted in June when water in paddy fields receded during the summer of 2019. In each plot, five soil cores (4.5 cm in diameter) were taken at a depth of 0–10 cm and mixed to form a pooled sample. The soil samples were stored in airtight polypropylene bags and placed in a cool box at 4°C during transportation to the laboratory. After the removal of visible litter, roots, worms, and other debris, the soils were sieved through a 2-mm mesh. Each sample was divided into three subsamples. Subsamples for determining the cyanobacterial community composition were stored at −80°C until cyanobacterial gene extraction. Subsamples for the analysis of dissolved organic carbon (DOC), dissolved nitrogen (DN), ammonium N, and nitrate N content were stored at 4°C and analyzed within 48 h. Subsamples used to measure soil pH, soil organic matter (SOM), soil alkali-hydrolysable N (N_alk), extractable P (P_extra), extractable K (K_extra), total soil N, P, K, Ca, and Mg content were air dried.

Soil pH was determined in a 1:2.5 soil: water slurry using a glass pH electrode (FiveGO™, METTLER TOLEDO, Zurich, Switzerland). SOM was determined using H₂SO₄-K₂Cr₂O₇ oxidation method. Total soil N was quantified by the Kjeldahl acid digestion method. Total soil P was quantified using the molybdate blue method after acid digestion. To measure total soil K, Ca and Mg, air-dried soil were acid-digested. The solution after the digestion was analyzed following the method described in Liu et al. (1996). NH₃ produced from 2 g air-dried soil mixed with 7 mL 1 M NaOH solution for 24 h was measured to determine the soil alkali-hydrolyzable N content. Soil N_extra was the sum of ammonium-N and nitrate-N. Soil ammonium-N and nitrate-N were extracted with 10 g fresh soil by 100 mL 2 M KCl for 30 min within 48 h after soil samplings. Soil P_extra was extracted with 5 g air-dried soil via the Bray 1 method (0.03 M NH₄F and 0.025 M HCl, 50 mL) for 5 min. Soil K_extra was extracted with 2.0 g air-dried soil in 100 mL 1 M CH₃COONH₄ solution for 5 min. Dissolved soil organic C and dissolved soil N were extracted in K₂SO₄ solution and analyzed using a TOC analyzer (TOC-VCSH, Shimadzu GmbH, Duisburg, Germany).

Soil cyanobacterial community genomic DNA extraction was performed using a E. Z.N.A. Soil DNA Kit (Omega, M5635-02, United States). The concentration of the DNA was measured using a Qubit 4.0 (Thermo, United States) to ensure that adequate amounts of high-quality genomic DNA had been extracted. Trimmomatic was used for filtering and evaluating the quality of sequenced data. Raw reads were filtered following these steps: 1) Removing adaptor sequence; 2) Removing low quality bases from reads 3′ to 5’ (Q < 20), using a sliding window method to remove the base value less than 20 of reads tail (window size is 4 bp); 3) Finding overlap of each pair of reads and properly correct inconsistent bases within the interval; 4) Removing reads with length less than 35 nt and its pairing reads. And the remaining clean data was used for further analysis. MEGAHIT was used to perform multi-sample mixed splicing to obtain preliminary spliced contig sequences. Clean reads were mapped back to the spliced results, unmapped reads were extracted, and spliced again using SPAdes to obtain low-abundance contigs. MetaWRAP was used to perform a series of binning, and bin identification are performed in sequence. After filtering, a draft genome of a single bacteria with high integrity and low contamination is obtained. We used DIAMOND to compare the gene set with NR databases to obtain species annotation information of genes. Screening conditions: e-value <1×e⁻⁵, Score >60. Based on gene set abundance information and annotation information, species abundance are obtained.

Structural equation model (SEM) analyses were carried out in Amos 21.0 (IBM SPSS Statistics). Other statistical analyses were conducted in R version 4.1.2 (R Core Team, 2020). Shannon’s diversity index (H) was calculated as diversity index using the function diversity () in R package “vegan” (Oksanen et al., 2018). We used two-way ANOVA to examine the effects of organic fertilizer and biochar applications on soil cyanobacterial abundance (gene copy number), species richness (observed OTUs number), Shannon’s H, gene copy number of each genus and soil property. To compare these characteristics among each treatment, one way ANOVA was carried out followed by Tukey’s HSD. Multivariate effects of soil improvement on soil cyanobacterial community structure were illustrated by plotting ordination structure of a canonical correlation analysis (CCA) dependent on organic fertilizer and biochar applications. Environmental factors were plotted on the ordination using envfit () function in “vegan”. Heatmaps of OTUs representing their abundances and relative abundance in each treatments were displayed use heatmap () in “pheatmap” package. Linear regression analyses were conducted to reveal the relationships between change rates due to soil improvements and gene copy number for each cyanobacterial genus. The change rates were the differences of gene copy number between the soil improvement treatments and Ctr treatment divided by gene copy number in the Ctr treatment. Linear regression analyses were also used to explore responses of soil cyanobacterial abundance, species richness and Shannon’s H to soil property changes. The best models including different explanatory variables number were obtained using the regsubsets () function in “leap” package. To avoid overfitting, models with more than 2 independent variables were not considered.

