Predator-driven local convergence fosters global microbial community divergence

Predator-driven local convergence fosters global microbial community divergence | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Predator-driven local convergence fosters global microbial community divergence View ORCID Profile Rasit Asiloglu , Hayato Kuno , Mayu Fujino , Seda Bodur , Murat Aycan , Haruka Ishizuka , Shiori Kazama , Shinya Iwasaki , Jun Murase , Naoki Harada , Miwa Arai , Kenta Ikazaki doi: https://doi.org/10.1101/2025.08.15.670637 Rasit Asiloglu 1 Graduate School of Science and Technology, Niigata University , Niigata 950-2181, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Rasit Asiloglu For correspondence: asiloglu{at}agr.niigata-u.ac.jp Hayato Kuno 1 Graduate School of Science and Technology, Niigata University , Niigata 950-2181, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Mayu Fujino 1 Graduate School of Science and Technology, Niigata University , Niigata 950-2181, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Seda Bodur 1 Graduate School of Science and Technology, Niigata University , Niigata 950-2181, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Murat Aycan 1 Graduate School of Science and Technology, Niigata University , Niigata 950-2181, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Haruka Ishizuka 1 Graduate School of Science and Technology, Niigata University , Niigata 950-2181, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Shiori Kazama 1 Graduate School of Science and Technology, Niigata University , Niigata 950-2181, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Shinya Iwasaki 2 Crop, Livestock and Environment Division, Japan International Research Center for Agricultural Sciences (JIRCAS) , Ibaraki 305-8686, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Jun Murase 3 Graduate School of Bioagricultural Sciences, Nagoya University , Nagoya 464-8601, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Naoki Harada 1 Graduate School of Science and Technology, Niigata University , Niigata 950-2181, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Miwa Arai 2 Crop, Livestock and Environment Division, Japan International Research Center for Agricultural Sciences (JIRCAS) , Ibaraki 305-8686, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Kenta Ikazaki 2 Crop, Livestock and Environment Division, Japan International Research Center for Agricultural Sciences (JIRCAS) , Ibaraki 305-8686, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site Abstract Full Text Info/History Metrics Preview PDF Abstract Understanding the rules that govern microbial community assembly is essential for predicting ecosystem function. While microbial predators are key biotic agents that shape bacterial communities through predation, yet their ecological consequences have been studied mostly in isolated and in vitro systems 1 – 3 . In contrast, large-scale studies of microbial diversity have primarily emphasized abiotic factors as drivers of community assembly 4 – 7 , while the role of microbial predators modulating global microbial divergence and convergence patterns remains largely neglected. Here, we show that bacterivorous protists (predators) exert dual, scale-dependent effects on microbial communities: promoting local convergence by suppressing dominant bacterial taxa, while generating global divergence through species-specific predation effects. By integrating global meta-analyses, controlled field experiments, and reconstructions of natural and synthetic communities, we find that predator identity and prey susceptibility jointly determine convergence outcomes. Communities dominated by predator-resistant taxa exhibit reduced convergence under predation pressure, revealing a predictable trait-based filtering mechanism. This framework reconciles previous contradictory findings 3 , 8 – 11 and highlights predators as selective, context-dependent agents of microbial biogeography. Predator-driven convergence suggests new opportunities for microbiome engineering 12 : targeted use of predators may steer microbial communities toward functional configurations that enhance soil health, disease suppression, carbon cycling, and ecosystem resilience 2 , 13 , 14 . Main Understanding how microbial communities assemble is a central question in ecology, with broad implications for ecosystem management and microbiome engineering 12 . The interplay between biotic and abiotic variation at both local and global scales influences microbial divergence (increased dissimilarity among communities) and convergence (increased similarity among communities) patterns 15 – 17 , making it challenging to predict the trajectories of microbial communities. Recent studies revealed that abiotic factors such as pH, moisture, temperature, salinity, nutrient content, and nutrient complexity drive microbial divergence patterns 4 – 7 . While abiotic factors are important, myriad interactions among species drive microbial community composition against a background of environmental variability 1 , 15 , 18 . However, the role of biotic factors, particularly prey-predator interactions, in modulating divergence and convergence patterns remains largely neglected. Predatory protists (hereafter, predators) represent the primary and dominant consumers of bacteria and are one of the most crucial factors controlling bacterial community assembly through selective predation 1 , 19 . Predators target bacterial taxa based on bacterial traits or dominance, while several bacterial taxa can take advantage of predation through accelerated nutrient turnover 1 , 20 . Different predators drive bacterial communities in distinct directions through species-specific effects 21 – 23 . Nevertheless, the consequences of predation for microbial community assembly have been largely studied in isolated and in vitro systems 19 , 20 , and their influence on broader patterns such as community divergence and convergence remain poorly understood. Predators can either promote divergence or convergence in ecological communities depending on how they interact with prey species, the spatial and temporal context, and the nature of selective pressures they impose. Two contrasting hypotheses have been proposed to explain predators’ ecological impact ( Fig. 1a ). The divergence hypothesis posits that predators diverge communities by reducing the abundance and richness of the same bacterial taxa across ecosystems through selective-feeding 8 , 9 . In contrast, the convergence hypothesis suggests that predators converge communities by preventing dominance by a few species and allowing coexistence of a broader range of species via density-dependence feeding 3 , 10 , 11 . These hypotheses provide a framework for testing in both natural and experimental systems. Yet, despite their theoretical appeal, empirical evidence to establish which of these hypotheses explains how predators control community convergence and divergence across multiple scales and contexts is scarce. Download figure Open in new tab Figure 1 Context-dependent roles of predators in driving divergence and convergence at the global scale. a , Two contrasting hypotheses on predators’ effect on bacterial community assembly were illustrated under theoretical conditions. Stacked bar charts on top depict microbial community composition and community divergence (below) depict dissimilarities between two communities (A and B) under two scenarios. Size shows abundances, and colors corresponding to different bacterial taxa. The divergence hypothesis (left) posits that communities diverge due to predation on the same bacterial taxa (blue, orange, and green) across communities. The convergence hypothesis (right) posits that predators feed on dominant taxa in each community, leading to similar community composition across environments. b , Map showing the locations of 138 soil sampling sites (red dots) across six continents where both bacterial and protistan sequencing data were available in public databases 5, 24 . c , Variable importance derived from random forest permutation analysis identifying key environmental predictors of global-scale bacterial community divergence. Importance is expressed as the percent increase in mean squared error (%IncMSE) when each variable is permuted; higher values indicate stronger predictive influence. Predator traits are highlighted in bold. Bar colors indicate significance levels (red, P < 0.01; black, P < 0.05; light grey, not significant). d , Linear correlation between protist and bacterial community divergence based on Bray–Curtis distances. e , f , Linear correlations between the relative abundance of endemic ( e ) and ubiquitous ( f ) predators and bacterial community divergence. g , h , Linear correlations between the richness of dominant ( g ) and ubiquitous ( h ) bacterial taxa and the relative abundance of ubiquitous predators. Linear regression lines are shown with 95% confidence intervals (shaded). Pearson correlation coefficients ( P values) and coefficients of determination ( R² ) indicate the strength and significance of each association. Here, we investigated the role of predators in driving both divergence and convergence of soil bacterial communities, combining 1) global meta-analyses across six continents, 2) controlled field trials in a local island where community divergence was environmentally manipulated, 3) microcosm studies with natural communities with and without defined predator introductions, and 4) synthetic community (SynCom) reconstructions with selected bacterial taxa. Together, our integrative approach provided multi-scale evidence for the context-dependent role of predators based on predator identity and prey susceptibility in shaping bacterial community divergence and convergence. Dual role of predators in driving both divergence and convergence We integrated and re-analysed existing data from two studies, matching protist 24 and bacterial 5 sequence data in 138 locations across six continents ( Fig. 1b ), to understand how predators affect community divergence and convergence. Predators were extracted from publicly available amplicon sequence variant (ASV) table 24 (methods, Extended Data Fig. 1a-b). Both bacterial and predator communities and divergences showed similar distribution patterns across ecosystems and continents (Extended Data Fig. 1c-h). As predictors of predators, we used alpha (Shannon index, Evenness, and Richness) and beta (community dissimilarity) diversities. Since predatory effects are species-specific 1 , 21 , relative abundances (RA) of ubiquitous (commonly found ASVs across diverse ecosystems) and endemic (restricted ASVs to specific ecosystems) predators were also included. Abundance and divergence of predators were affected by environmental factors, especially pH, aridity, temperature, and soil carbon 24 , where their effect on endemic and ubiquitous predators followed opposite trends (Extended Data Fig. 2). Although we found a positive linear correlation between predator RA and bacterial divergence (Extended Data Fig. 3a), RA of each one of the top five predator ASVs that were commonly observed in most of the samples showed negative patterns (Extended Data Fig. 3b-f), indicating their impact on community convergence. The alpha diversity indexes of predators had no effect (Extended Data Fig. 3g-i). Random forest model that identifies the key predictors of bacterial community divergence showed that RA of both ubiquitous and endemic predators and the divergence of predators were among the top factors along with pH, aridity and temperature ( Fig. 1c ), showing global importance of predators for bacterial community assembly. While predator divergence and RA of endemic predators showed a positive linear relationship with bacterial divergence, RA of ubiquitous predators followed an opposite trend ( Fig. 1d-f ). These results suggest a dual role of predators: while the global divergence of bacterial communities may arise from differences in the composition of predator communities across ecosystems through likely their species-specific effects, ubiquitous predators drive bacterial community converge by exerting consistent effects across the globe 3 , 9 , 11 . Since our results support the convergence hypothesis—predators converge communities by suppressing dominant taxa—we further examined relationship between richness of dominant bacterial taxa and ubiquitous predators. As ubiquitous predators are likely to target widespread (ubiquitous) bacterial taxa, we also assessed the correlation between them. We focused on richness over relative abundance, as richness more directly captures the divergence and convergence patterns in bacterial communities 6 . We found that ubiquitous predators tend to enhance the richness of both ubiquitous and dominant bacterial taxa ( Fig. 1g-h ), suggesting that predation disrupts dominance by a few taxa, instead promote dominance by broader range of species. Taken together, we do not interpret predator-driven convergence as a universal effect. Rather, predators are likely to affect microbial communities in a species-specific manner, which may explain how convergence and divergence can coexist ( Fig. 1e-f ). This mechanism aligns with the Janzen–Connell hypothesis 25 , which posits that natural enemies reduce the dominance of abundant species, thereby promoting diversity through negative density dependence. In microbial communities, such top-down control by predators may function as a stabilizing mechanism of coexistence 26 , where predation suppresses competitive dominants and allows subordinate taxa to persist. This dynamic is conceptually analogous to clonal interference in microbial evolution 27 , in which multiple high-fitness genotypes compete for dominance, preventing any single lineage from sweeping to fixation. Similarly, predator-driven suppression of dominant bacterial taxa may allow multiple strong competitors to coexist, thereby increasing community richness among dominant taxa and compositional convergence. Local-scale bacterial community convergence by predators To gain deeper insight into predator-driven convergence in bacterial communities suggested by global observations, we conducted a local field trial under controlled ecological conditions on Ishigaki, a small tropical island (222 km 2 ) in Japan ( Fig. 2a ). The relatively low initial divergence of microbial communities in this setting allows the effects of similar predators on community convergence to be more clearly observed. We artificially introduced community divergence in three fields with eight different fertiliser treatments and a pH adjustment treatment by CaCO3 addition (Extended Data Table 1-2), which are known to strongly drive bacterial diversification ( Fig. 1c ) 5 , 6 , 17 , 28 . We hypothesised that in fields with relatively less divergent and high abundant predators, ubiquitous predators will counteract the pH- and nutrient-driven divergence by converging bacterial communities, while endemic predators would not have a divergence effect at the local scale. Therefore, we analysed microbial community divergence within each field. Download figure Open in new tab Figure 2 Local-scale bacterial community convergence driven by predators in sugarcane fields, Ishigaki Island, Japan. a , Map showing the locations of the three sugarcane fields (A, B, and C) used in the field trial. b , Variable importance derived from random forest permutation analysis identifying key environmental predictors of local-scale bacterial community divergence. Importance is expressed as %IncMSE. Predator traits are highlighted in bold. Bar colors indicate significance levels (red, P < 0.01; black, P < 0.05; light grey, not significant). c , Linear correlation between protist and bacterial community divergence based on Bray–Curtis distances. d , e , Linear correlations between the relative abundance of endemic ( d ) and ubiquitous ( e ) predators and bacterial community divergence. f , g , Linear correlations between the richness of dominant ( f ) and ubiquitous ( g ) bacterial taxa and the relative abundance of ubiquitous predators. Regression lines with 95% confidence intervals are shown. Pearson correlation coefficients ( P values) and R² indicate statistical strength and significance. Aligned with our expectations, bacterial divergence was generally increased with fertilizer addition (Extended Data Fig. 4a) and we observed differences in abundance and diversity of predators among treatments and fields (Extended Data Fig. 4b-c). Nonmetric multidimensional scaling (NMDS) analyses showed that predator community composition was relatively more similar to each other than those of bacteria among fields (Extended Data Fig 4d-e). RA of ubiquitous predators and divergence of predators were oppositely affected by environmental factors, especially pH, C, N, and C/N ratio, while endemic predators were not affected (Extended Data Fig. 5). To identify the key predictors of bacterial community divergence, we performed random forest analysis, showing RA of ubiquitous predators was among the top factors, along with C and pH ( Fig. 2b ). The divergence of predators was also detected as an important factor predicting bacterial community divergence ( Fig. 2b ). While the correlation of bacterial and predator divergences and RA of ubiquitous predators showed an opposite trend, RA of endemic predators showed no significant correlation ( Fig. 2c-e ), as expected. Consistent with the convergence hypothesis, we found that ubiquitous predators tend to enhance the richness of both ubiquitous and dominant bacterial taxa ( Fig. 2f-g ). Taken together, these results support our hypothesis that under local conditions where predator taxa are similar and abundant, they tend to converge bacterial communities or at least reduce the nutrient- and pH-driven divergence, while endemic predators have no divergence impact due to their low abundance at local conditions. Predators promote convergence by feeding on dominant bacterial taxa Since both global and local studies consistently showed that predators are likely to promote bacterial convergence primarily by enhancing richness of dominant and ubiquitous bacteria, we used structural equation modelling (SEM) to mechanistically evaluate these effects. By integrating the global and local datasets, we constructed a simple model with a reduced number of predictors based on our random forest results ( Figs. 1c and 2b ). SEM analysis demonstrated that bacterial community convergence was largely governed by the dominant bacterial richness that was driven by RA of ubiquitous predators and the community similarity of predators ( Fig. 3a ). Although these results provide strong support for the convergence