A one-way ANOVA indicated significant variations in soil cyanobacterial abundance (p = 0.004), species richness (p = 0.025), and Shannon’s H (p = 0.009) among Ctr, +OF, +BC, and +Mix (Figure 1). The soil cyanobacterial abundance in Ctr was the highest, significantly higher than those of soil improvement treatments (+OF, +BC, and +Mix) (all p < 0.017). In comparison to Ctr, both + OF (p = 0.031) and +BC (p = 0.048) treatments significantly reduced soil species richness. The + Mix treatment showed a tendency to decrease species (p = 0.070). Conversely, soil improvement practices led to an increase in Shannon’s H. The + OF treatment significantly elevated Shannon’s H (p = 0.006), the +BC exhibited a tendency to decrease it (p = 0.059), while the +Mix’s enhancement was not statistically significant (p = 0.327). A two-way ANOVA revealed significant interactive effects between organic fertilizer and biochar applications on soil cyanobacterial abundance (p = 0.005), species richness (p = 0.027), and Shannon’s H (p = 0.003) (Supplementary Table S2).

Canonical correlation analysis (CCA) indicated a significant interactive effect between organic fertilizer (OF) and biochar (BC) application (p = 0.035, Figure 2). The ordination plot of the CCA further illustrated that both organic fertilizer (p = 0.029) and biochar (p = 0.007) applications had a substantial impact on soil cyanobacterial structure. All soil improvement plots were distinct from Ctr plots, while + OF plots showed a tendency to differ from +BC and +Mix.

A heatmap featuring all Operational Taxonomic Units (OTUs) also indicated that the 12 plots could be broadly grouped into two major clusters, labeled as A and B (Supplementary Figure S1), aligning with the Canonical Correlation Analysis (CCA) results. Cluster A consisted of 3 Ctr plots, while cluster B encompassed soil improvement plots. In Figure 3, the abundance and relative abundance of the top 200 abundant cyanobacterial OTUs in each treatment were illustrated. Soil improvement led to a decline in the majority of OTUs (Figure 3A). Interestingly, the relative abundance of several of the most abundant OTUs in Ctr was higher than those in other treatments, whereas most other OTUs were generally more prevalent in soil improvement treatments (Figure 3B). These response patterns suggested that the most abundant OTUs experienced a more significant decrease compared to other OTUs due to soil improvements, contributing to the alterations in the soil cyanobacterial community. The heatmap analysis involving all OTUs indicated that, for those OTUs with very low abundance, there were no apparent differences in abundance or relative abundance between Ctr and soil improvement treatments. However, owing to their extremely low read numbers, they exerted minimal influence on the overall size of the community.

At the genus level, notable differences were observed among the four treatments, particularly between Ctr and soil improvement treatments. In Ctr plots, the six most abundant cyanobacterial genera, in descending order, were Leptolyngbya, Microcoleus, Phormidium, Nostoc, Calothrix and Scytonema. Conversely, in the soil improvement plots, the order shifted to Leptolyngbya, Microcoleus, Nostoc, Phormidium, Calothrix and Hassallia. Differences in the abundance of each cyanobacterial genus among the four treatments were examined through one-way ANOVA followed by Tukey’s HSD test (Table 1), and the results were consistent with the heatmap analyses. The relative abundance of Leptolyngbya and Phormidium, the two most abundant cyanobacterial genera, decreased due to soil improvements, while the majority of others increased. This finding aligns with the results obtained from the analysis of OTUs data, providing insights into specific changes at the cyanobacterial genus level. Linear regressions suggested that the more abundant a genus was, the more it decreased due to soil improvements (Supplementary Figure S2).