hypothesis, they did not offer definitive causal evidence due to the background of environmental variability. To validate our findings ( Fig. 3b ), we conducted a complementary in vitro microcosm experiment under defined conditions. We first obtained indigenous predator-free bacterial communities from five distinct soil types (forest, Com1; grassland, Com2; paddy field, Com3; soybean field, Com4; sugarcane field, Com5). Then we incubated them in calcined clay as an inert soil substitute 29 , 30 under defined nutrient conditions for 5 weeks. Each bacterial community was grown with and without one of three different model predator species, applied separately. This design enabled us to directly assess the exclusive role of individual predator species in driving bacterial convergence, independent of environmental heterogeneity 21 . Download figure Open in new tab Figure 3 Predators promote convergence by feeding on dominant bacterial taxa. a , Structural equation model (SEM) showing the effects of protist traits on microbial community convergence mediated by dominant and ubiquitous bacterial taxa. Numbers adjacent to arrows represent standardized path coefficients; arrow width is proportional to coefficient magnitude. Green and red arrows indicate positive and negative relationships, respectively. Solid arrows denote significant, and dashed arrows non-significant, paths. R² values indicate explained variance. Significance levels: *** P < 0.001. b , Conceptual diagram comparing predator-absent (top) and predator-present (bottom) conditions, illustrating how predators influence bacterial community composition and dominant bacterial richness by preying on dominant taxa. Stacked bar charts show abundances, with colors corresponding to different bacterial taxa. c , Relative abundances of bacterial genera in control (Ctrl), and predator-present treatments: Ac ( Acanthamoeba castellanii ), Hg ( Heteromita globosa ), and Vv ( Vermamoeba vermiformis ). Community evenness is indicated above each bar; bold values denote significant differences ( P < 0.05, ANOVA) relative to Ctrl. d–g , Nonmetric multidimensional scaling (NMDS) of Bray–Curtis dissimilarities showing community shifts across treatments: d , All bacterial ASVs; e , predator-depleted ASVs; f , predator-enriched ASVs; g , neutral ASVs unaffected by predators. Points represent communities, with different shapes for each. h , k , Richness of dominant ( h ) and rare ( k ) ASVs. l , m , Relative abundance of dominant ( l ) and rare ( m ) ASVs in the control and predator treatments. Box plots show medians (central line), interquartile ranges (hinges), and range (whiskers). Asterisks indicate significant differences ( P < 0.05) compared to the control. Bacterial communities consisted of a total of 2121 ASVs with an average richness of 88.9±25.1, indicating successful colonisation of diverse and rich communities. Results after 3 and 5 weeks showed highly similar communities in each treatment. This might be due to most communities stabilising after 60 generations, reaching stable population equilibria 6 , 17 . The evenness of communities increased in the presence of predators ( Fig. 3c ), and predator treatments, especially with Acanthamoeba castellanii (Ac) and Vermamoeba vermiformis (Vv), showed more uniform distribution of bacterial taxa at the genus level ( Fig. 3c ). We found that the five communities were grouped closer to each other in predator treatments, while the control group with no predator addition showed clear separation from each other ( Fig. 3d ). This pattern was more strongly observed in NMDS analysis based on predator-enriched and predator-depleted ASVs, while predator-neutral ASVs grouped based on source communities ( Fig. 3e-g ). Consistent with NMDS visualisation, the divergence of bacterial communities was higher in the control group than that of predator treatments, including predator-enriched and predator-depleted ASVs, while no difference in bacterial community divergence was observed for predator-neutral ASVs (Extended Data Fig. 6a-d). These results evidently validate the predator-driven convergence: predator-affected ASVs caused the similarities of bacterial communities. Although NMDS revealed broadly similar distributions among the three predator treatments, Mantel tests indicated that each predator generated distinct bacterial communities ( P < 0.01). Indeed, NMDS analyses of each community (Extended Data Fig. 7) showed that Heteromita globosa (Hg), a phylogenetically distinct predator from the supergroup Rhizaria, drove bacterial communities in a different compositional direction compared to the amoeba treatments from the supergroup Amoebozoa (Ac and Vv). This supports taxonomy-based, species-specific convergence effects that ultimately lead to global divergence, consistent with our expectations and previous findings 20 , 21 . Similar with field observations, predators increased the richness of dominant bacterial taxa ( Fig. 3h ), while they did not necessarily increase the richness of rare bacterial taxa ( Fig. 3i ). To double-check the effect of predators on dominant taxa, we further showed that the RA of dominant ASVs in the control group decreased in the predator treatments ( Fig. 3j ), while RA of rare bacterial ASVs in the control group increased in the presence of predators, leading to an increase in abundant bacterial richness attributed to predators ( Fig. 3k ). Taken together, in line with global and local field observations, we showed evidence for the convergence hypothesis: predators converge bacterial communities through consistently reducing dominant bacterial taxa and increase their richness. Revisiting the divergence hypothesis ( Fig. 1a ), we found over 99% of the ASVs that affected by predators were unique to each community (Extended Data Fig. 8), which was consistent with previous findings 21 . This suggests that selective predation is not uniformly directed at the same bacterial taxa. Instead, predation effects appear to be highly context-dependent, shaped by initial community composition, which disables taxonomic-based predictions 21 , 31 . Although our data do not support widespread targeting of the same bacterial taxa across ecosystems, we do not reject the possibility that certain protists may feed on similar prey under specific conditions 11 , 32 . Despite this nuance, the convergence hypothesis provides a more compelling explanation for the observed community divergence and convergence patterns in large-scale experiments across varied environmental conditions. Convergence depends on the predation resistance of dominant taxa Several bacterial taxa have evolved to successfully escape from predation 33 , which may hinder the predator-driven convergence. Therefore, we hypothesised that communities dominated by predator-resistant bacteria should have less convergence under predation. We constructed seven SynComs, consisting of six bacterial species (three predator- resistant and three predator-vulnerable species based on the results of microcosm experiment). The first SynCom received equal abundances of all six bacterial species, while the other six each had one dominant species and five rare ones ( Fig. 4a ). The SynComs were then grown with and without the three predators, separately. While communities without predators showed distinct patterns in each SynCom group, predator-resistant bacterial taxa dominated in the presence of predators ( Fig. 4a ). NMDS grouped predator treatments together, while the control group showed higher dissimilarities ( Fig. 4b ). Indeed, community divergence in the control group was higher than that of predators ( Fig. 4c ). Crucially, the community divergence was lower when initially dominated by predator-resistant bacteria ( Fig. 4d ), supporting our hypothesis. Our findings suggest that the extent of predator-driven convergence is shaped by the traits of dominant bacterial taxa and that resistance to predation plays a key role in determining the trajectory of microbial community assembly under predator pressure. Download figure Open in new tab Figure 4 Convergence depends on the predation resistance of dominant taxa. a , Relative abundances of bacterial genera in initial, control (Ctrl), and predator treatments (Ac: Acanthamoeba castellanii ; Hg: Heteromita globosa ; Vv: Vermamoeba vermiformis ). Predator-resistant taxa are shown in green tones. SC, synthetic community. Pk, Pseudomonas kilonensis ; Ad, Achromobacter denitrificans ; Pm, Priestia megaterium (formerly known as Bacillus megaterium ; Lg, Luteimonas galliterrae ; Ks, Knoellia sinensis ; Tt, Telluria terrae (formerly known as Masilla terrae ). b , NMDS ordination (Bray–Curtis dissimilarity) of log-transformed ASV tables showing compositional shifts across treatments. Colors indicate protist treatments; shapes indicate synthetic communities. c , d , Box plots of within-group variance in community divergence across predator treatments ( c ) and synthetic communities ( d ). Box plots show medians (central line), interquartile ranges (hinges), and range (whiskers). Context-dependent predator-driven convergence is mediated by both predator identity and prey characteristics, suggesting an evolutionary layer to microbial community responses. The effects of predators on bacterial diversity are crucial for functional redundancy and ecosystem resilience 1 , 19 . Although our