The applications of organic fertilizer and biochar exerted significant effects on soil properties. Organic fertilizer application resulted in notable changes in soil pH, TP, P_extra, K_extra, Ca, DN and N_extra (all p < 0.050) (Table 2). In comparison, biochar application demonstrated a more pronounced impact on soil properties than organic fertilization, significantly altering soil pH, SOM, TN, TP, TK, P_extra, K_extra, Ca, DN and N_extra (all p < 0.050). Furthermore, significant interactive effects between organic fertilizer and biochar application were observed concerning soil pH, P_extra, K_extra, Ca, DOC and N_extra (all p < 0.050). Detailed changes and the results of Tukey’s HSD can be found in Supplementary Table S3 in the Supplementary Material.

A model included only N_extra explained as high as 78% variation of soil cyanobacterial abundance (p < 0.001, Table 3). The model also suggested soil cyanobacterial abundance was negatively correlated with N_extra (p < 0.001). Soil cyanobacterial species richness was negatively affected by N_extra (R ² = 0.709, p < 0.001). Incorporating N_alk into the model slightly improve the model’s r ² (R ² = 0.841, p < 0.001). N_extra explained 52% variation of Shannon’s H (p = 0.005), while pH and TK together explained 79% (p < 0.001). Shannon’s H was positively correlated with N_extra (p = 0.005) and pH (p < 0.001), but negatively correlated with TK (p = 0.018). The present study found that N_extra, TK, pH, Ca, TN, MBC, K_extra, and Mg were significantly correlated with soil cyanobacterial community structure (all p < 0.016) (Figure 2), and N_extra had the highest R ² .

Our structural equation models (SEM) revealed how organic fertilizer and biochar applications affected soil cyanobacterial abundance, species number and Shannon’s H (Figure 4) through influencing soil physicochemical characteristic. Both organic fertilizer and biochar applications affect cyanobacterial abundance through its effects on soil pH and the consequent N_extra changes (p = 0.468, χ² = 3.569, RMSEA <0.001) (Figure 4A). The model indicated N_extra was strongly and negatively impacted cyanobacterial abundance (r = −0.90, p < 0.001). The change of N_extra could be directly influenced by organic fertilizer (r = 0.27, p = 0.055) and biochar (r = −0.30, p = 0.081) applications. However, N_extra was more closely correlalated with soil pH (r = 0.92, p < 0.001) which was postively affected by both organic fertilizer (r = 0.39, p = 0.043) and biochar (r = 0.66, p < 0.001) applications. The SEM for soil cyanobacterial species richness revealed a mechanism illustrating the impact of soil improvements on it, similar to soil cyanobacterial abundance (p = 0.497, χ² = 3.375, RMSEA <0.001) (Figure 4B). In contrast, organic fertilizer and biochar applications influenced soil cyanobacterial Shannon’s H diveristy mostly through soil pH and K (p = 0.878, χ² = 2.417, RMSEA <0.001) (Figure 4C). The model suggested organic fertilizer mostly influenced cyanobacterial diveristy through increasing soil pH (r = 0.81, p = 0.009). Interestingly, the application of biochar led to an increase in soil pH, thereby enhancing cyanobacterial diversity, but, it also improved soil K availability (r = 0.86, p < 0.001), which in turn suppressed cyanobacterial diversity (r = −0.73, p < 0.001). Although soil N_extra was contained in the final model, it was not the principle factor mediating the effects of organic fertilizer and biochar applications (all p > 0.050).

Consistent with our hypothesis, the application of organic fertilizer and the addition of biochar, although exhibiting somewhat different effects, both positively contribute to the enhancement of soil nutrition. As shown in previous studies (Maltas et al., 2013; Ren H. et al., 2020; Viketoft et al., 2021), organic fertilizer excels in elevating nutrient levels, leading to a significant and rapid increase in extractable and soluble N, P, and K. In contrast, biochar has higher levels of C and K, but lower levels of N and P. Due to its slow decomposition, biochar made notable contributions to soil structure through its inactive structure (Sohi et al., 2010; Novair et al., 2023). As expected, compared to organic fertilizer application, biochar application had stronger effects on total soil C and K, as well as soil pH. On the other hand, organic fertilizer had a more pronounced impact on soluble soil nutrients, such as extractable NPK, DOC and DN. The distinct effects between organic fertilizer and biochar applications suggested combining biochar and organic fertilizer may offer a comprehensive approach, harnessing the nutrient-rich attributes of organic fertilizer and the structural enhancements provided by biochar. Our results also suggested that there were significant interactive effects between biochar and organic fertilizer applications on soil pH, extractable N, P and K, and other soil characteristics, indicating a combination of these two amendments could provide valuable insights into optimizing soil health and fertility. Therefore, exploring their synergistic potential emerges as a promising avenue for future research.