results indicated that predator-driven convergence occurs mainly due to their predation on dominant taxa, the role of resource redistribution or increased nutrient turnover following predation likely contributes to predator-driven convergence. Our previous study showed that while predator-preferred bacterial taxa overlap to some extent among distinct predator species, predators assert a stronger species-specific effect on predator-enhanced bacterial taxa 21 . This suggests that convergence also arises indirectly through trophic cascades or facilitative interactions, wherein predator activity releases nutrients or alters the competitive environment in ways that favour opportunistic bacterial taxa 20 . These context-dependent interactions could result in selective growth of functionally similar bacteria across communities, further reinforcing convergence at the compositional or functional level. For instance, increased predation has been shown to drive functional convergence among bacterial taxa exhibiting traits such as antibiotic resistance 14 , 34 , 35 , exopolysaccharide production 1 , 36 , and pathogen suppressiveness 37 . Here we predicted taxonomic convergence, and future work incorporating functional metagenomics or isotope tracing may help disentangle these intertwined effects and clarify the extent to which predators drive functional convergence. Net convergence effect of predators Despite the strength of our results, the interplay between abiotic and biotic drivers also remains complex. The predator diversity and abundance are controlled by climatic variables (aridity and temperature), and to some extent, by soil characteristics (pH and total C), making the generalisability of predatory effects across ecosystems remain uncertain, with likely varying magnitudes across ecosystem types 24 . Moreover, extreme abiotic divergence, where environmental filtering outweighs biotic interactions, may diminish the influence of predators. For instance, as most predators are aquatic organisms restricted to the water-filled pores in the soil ecosystem 38 , their effects on bacterial assembly likely to be reduced under dry conditions 24 , which was evidenced by their low relative abundances in dry ecosystems (Extended Data Fig. 1a). In addition, environmental factors, especially nutrients, shape the predation patterns on bacteria 21 , 31 , 39 . As our in vitro experiments were conducted under defined nutrient conditions, the degree of predator- driven convergence and the outcome of bacterial assembly may vary under different circumstances 21 . Despite this nuance, our model ( Fig. 5a ), which quantifying the net convergence effect of predatory protists shows a highly predictive linear relationship with bacterial convergence across our global and local datasets (R² = 0.90), effectively capturing the dual role of predators. Notably, this relationship persists despite extreme abiotic gradients in climate, nutrient availability, and pH 5 , 24 . The robustness of this correlation underscores the significant role of predators in driving bacterial community assembly, likely operating in synergy with abiotic environmental filters at multiple scales. Download figure Open in new tab Figure 5 Predator-driven bacterial convergence. a, linear correlation between net convergence effect of predators and bacterial convergence. Net convergence effect of predators was calculated by subscribing the divergence effect of predators (relative abundance of the endemic predators × predator community dissimilarity) from convergence effect (relative abundance of the ubiquitous predators × predator community dissimilarity -1 ) of predators. Linear regression lines are shown with 95% confidence intervals (shaded). Pearson correlation coefficients ( P values) and coefficients of determination ( R² ) indicate the strength and significance of each association. b , Conceptual framework of context-dependent protist-driven convergence. Predator communities (A–D) and bacterial communities (1–3) are represented using color-coded bar plots. Predator colors indicate distinct taxonomic identities, while bacterial colors represent predation vulnerability: red tones for vulnerable taxa and green for resistant taxa. Dashed grey arrows denote the influence of predators on bacterial communities. When bacterial communities are dominated by predator- vulnerable taxa and encounter predator communities with similar taxonomic composition (e.g., A and B, or C and D), the bacterial communities shift in similar directions—this pattern is defined as predator-driven convergence (PDC). When different predator taxa dominate predator communities, their PDC effect drive bacterial communities in distinct directions through species-specific effects, resulting in predator-driven divergence (PDD). When bacterial communities are dominated by predator-resistant taxa, they remain compositionally stable regardless of predator identity—demonstrating community resistance to predation (CRP). Conclusion Microbial predators have long been recognized as key biotic agents in food webs, yet their role in shaping microbial community assembly across scales has remained largely overlooked. Here, we integrate global meta-analyses, field experiments, controlled microcosms, and synthetic communities to show that predatory protists simultaneously drive local convergence and global divergence of bacterial communities. This dual effect arises from a consistent mechanism: predators suppress dominant bacterial taxa, promoting compositional similarity in local settings, while species-specific predation effects lead to divergent outcomes across ecosystems, which were affected by bacterial traits ( Fig. 5b ). Our findings reconcile previously conflicting models of microbial community assembly 3 , 8 – 11 and provide empirical support for a refined convergence hypothesis: predators act as context-dependent biotic filters that mediate microbial diversity through trait-based interactions. This scale-dependent framework highlights predation as a fundamental driver of microbial biogeography and diversity patterns. The predictable nature of predator-driven convergence under defined conditions opens new avenues for microbiome engineering, where targeted use of predator species may help steer microbial communities to enhance functions such as nutrient cycling, disease suppression, and carbon storage 2 , 13 , 14 Author Contributions RA conceived and designed the study, performed bioinformatics and statistical analyses, interpreted the data, planned the direction and prepared the manuscript. IK and M Arai were in charge of local field trials. RA, HK, MF, M Aycan, SI, SB, HI, SK, M Arai, and KI performed experiments. HK, MF, M Aycan, JM, NH, SI, M Arai, and IK contributed to interpretation and presentation of the results. All authors provided feedback and suggestions on the manuscript and approved the final manuscript. Data availability The raw sequence data obtained in this study have been deposited in the NCBI database under the BioProject ID PRJNA1290915. Code availability The codes are available upon request to RA. Competing interests The authors declare no competing interests. Correspondence and requests for materials should be addressed to RA. Methods Global meta-data We integrated and re-analysed existing data from two studies that analysed global bacterial 5 and protist 24 sequences. We first matched soil samples from protist and bacterial data by using exact longitude and latitude information, allowing us to match 138 samples across six continents (source data). The bacterial amplicon sequences were downloaded 5 and re-analysed using QIIME2 40 . DADA2 41 in QIIME2 was used for error correction, removal of forward and reverse primers, quality filtering, doubleton removal, and chimera removal of the amplicon sequences, corresponding to a quality score > 30. We denoised the paired-end sequences into amplicon sequence variants (ASVs). QIIME2’s q2-feature-classifier plugin was used for taxonomy assignment against the latest SILVA reference database 42 . We obtained 27,148 ASVs, which is comparable with the original study 5 . For protists, we downloaded available ASV table with taxonomic assignment 24 . Then functional group of bacterivorous protists (predators) was manually assigned and extracted as previously shown 24 , 43 , 44 . The relative abundance (RA) of predators used in all steps of our study represented their abundance distribution among all protists, which was comparable with the previous study 24 . Environmental variables were obtained from the bacterial metadata available online 5 . The remaining bioinformatics and all statistical analyses were conducted in R version 4.5.0 (2025.4.11; https://www.r-project.org/ ). To visualize the geographic distribution of global sampling sites, we generated a world map using the ggplot2 package. A base map was created using the map_data function. Sampling locations were overlaid based on their corresponding longitude and latitude coordinates from the metadata. Community divergence was calculated as previously shown 6 . Briefly, beta diversities of bacteria and predators was calculated using Bray–Curtis dissimilarity matrices and quantified with the betadisper function from the Phyloseq and Vegan packages in R. We then calculated mean divergence by averaging Bray–Curtis distances between samples. Bray–Curtis dissimilarity is a widely used ecological metric that captures compositional differences based on relative abundance. While alternative approaches such as Aitchison distance, Jaccard index, or UniFrac are also commonly used, Bray–Curtis was selected for its ecological interpretability in community ecology. Specifically, Bray–Curtis emphasizes differences in abundant taxa—which are typically the most ecologically influential in shaping ecosystem function—while down-weighting rare taxa that may introduce noise in divergence estimates. This focus aligns with studying prey-predator interactions. In contrast, Aitchison distance is designed for compositional data and compares taxa based on log-ratios of their relative abundances. While this method is appropriate for analyzing proportional relationships, it may overemphasize low-abundance taxa due to the nature of the log-ratio transformation, especially when rare taxa exhibit high variability as in our global dataset. Thus, Bray–Curtis provides a robust and ecologically meaningful measure of community structure change in response to predation pressure. Alpha diversity was computed in QIIME2 using diversity metrics of the richness (observed ASVs), Evenness and Shannon index. To visualize beta diversity of bacterial and protist communities, we performed NMDS based on Bray– Curtis dissimilarity using relative abundance data. We employed a random forest permutation using rfPermute package in R with 1,000 trees and 1,000 permutations to identify environmental and biotic predictors of bacterial community divergence or environmental predictors of protist divergence and RA. Importance scores and associated permutation-based p -values were extracted and exported for interpretation. To visualize the linear correlation relationships, we generated a scatter plot using ggplot2, displaying a linear regression fit with 95% confidence interval between factors. Pearson’s correlation coefficient ( P -value) and the coefficient of determination ( R² ) were annotated on the plot. To identify ubiquitous ASVs, we calculated their prevalence across all samples. ASVs present in more than 60% of samples were classified as ubiquitous, while ASVs present in fewer than 20% of samples were classified as endemic, representing habitat-specific or spatially restricted taxa. Since only two bacterial ASVs were present in more than 60% of samples, we lowered the ubiquity threshold to 40% for bacteria. ASVs with mean relative abundance >0.1% across all samples were classified as dominant bacterial taxa, while those with mean relative abundance <0.01% were considered rare. Field Trial We conducted a local field trial in sugarcane fields on a small tropical island, Ishigaki, Japan due to its relatively stable agricultural management, distinct climatic conditions, and ecological isolation. These features minimized confounding effects from long-range dispersal and heterogeneous land use, allowing us to more clearly assess predator-driven microbial community dynamics under partly defined environmental conditions. Our experiment site consisted of three fields and nine treatments in each field (Extended data Table 1 and 2, and source data). All treatments received chemical fertilizer addition. While Treatments 1, 7-9 received conventional chemical fertilisers (Extended data Table 1) and the other treatments (T2-6) received reduced (70%) chemical fertilizers. Type of fertilizers included filter cake (a nutrient-rich organic byproduct of the sugarcane industry, specifically from the clarification and filtration of sugarcane juice during sugar processing), cattle manure, bagasse (a fibrous residue left after sugarcane stalks are crushed to extract their juice during sugar production), and wood biochar (Extended data Table 1-2). Filter cake, cattle manure, bagasse, and biochar were applied at a rate of 10 tons of carbon per hectare (T4-7 and 9). To create a soil carbon gradient, filter cake and cattle manure were also applied at 5 tons of carbon per hectare (T2-3). For pH adjustment, CaCo3 was applied (10.7 Mg/ha). Fertilizer types and amounts were selected based on their potential to influence bacterial diversification by increasing soil carbon, pH, and nutrient complexity (Extended data Table 1-2). Biochar was purchased from Shirotori Mokuzaikakou Cooperative Society (Gifu, Japan). The raw material consisted of 60% softwood ( Cryptomeria japonica and hinoki Chamaecyparis obtusa ) and 40% hardwood ( Quercus spp. and Zelkova serrata ). The pyrolysis was conducted at 600 °C, indicating that the material was fully carbonized. Soil samples were taken from each field at 0–10 cm depth in March 2023. Each treatment had two plots and we sampled soils from five location in each plot and then homogenized after sieving through a 2 mm mesh. Microbial communities within each plot represented with two replicates, making four replicates per treatment (source data). Soil pH (H2O, KCl) and electric conductivity (EC) were measured using a soil to solution ratio of 1:5 after shaking for 2 h with a pH electrode (LAQUA F-72, Horiba, Ltd., Kyoto, Japan) and with a conductivity meter (ES-51, Horiba, Ltd., Kyoto, Japan), respectively. Total carbon and total nitrogen contents were determined using the dry combustion method with an elemental analyser (Sumigraph NC-220F; Sumika Chemical Analysis Service, Osaka, Japan). Exchangeable bases (Ca 2+ , Mg 2+ , K + , Na + ), exchangeable Al 3+ , and cation exchangeable capacity (CEC) were measured with inductively coupled plasma atomic emission spectroscopy with a spectrometer (ICPE-9000, Shimadzu Corporation, Kyoto, Japan). Base saturation (BS) was calculated by dividing exchangeable cations (Ca 2+ + Mg 2+ + K + + Na + ) by CEC. Available phosphorus was determined using the Bray-I method with a spectrophotometer (UV-1800, Shimadzu Corporation). Soils in three sugarcane fields were Argic Red-Yellow soils according to e-SoilMap II ( https://soil-inventory.rad.naro.go.jp/eSoilMap.html ), which generally correspond to Acrisols (IUSS Working Group WRB 2022). DNA was extracted from the soil (0.5 g) using ISOIL for Bead Beating (Nippon Gene, Tokyo, Japan) according to the manufacturer’s instructions and then eluted in TE buffer (50 μL). The V4 region of the 16S rRNA gene was amplified from the extracted DNA using the universal bacterial primers (515F and 806R) 45 and V9 region of 18S rRNA gene was amplified from the extracted DNA using the universal protist primers (1389F-1510R) 46 . Both primers tailed with Illumina barcoded adapters (San Diego, CA, USA). Negative control was used in all analyses from the DNA extraction through PCR to make sure contamination did not occur. Illumina MiSeq sequencing and primary analyses of raw FASTQ data were performed as described previously 21 , 43 . Bioinformatic analyses for bacteria and protists were conducted as describe above. For protists, QIIME2’s q2-feature-classifier plugin was used for taxonomy assignment against the latest of the Protist Ribosomal Reference (PR2) database 47 . To obtain exclusive protist data, non-protist sequences (Fungi, Metazoa, unidentified Opisthokonta, Streptophyta, Rhodophyta, and unclassified eukaryotes) were removed from all samples using the Qiime2 (taxa filter-table/seq). Then predators were extracted as described above. Rest of the bioinformatics and statistics were conducted in the same way as described above with a modification. Due to our aim, we calculated the community divergence within each field. To evaluate the effects of protistan traits on microbial community convergence, we employed piecewise structural equation modeling (SEM) using the piecewiseSEM package (v2.1.2) in R (v4.3.2). This approach allows for the estimation of complex causal relationships by combining multiple linear models into a single path framework, accommodating both direct and mediated effects between variables. The SEM included three component models that were selected based on important predictors of random forest model in both global and local analysis. Predator divergence, RA of ubiquitous and endemic predators were modeled as predictors of dominant and ubiquitous bacterial richness. Finally, both bacterial richness metrics were included as predictors of the ultimate outcome, community convergence. Model evaluation was based on individual model fit and the overall Fisher’s C statistic for directed separation. Standardized path coefficients were extracted and interpreted to assess the strength and direction of hypothesised causal relationships. This framework allowed us to test whether protist-mediated effects on microbial community convergence are channeled through shifts in bacterial community traits, especially those affecting dominant and ubiquitous taxa. Microcosm experiment To obtained diverse bacterial communities, we collected five soil samples from diverse fields (source data). Forest soil (Com1) was obtained from a natural forest at Nishikan ward, Niigata prefecture (N37.785488, E138.846804); a soil from grassland (Com 2) was obtained from near shoreline at Nishikan ward, Niigata prefecture (N37.787003, E138.817610); a paddy field soil (Com 3) and a soybean field (Com 5) was sampled from Shindori Station in the Field Center for Sustainable Agriculture and Forestry, Niigata University, Niigata (N37.85.69, E138.96.22); a sugarcane field (Com 4) was sampled from Sugarcane