Although the effects on paddy soil property between the applications of organic fertilizer and biochar were different, both of them induced significant decreases in overall cyanobacterial abundance and species numbers. In paddy fields, the application of organic fertilizer (Lv et al., 2011; Ponge et al., 2013; Yang et al., 2019; Liu et al., 2021) and biochar (Lehmann et al., 2011; Liu et al., 2022) increased total microbial biomass, which were consistent with the findings of the present study. However, such studies treated the entire soil microbiota as a whole, ignoring the distinctions between cyanobacteria and other microorganisms. The present study revealed that cyanobacterial abundance’s response to organic fertilizer application was contrary to total soil microbial biomass. There were only a few studies studying the effects on soil cyanobacteria (Li et al., 2020). In line with the present study, most of them found organic fertilizer application depressed soil cyanobacterial abundance (Zhang et al., 2017; O'Brien et al., 2018; Li et al., 2020; Song W. et al., 2022). In a 5-year organic fertilization experiment in paddy soils, Song W. et al. (2022) suggested that replacement of replacing chemical fertilizers with organic fertilizers turned cyanobacteria to be non-dominant, suggesting the organic fertilizer appliaction inhibited the soil cyanobacteria. In a cabbage field, rhizosphere cyanobacterial abundance in synthetically fertilized soils also delined relative to control (O'Brien et al., 2018). As a result, we concluded that organic fertilizer application will decreased soil cyanobacteria. The impact of biochar application on soil cyanobacterial communities was seldomly studied and remained controversial (Kholssi et al., 2018; Cheng et al., 2019; Ren H. et al., 2020). Consistent with the present study, in rapeseed and maize fields in karst regions of southwestern China, long-term biochar addition decreased soil cyanobacterial abundance (Cheng et al., 2019). In contrast, in a eucalyptus seedling plantation in the neigbour province, biochar application increased the relative abundance of cyanobacteria (Ren H. et al., 2020). And a 20 days of incubation suggested the growth of filamentous cyanobacteria was increased on biochar solid supports (Kholssi et al., 2018). These contrasted results suggested that the impacts of biochar application on soil cyanobacteria may be dependent on specific ecosystem. In crop lands, as in our study, biochar application seemed to inhibit soil cyanobacteria, while in forests and lab incubation, biochar application tended to favour soil cyanobacteria.

While the majority of cyanobacterial species declined, the prominent cyanobacterial species, particularly the two most abundant genera, Leptolyngbya and Phormidium, exhibited more pronounced decreases. The responsiveness of soil microbial species to environmental changes can vary, influenced by the nature and extent of changes, as well as the specific characteristics of a microbial community and a ecosystem (Fierer and Jackson, 2006; Allison and Martiny, 2008; Rousk et al., 2010). It is important to acknowledge that microbial communities are highly diverse, and their responses to environmental changes are complex and context-dependent. Previous studies have yielded inconsistent results regarding whether dominant species are more sensitive to environmental changes. For examples, a field study in subtropical forest suggested the most dominant bacteria phyla were more responsive to N deposition (Dai et al., 2021). In contrary, in a maize filed, there were not obvious response difference between dominant and other microbes (Sun et al., 2019). The present study indicated that applications of organic fertilizer and biochar inhibited dominant soil cyanobacteria more than other species in paddy fields. However, the underlying mechanism remained unclear. Nevertheless, these divergent responses among cyanobacterial species should be considered in accounting for changes of the soil cyanobacterial community structure.

Soil available N and pH were the key soil property influencing soil cyanobacterial abundance, species richness and community structure due to the organic fertilizer and biochar applications.

Both organic fertilizer and biochar applications significantly afftected soil physicochemical property, potentially accounting for their effects on soil cyanobacterial community. In our experimental paddy field, the predominant factor influencing soil cyanobacterial community appeared to be the N availability. In terrestrial environments, oligotrophic conditions is more favorable for cyanobacterial growth. Soil cyanobacteria was usually the pioneer species in N-poor bare soil and weathered rock surface (Belnap, 2002; Housman et al., 2006). The copiotrophic hypothesis suggests that N addition reduces the relative abundance of oligotrophic taxa as relieved N limitation allows them to be outcompeted by more copiotrophic (Ramirez et al., 2010; Ramirez et al., 2012). As a consequence, the competitive advantage of N-fixing cyanobacteria is expected to wane with N supplementation. As the ecosystem undergoes succession, the dominance of cyanobacteria was gradually replaced by other species, usually coinciding with the increase in N availability within the ecosystem (Benavent-Gonzalez et al., 2019; Cano-Diaz et al., 2022). Notably, some previous studies suggested that increasing nutrient levels due to fertilizer application, particularly N availability, inhibited soil cyanobacteria (Sun et al., 2019; Sido, 2021), aligning with our findings.