fields in Ishigaki as described above. Each soil collected from a 0–20 cm depth at five locations per field, sieved to 2 mm, and homogenized. The protist-free indigenous bacterial community was obtained with a filtration method (1.2 µm pore size mixed cellulose ester membrane filters [Advantec, Tokyo, Japan]) from the collected paddy field soils as described previously 21 . The 50 μL of protist-free bacterial inoculum was cultured in 100 μL of the amoeba saline solution 48 in 96-well microtiter plates for three weeks at 20°C. The absence of protists was confirmed weekly with an inverted microscope at ×100, ×200 and ×400 magnifications (Nikon Eclipse TE2000-S, Tokyo, Japan). The preparation and growth of axenic cultures of predatory protists, Acanthamoeba castellanii (Ac; Amoebozoa, Tubulinea; ∼20-30 µm), Vermamoeba vermiformis (Vv; Amoebozoa, Discosea; 15-35 µm), and Heteromita globosa LAP3-2017 (Hg; Rhizaria, Cercozoa; ∼10 µm) was described elsewhere 21 . Both global and local field trials demonstrated that taxonomically distinct predators show distinct impact on bacterial communities. Therefore, the three protist species were selected based on their distinct phylogenetic backgrounds, ecological relevance as common bacterivorous predators, and ability of axenic culturing. We used calcined clay as an inert soil substitute allowing us to control nutrient levels 29 , 30 . The calcined clay was prepared as described elsewhere 21 . The sterile plastic tubes (volume: 100 mL) were filled with calcined clay (27 g DW, 40 mL) and a defined nutrient media (nutrient broth [18 g L - 1 ], Eiken Chemical Co. Ltd., Tokyo, Japan) that supports the growth of a wide range of bacterial taxa and consist of peptone, beef extract, and sodium chloride. All of the microcosms were inoculated with the protist-free bacterial media (∼10 7 cells per g dry calcined clay). About 500 cells g -1 soil of each axenic protist species were added into the microcosms, while control treatment received the same amount of sterile water. We had 4 predator treatments (Ctrl; Control with no protist addition; Ac, Acanthamoeba castellanii; Vv, Vermamoeba vermiformis; Hg, Heteromita globosa) for each community, making in total 20 treatments. Each treatment was prepared with 6 replications. Microcosms were incubated at 25°C in the dark. An equal volume of nutrient medium was added every three days to compensate for water loss due to evaporation and uptake. Our previous experiments showed that the effect of protists on bacteria at the community level can be clearly observed in 3 weeks 21 , 22 . Therefore, we conducted sampling on the 3 rd and 5 th weeks. Three replicative microcosms in each treatment were destructively sampled at 3 rd and 5 th weeks. The calcined clay mixed through and 0.5 g of sample transferred to a new sterile tube. The afterward molecular analysis (DNA extraction, PCR, Illumina Miseq Sequencing), and the bioinformatics analyses were the same as previously described. Here we analysed protist enriched and protist depleted bacterial taxa, using the DESeq function in the DESeq2 package in the R program, which models raw counts using a negative binomial GLM, taking into account sample library size and the dispersion for each ASV 49 , 50 . Although differential abundance analysis tools have limitations, DESeq2 is one of the recommended tools for microbial datasets 51 . Using this model, we compared each protist treatment with the control treatment with relative abundances. DESeq analyses were performed separately for each community and each protist treatment. Since microcosm experiment yielded less ASV counts compared to field studies, ASVs with mean relative abundance >1% across all samples were classified as dominant bacterial taxa, while those with mean relative abundance <0.1% were considered rare. To identify dominant and rare bacterial ASVs of control group, we first isolated treatments without protist addition and normalized the ASV counts to relative abundances. ASVs with mean relative abundance >1% across control samples were classified as dominant in control, while those with mean relative abundance <0.1% were considered rare in control. While many convergence/divergence studies compare initial and final communities to assess shifts over time 6 , 17 , our microcosm approach diverges from this by directly contrasting predator-present and predator-absent communities. This design, although lacking a defined initial baseline, enables a focused examination of predator-induced community convergence under controlled conditions as we isolate the effect of predation itself — a critical, yet often underexplored, biotic driver of microbial assembly. Moreover, in natural settings, initial community states are rarely known, and microbial communities are shaped by ongoing biotic interactions 5 , 24 , 28 . Our predator-absent controls serve as realistic ecological baselines, allowing us to infer the relative influence of predation without relying on potentially artificial starting conditions 10 , 11 . SynCom reconstructions To understand how predator-resistant and predator-vulnerable bacterial taxa response predator-driven convergence, we selected three bacterial ASVs that were significantly depleted and three bacterial ASVs that were significantly enriched in response to protist presence in the microcosm experiment. Then, we obtained bacterial species from Japan Microbe Collection (JCM, https://jcm.brc.riken.jp/en ) that showed 100% match with our raw sequences. Our selection criteria were based on several factors. Bacterial taxa were included only if they met all of the following criteria: ubiquitous (presence in >60% of samples), either enriched or depleted by at least two out of three predators in at least 3 communities. The selected protist-depleted bacterial taxa were Telluria terrae (Tt, formerly known as Masilla terrae, ASV: 45af176ccb220ac7f05edaa76f14fb1; JCM No: 31606), Knoellia sinensi (Ks, ASV: a7da93a9744a7863292bcf97d47b6a10; JCM No: 11536) , and Luteimonas galliterrae (Lg, ASV: 33a198a979ec6f8252b36eb5ae643d5e; JCM No: 34401), while protist-enriched bacterial taxa were Priestia megaterium (Pm, formerly known as Bacillus megaterium, ASV: 83e948f389f9d97b0c975afc880989e8; JCM No: 2506 ), Achromobacter denitrificans (Ad, ASV: ecc423b8fefd03a92aa1a97fc74a654c; JCM No: 9657), and Pseudomonas kilonensis (Pk, ASV: 0ddcd311e02f742e2e0e61ce02cf9c29; JCM No: 11939). Each bacterial species was grown according to the instructions of JCM. Among those bacteria, Pk, Pm, and Ad are known as competitive and fast-growing species making them less vulnerable for predation, while Ks, Lg, and Tt are known as slow growing and less competitive species. In addition, both Pk and Pm are known to produce secondary metabolites that may increase their survival from predation chances 52 , 53 , while Ad is known for resistance to antimicrobial compounds 54 that are likely to be produced by Pk or Pm 52 , 53 . The first SynCom received equal abundances of all six bacterial species (1.8×10 6 cells g calcined clay -1 ), while the other six each had one dominant species (1.0×10 7 cells g calcined clay -1 ) and five rare ones (1.0×10 5 cells g calcined clay -1 ). All SynComs had the same total amount of bacteria (1.1×10 7 cells g calcined clay -1 ). The SynComs were incubated for 3 weeks under same experimental conditions as described above with one modification. Here we used one gram of calcined clay per SynCom in 2 mL tubes. Sampling and DNA extraction was also same as described above. To quantify the gene abundances of each bacterial species, we used a quantitative real-time PCR (qPCR), for which we have designed species-specific primers as follows. View this table: View inline View popup Each bacterial species in each treatment was separately amplified making total 504 samples. The qPCR analysis was conducted as previously described 21 . Briefly, one µL of the standardized DNA extracts (5 ng µL -1 ) was used in the qPCR analysis. The qPCR reaction (25 μL) contained 10 pmol of each primer and 5 μL of SsoFastTM EvaGreen supermix (Bio-Rad, Hercules, CA, USA) and ran in a CFX96TM Real-Time System (Bio-Rad). The qPCR program started with an initial denaturation step of 180 s at 95 °C, followed by 40 cycles of denaturation (45 s, 95 °C) and primer annealing (60 s, 57°C to 60 °C depending on the primer set) with a final step of primer extension of 30 s at 72 °C. The quality and size of the generated amplicons were checked by gel electrophoresis and melting curve analysis. We conducted these analyses using six 96-well plates each contained only one primer set for only one bacterial species and dilutions of original of each bacterial media for calculating copy numbers. Copy numbers for each gene were calculated using a regression equation for each assay relating the cycle threshold (Ct) values to the known number of copies in the standards of each bacterial taxa. The remaining bioinformatics and statistical analyses were conducted as described above with one modification. In NMDS analysis, a log transform was applied to the ASV table consisting of 16S rRNA gene copy numbers to handle heteroscedasticity in the data 55 using log function in R, then Bray–Curtis distance matrix was prepared. Then, the community divergence was calculated from the Bray–Curtis distance matrix using log transformed ASV table (source data). Net convergence effect To create a model that predicts predator-driven divergence and convergence, we formulated divergence and convergence effects of predators. As high community divergence of predators and high abundance of endemic predators correlates with bacterial community divergence, we multiply the predator divergence and RA of endemic predators for divergence impact (see formula below). We divided RA of ubiquitous predators to predator divergence for convergence impact as high abundance of ubiquitous predators and low community divergence of predators correlates with bacterial community convergence. Then we substate predator divergence effect from predator convergence effect. Positive values imply that predators are more likely to be responsible of bacterial community convergence, while negative values imply the opposite. Predators’ divergence impact (PDI) = RAE × DP Predators’ convergence impact (PCI) = RAU / DP Net convergence effect (NCE) = PCI – PDI Where, RAE, Relative abundance of endemic predators; RAU, Relative abundance of ubiquitous predators. Extended data Download figure Open in new tab Extended data Figure 1 Predators and Bacterial community composition at global scale. Relative abundance of bacterivorous protists (predators) grouped by ecosystem type ( a ) and by continent ( b ). The relative abundance of predators represents their abundance distribution among all protists. Community divergence of bacteria ( c and d ) and predators ( c and d ) grouped by ecosystem type ( c and e ) and by continent ( d and f ). Box plots show medians (central line), interquartile ranges (hinges), and range (whiskers). Nonmetric multidimensional scaling (NMDS) of Bray– Curtis dissimilarities showing bacterial ( g ) and predator ( h ) communities across samples. Colors indicate ecosystem type, shapes indicate continent. Download figure Open in new tab Extended data Figure 2 Relationship between predator traits and environmental variables. Left column, predator divergence, middle column, relative abundance of endemic predators, and right column, relative abundance of ubiquitous predators. From top to bottom, random forest predictor, linear correlation between protist traits and soil pH, aridity index, minimum temperature, and soil carbon. Random Forest importance is expressed as the percent increase in mean squared error (%IncMSE) when each variable is permuted; higher values indicate stronger predictive influence. Bar colors indicate significance levels (red, P < 0.01; black, P < 0.05; light grey, not significant). Regression lines with 95% confidence intervals are shown. Pearson correlation coefficients ( P values) and R² indicate statistical strength and significance. Download figure Open in new tab Extended data Figure 3 Relationship between bacterial community divergence and predator traits. In all figures, the y-axis represents bacterial community divergence, while the x-axis indicates predator traits. ASV 1 to 5 indicates relative abundances of the top five dominant ASVs in predator community. Regression lines with 95% confidence intervals are shown. Pearson correlation coefficients ( P values) and R² indicate statistical strength and significance. Download figure Open in new tab Extended data Figure 4 Predators and Bacterial community composition at local field trial. Divergence of bacterial ( a ) and protist ( b ) communities, and relative abundance ( c ) of bacterivorous protists (predators) grouped by treatments in each field. The relative abundance of predators represents their abundance distribution among all protists. Box plots show medians (central line), interquartile ranges (hinges), and range (whiskers). Nonmetric multidimensional scaling (NMDS) of Bray–Curtis dissimilarities showing bacterial ( d ) and predator ( e ) communities across samples. Colors indicate treatments, shapes indicate fields. Download figure Open in new tab Extended data Figure 5 Relationship between predator traits and environmental variables. Left column, predator divergence, middle column, relative abundance of endemic predators, and right column, relative abundance of ubiquitous predators. From top to bottom, random forest predictor, linear correlation between protist traits and soil carbon, soil pH, soil nitrogen, and soil carbon to nitrogen ratio. Random Forest importance is expressed as the percent increase in mean squared error (%IncMSE) when each variable is permuted; higher values indicate stronger predictive influence. Bar colors indicate significance levels (red, P < 0.01; black, P < 0.05; light grey, not significant). Regression lines with 95% confidence intervals are shown. Pearson correlation coefficients ( P values) and R² indicate statistical strength and significance. Download figure Open in new tab Extended data Figure 6 Bacterial community composition within each community. Nonmetric multidimensional scaling (NMDS) of Bray–Curtis dissimilarities showing bacterial dissimilarities within Com 1 to 5 ( a-e ) in control (Ctrl) and predator treatments (Ac: Acanthamoeba castellanii ; Hg: Heteromita globosa ; Vv: Vermamoeba vermiformis ). Download figure Open in new tab Extended data Figure 7 Bacterial community divergence. Box plots of within-group variance in community divergence across predator treatments in all ( a ), predator-depleted ( b ), predator-enriched ( c ), and predator-neutral ASVs ( d ). Box plots show medians (central line), interquartile ranges (hinges), and range (whiskers). Download figure Open in new tab Extended data Figure 8 Venn diagram showing overlap of the protist enriched (a) and depleted (b) bacterial ASVs among the five communities for each protist. ASVs were detected by DESeq based on the differences in relative abundances between each protist treatment and its control. Red, Acanthamoeba castellanii (Ac); blue, Heteromita globosa (Hg); green, Vermamoeba vermiformis (Vv). View this table: View inline View popup Download powerpoint Extended data Table 1. Experimental set-up. View this table: View inline View popup Download powerpoint Extended data Table 2. Chemical composition of organic fertilisers Acknowledgements We acknowledge the invaluable contributions of Prof. Dr. Harada Naoki, who sadly passed away on May 26 th , 2025. This paper is dedicated to his memory. This research was partly funded by the Japan Society for the Promotion of Science (JSPS) to Asiloglu R (JP22K14804), Ikazaki K(JP25K02147), Murase J (JP24K01654). We thank the authors of Delgado-Baquerizo et al., 2018 and Oliverio et al., 2020 for making their data publicly available, which contributed to our global meta-analysis. We are grateful to Mr. Iritakenishi Atsushi, Dr. Anzai Toshihiko, Dr. Terashima Yoshifumi, and Dr. Kanda Takashi (JIRCAS), for their support in the field trials in Ishigaki (Ishigakijima Sugar Manufacturing Co., Ltd.). Funder Information Declared Japan Society for the Promotion of Science (JSPS) , JP22K14804 , JP25K02147 , JP24K01654 Footnotes ↵ † Prof. Dr. HARADA Naoki passed away on May 26 th , 2025. This paper is dedicated to his memory. References 1. ↵ Leander , B. S. Predatory protists . Curr. Biol. 30 , R510 – R516 ( 2020 ). OpenUrl CrossRef PubMed 2. ↵ Hu , S. K. et al. Protistan grazing impacts microbial communities and carbon cycling at deep-sea hydrothermal vents . Proc. Natl Acad. Sci. USA 118 , e2102674118 ( 2021 ). OpenUrl Abstract / FREE Full Text 3. ↵ Burian , A. et al. Predation increases multiple components of microbial diversity in activated sludge communities . ISME J . 16 , 1086 – 1094 ( 2022 ). OpenUrl PubMed 4. ↵ Lozupone , C. A. & Knight , R . Global patterns in bacterial diversity . Proc. Natl Acad. Sci. USA 104 , 11436– 11440 ( 2007 ). 5. ↵ Delgado-Baquerizo , M. et al. A global atlas of the dominant bacteria found in soil . Science 359 , 320 – 325 ( 2018 ). OpenUrl Abstract / FREE Full Text 6. ↵ Silverstein , M. R. , Bhatnagar , J. M. & Segrè , D . Metabolic complexity drives divergence in microbial communities. Nat . Ecol. Evol . 8 , 1493 – 1504 ( 2024 ). OpenUrl 7. ↵ Knight , C. G. et al. Soil microbiomes show consistent and predictable responses to extreme events . Nature 636 , 690 – 696 ( 2024 ). OpenUrl CrossRef PubMed 8. ↵ McClean , D. et al. Single gene locus changes perturb complex microbial communities as much as apex predator loss . Nat. Commun . 6 , 8235 ( 2015 ). OpenUrl CrossRef PubMed 9. ↵ Guo , P. , Li , C. , Liu , J. & Chai , B . Predation has a significant impact on the complexity and stability of microbial food webs in subalpine lakes . Microbiol. Spectr . 11 , e02411 – 23 ( 2023 ). OpenUrl 10. ↵ Bell , T. et al. Protists have divergent effects on bacterial diversity along a productivity gradient . Biol. Lett . 6 , 639 – 642 ( 2010 ). OpenUrl CrossRef PubMed 11. ↵ Saleem , M. , Fetzer , I. , Harms , H. & Chatzinotas , A . Diversity of protists and bacteria determines predation performance and stability . ISME J . 7 , 1912 – 1921 ( 2013 ). OpenUrl CrossRef PubMed 12. ↵ Lawson , C. E. et al. Common principles and best practices for engineering microbiomes . Nat. Rev. Microbiol . 17 , 725 – 741 ( 2019 ). OpenUrl CrossRef PubMed 13. ↵ Guo , S. et al. Predatory protists reduce bacteria wilt disease incidence in tomato plants . Nat. Commun . 15 , 829 ( 2024 ). 14. ↵ Jiang , G. , Liu , C. , Xiong , W. , Shen , Q. & Wei , Z . Protist predation selects for the soil resistome . ISME J . 18 , ( 2024 ). 