Soil pH also strongly mediated the effects of organic fertilizer and biochar on soil cyanobacteria. Soil pH was commonly an important factor influencing soil microbial community in soil improvement experiments (Zhang et al., 2017; Li et al., 2020). In the present study, both organic fertilizer and biochar additions elevated soil pH to nearly neutral levels. Typically, aquatic cyanobacteria prefers neutral to alkaline pH conditions (Metting, 1981), and thus the increase in pH due to soil improvement should significantly and positively impact the cyanobacterial community. Consistent with our hypothesis, the increased soil pH directly enhanced soil cyanobacterial diveristy. Although soil pH did not directly impacted soil cyanobacterial abundance and species richness, it strongly influenced them through affecting soil N_extra. Surprisingly, the cyanobacterial abundance and richness in the paddy soil did not benefit from the increased soil pH. Conversely, the cyanobacteria were even less abundant in the soil improvement plots. Interestingly, a previous study suggested that cyanobacteria was more abundant in acidic soil (Zhang et al., 2017), indicating the cyanobacterial preference on neutral to alkaline pH may not be universal in soil ecosystem. In the present study, our SEMs suggested that the applications of organic fertilizer and biochar increased soil pH, thus increasing soil N_extra, which, in turn, inhibited soil cyanobacterial abundance and species richness.

Other soil properties might have only minor effects on cyanobacterial community. Cyanobacteria was sensitive to increased P availability both in aquatic and terrestrial ecosystems (Correll, 1998; Conley et al., 2009; Zou et al., 2023), and sometimes blooms with P enrichment, especially in aquatic environments (Reinl et al., 2021). As a result, the increased P availability induced by the soil improvements, especially the soil fertilizer application, was expected to increase soil cyanobacterial abundance. However, soil cyanobacterial abundance decreased with the organic fertilizer and biochar applications, and was not significantly correlated to soil P availability. The decreased soil cyanobacterial abundance and species number might be also attributed to altered specie interaction resulting from the soil improvements. Consistent with many other studies, the applications of organic fertilizer and biochar led to increases in total microbial biomass, predominantly comprising heterotrophic microorganisms (Werner and Dindal, 1990; Lehmann et al., 2011; Zhang et al., 2018; Viketoft et al., 2021). These heterotrophic microorganisms may compete with cyanobacteria for mineral nutrients or produce inhibitory substances on cyanobacteria (Wang et al., 2021), inhibiting soil cyanobacterial growth. On the other hand, other phototrophic microorganisms, such as green algae, tend to proliferate extensively with increasing N availability, covering the surface of soil and water, subsequently hindering cyanobacteria from obtaining light energy (Hu and Liu, 2003). Moreover, the heightened soil nutrient levels due to soil improvements could increase rice biomass, reducing the light penetration to soil surface. This reduction in light availability may also contribute to the decrease of soil cyanobacterial abundance, particularly those with photosynthetic capabilities (Rejmankova et al., 2008; Cerna et al., 2009; Santruckova et al., 2010).

The soil sampling in the present study was carried out only once, which limits the generalizability of the results. The response of soil cyanobacteria to soil improvements may be seasonally dependent. The temperature in the research area exhibited only a small seasonal variation, with the lowest temperature in January reaching approximately 5 degrees Celsius while the highest temperature reaches around 40 degrees Celsius in either December or January. In contrast, there was a significant alternation between wet and dry seasons in terms of precipitation, which may lead to pronounced seasonal differences in the response of cyanobacterial community to soil improvements. Studies on other soil microorganisms have indeed shown this possibility (Yu et al., 2022; Zhu et al., 2023). Since our samplings were conducted during the wet season, it should be noted that the response of cyanobacteria to soil improvements during the dry season may be different. Furthermore, the response of cyanobacterial community to soil improvements may change over years as the materials produce their effects and then gradually decline (Rinnan et al., 2007; Bell et al., 2014). Our samplings were conducted 2 years after the improvements, a time considered optimal for soil improvement effects in practical productions. Therefore, the impact of soil improvement measures on cyanobacteria may lessen if it is shorter or longer than 2 years. These factors represent limitations of our study, and we urge readers to consider them when interpreting the results of this study.