15. ↵ Philippot , L. , Chenu , C. , Kappler , A. , Rillig , M. C. & Fierer , N . The interplay between microbial communities and soil properties . Nat. Rev. Microbiol . 22 , 226 – 239 ( 2024 ). OpenUrl CrossRef PubMed 16. Zhang , X. et al. Local community assembly mechanisms shape soil bacterial β diversity patterns along a latitudinal gradient . Nat. Commun . 11 , 5428 ( 2020 ). OpenUrl PubMed 17. ↵ Goldford , J. E. et al. Emergent simplicity in microbial community assembly . Science 361 , 469 – 474 ( 2018 ). OpenUrl Abstract / FREE Full Text 18. ↵ Hibbing , M. E. , Fuqua , C. , Parsek , M. R. & Peterson , S. B . Bacterial competition: surviving and thriving in the microbial jungle . Nat. Rev. Microbiol . 8 , 15 – 25 ( 2010 ). OpenUrl CrossRef PubMed Web of Science 19. ↵ Geisen , S. et al. Soil protists: A fertile frontier in soil biology research . FEMS Microbiol. Rev . 42 , 293 – 323 ( 2018 ). OpenUrl CrossRef PubMed 20. ↵ Gao , Z. , Karlsson , I. , Geisen , S. , Kowalchuk , G. & Jousset , A . Protists: Puppet masters of the rhizosphere microbiome . Trends Plant Sci . 24 , 165 – 176 ( 2019 ). OpenUrl CrossRef PubMed 21. ↵ Asiloglu , R. et al. Trophic modulation of endophytes by rhizosphere protists . ISME J . 18 , wrae235 ( 2024 ). 22. ↵ Asiloglu , R. et al. Top-down effects of protists are greater than bottom-up effects of fertilisers on the formation of bacterial communities in a paddy field soil . Soil. Biol. Biochem . 156 , 108186 ( 2021 ). 23. ↵ Bodur , S. O. , Suzuki , K. , Harada , N. & Asiloglu , R . Top-down predators shape soil bacterial community composition while bottom-up nutrients drive bacterial abundance . Sci. Total Environ . 957 , 177634 ( 2024 ). 24. ↵ Oliverio , A. M. et al. The global-scale distributions of soil protists and their contributions to belowground systems . Sci. Adv . 6 , eaax8787 ( 2020 ). 25. ↵ Janzen , D. H . Herbivores and the number of tree species in tropical forests . Am. Nat . 104 , 501 – 528 ( 1970 ). OpenUrl CrossRef Web of Science 26. ↵ Chesson , P . Mechanisms of maintenance of species diversity . Annu. Rev. Ecol. Syst . 31 , 343 – 366 ( 2000 ). OpenUrl CrossRef Web of Science 27. ↵ Good , B. H. , McDonald , M. J. , Barrick , J. E. , Lenski , R. E. & Desai , M. M . The dynamics of molecular evolution over 60,000 generations . Nature 551 , 45 – 50 ( 2017 ). OpenUrl CrossRef PubMed 28. ↵ Thompson , L. R. et al. A communal catalogue reveals Earth’s multiscale microbial diversity . Nature 551 , 457 – 463 ( 2017 ). OpenUrl CrossRef PubMed 29. ↵ Zhang , J. et al. NRT1.1B is associated with root microbiota composition and nitrogen use in field-grown rice . Nat. Biotechnol . 37 , 676 – 684 ( 2019 ). OpenUrl CrossRef PubMed 30. ↵ Lebeis , S. L. et al. Salicylic acid modulates colonization of the root microbiome by specific bacterial taxa . Science 349 , 860 – 864 ( 2015 ). OpenUrl Abstract / FREE Full Text 31. ↵ Becks , L. , Hilker , F. M. , Malchow , H. , Jürgens , K. & Arndt , H . Experimental demonstration of chaos in a microbial food web . Nature 435 , 1226 – 1229 ( 2005 ). OpenUrl CrossRef PubMed Web of Science 32. ↵ Saleem , M. , Fetzer , I. , Dormann , C. F. , Harms , H. & Chatzinotas , A . Predator richness increases the effect of prey diversity on prey yield . Nat. Commun . 3 , 1305 ( 2012 ). OpenUrl CrossRef PubMed 33. ↵ Matz , C. & Kjelleberg , S . Off the hook – how bacteria survive protozoan grazing . Trends Microbiol . 13 , 302 – 307 ( 2005 ). OpenUrl CrossRef PubMed Web of Science 34. ↵ Liu , C. et al. Protist predation promotes antimicrobial resistance spread through antagonistic microbiome interactions . ISME J . 18 , wrae169 ( 2024 ). 35. ↵ Nguyen , T. B.-A. et al. Protistan predation selects for antibiotic resistance in soil bacterial communities . ISME J . 17 , 2182 – 2189 ( 2023 ). OpenUrl PubMed 36. ↵ Hahn , M. , Lünsdorf , H. & Janke , L . Exopolymer production and microcolony formation by planktonic freshwater bacteria: defence against protistan grazing . Aquat. Microb. Ecol . 35 , 297 – 308 ( 2004 ). OpenUrl CrossRef 37. ↵ Fujino , M. , Bodur , S. O. , Harada , N. & Asiloglu , R . Guardians of plant health: roles of predatory protists in the pathogen suppression . Plant Soil 509 , 5 – 13 ( 2024 ). OpenUrl 38. ↵ Rutherford , P. M. & Juma , N. G . Influence of texture on habitable pore space and bacterial-protozoan populations in soil . Biol. Fertil. Soils 12 , 221 – 227 ( 1992 ). OpenUrl 39. ↵ Benincà , E. et al. Chaos in a long-term experiment with a plankton community . Nature 451 , 822 – 825 ( 2008 ). OpenUrl CrossRef PubMed Web of Science 39. ↵ Bolyen , E. et al. Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2 . Nat. Biotechnol . 37 , 852 – 857 ( 2019 ). OpenUrl CrossRef PubMed 40. ↵ Callahan , B. J. et al. DADA2: High-resolution sample inference from Illumina amplicon data . Nat. Methods 13 , 581 – 583 ( 2016 ). OpenUrl CrossRef PubMed 41. ↵ Quast , C. et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools . Nucleic Acids Res . 41 , D590 – D596 ( 2012 ). OpenUrl CrossRef PubMed Web of Science 42. ↵ Asiloglu , R. , Shiroishi , K. , Suzuki , K. , Turgay , O. C. & Harada , N . Soil properties have more significant effects on the community composition of protists than the rhizosphere effect of rice plants in alkaline paddy field soils . Soil Biol. Biochem . 161 , 108397 ( 2021 ). 43. ↵ Liao , H. et al. Protists regulate microbially mediated organic carbon turnover in soil aggregates . Glob. Chang. Biol . 30 , e17102 ( 2024 ). OpenUrl 44. ↵ Caporaso , J. G. et al. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms . ISME J . 6 , 1621 – 1624 ( 2012 ). OpenUrl CrossRef PubMed Web of Science 45. ↵ Amaral-Zettler , L. A. , McCliment , E. A. , Ducklow , H. W. & Huse , S. M . A method for studying protistan diversity using massively parallel sequencing of v9 hypervariable regions of small-subunit ribosomal RNA genes . PLoS One 4 , e6372 ( 2009 ). OpenUrl CrossRef PubMed 46. ↵ Guillou , L. et al. The Protist Ribosomal Reference database (PR2): a catalog of unicellular eukaryote Small Sub-Unit rRNA sequences with curated taxonomy . Nucleic Acids Res . 41 , D597 – D604 ( 2012 ). OpenUrl CrossRef PubMed 47. ↵ Page , F. C . A new key to freshwater and soil gymnamoebae: with instructions for culture . in Freshwater Biological Association vol. 107 121 (Ambleside, 1988). 48. ↵ McMurdie , P. J. & Holmes , S. Waste Not, Want Not: Why Rarefying Microbiome Data Is Inadmissible . PLoS Comput Biol 10 , e1003531 ( 2014 ). 49. ↵ Love , M. I. , Huber , W. & Anders , S . Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2 . Genome Biol . 15 , 550 ( 2014 ). 50. ↵ Calgaro , M. , Romualdi , C. , Waldron , L. , Risso , D. & Vitulo , N . Assessment of statistical methods from single cell, bulk RNA-seq, and metagenomics applied to microbiome data . Genome Biol . 21 , 191 ( 2020 ). 51. ↵ Jousset , A. , Lara , E. , Wall , L. G. & Valverde , C . Secondary metabolites help biocontrol strain Pseudomonas fluorescens cha0 to escape protozoan grazing . Appl. Environ. Microbiol . 72 , 7083 – 7090 ( 2006 ). OpenUrl Abstract / FREE Full Text 52. ↵ Santos , S. S. , Hendriksen , N. B. , Jakobsen , H. H. & Winding , A . Effects of Bacillus cereus endospores on free-living protist growth . Microb. Ecol . 73 , 699 – 709 ( 2017 ). OpenUrl PubMed 53. ↵ Isler , B. , Kidd , T. J. , Stewart , A. G. , Harris , P. & Paterson, D. L. Achromobacter Infections and treatment options . Antimicrob. Agents Chemother . 64 , 10 – 128 ( 2020 ). OpenUrl CrossRef 54. ↵ Barlow , J. T. , Bogatyrev , S. R. & Ismagilov , R. F . A quantitative sequencing framework for absolute abundance measurements of mucosal and lumenal microbial communities . Nat. Commun . 11 , 2590 ( 2020 ). OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted August 20, 2025. Download PDF Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Predator-driven local convergence fosters global microbial community divergence Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Predator-driven local convergence fosters global microbial community divergence Rasit Asiloglu , Hayato Kuno , Mayu Fujino , Seda Bodur , Murat Aycan , Haruka Ishizuka , Shiori Kazama , Shinya Iwasaki , Jun Murase , Naoki Harada , Miwa Arai , Kenta Ikazaki bioRxiv 2025.08.15.670637; doi: https://doi.org/10.1101/2025.08.15.670637 Share This Article: Copy Citation Tools Predator-driven local convergence fosters global microbial community divergence Rasit Asiloglu , Hayato Kuno , Mayu Fujino , Seda Bodur , Murat Aycan , Haruka Ishizuka , Shiori Kazama , Shinya Iwasaki , Jun Murase , Naoki Harada , Miwa Arai , Kenta Ikazaki bioRxiv 2025.08.15.670637; doi: https://doi.org/10.1101/2025.08.15.670637 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Ecology Subject Areas All Articles Animal Behavior and Cognition (7617) Biochemistry (17633) Bioengineering (13856) Bioinformatics (41841) Biophysics (21399) Cancer Biology (18529) Cell Biology (25422) Clinical Trials (138) Developmental Biology (13352) Ecology (19860) Epidemiology (2067) Evolutionary Biology (24281) Genetics (15582) Genomics (22461) Immunology (17700) Microbiology (40293) Molecular Biology (17140) Neuroscience (88413) Paleontology (666) Pathology (2823) Pharmacology and Toxicology (4813) Physiology (7632) Plant Biology (15107) Scientific Communication and Education (2042) Synthetic Biology (4284) Systems Biology (9808) Zoology (2267)