Effect of Herbicide Combinations on Bt-Maize Rhizobacterial DiversityS
Effect of Herbicide Combinations on Bt-Maize Rhizobacterial DiversityS
Journal of Microbiology and Biotechnology. 2014. Nov, 24(11): 1473-1483
Copyright © 2014, The Korean Society For Microbiology And Biotechnology
  • Received : May 22, 2014
  • Accepted : July 22, 2014
  • Published : November 28, 2014
Export by style
Cited by
About the Authors
José R. Valverde
Silvia Marín
Rafael P. Mellado

Reports of herbicide resistance events are proliferating worldwide, leading to new cultivation strategies using combinations of pre-emergence and post-emergence herbicides. We analyzed the impact during a one-year cultivation cycle of several herbicide combinations on the rhizobacterial community of glyphosate-tolerant Bt-maize and compared them to those of the untreated or glyphosate-treated soils. Samples were analyzed using pyrosequencing of the V6 hypervariable region of the 16S rRNA gene. The sequences obtained were subjected to taxonomic, taxonomy-independent, and phylogeny-based diversity studies, followed by a statistical analysis using principal components analysis and hierarchical clustering with jackknife statistical validation. The resilience of the microbial communities was analyzed by comparing their relative composition at the end of the cultivation cycle. The bacterial communites from soil subjected to a combined treatment with mesotrione plus s-metolachlor followed by glyphosate were not statistically different from those treated with glyphosate or the untreated ones. The use of acetochlor plus terbuthylazine followed by glyphosate, and the use of aclonifen plus isoxaflutole followed by mesotrione clearly affected the resilience of their corresponding bacterial communities. The treatment with pethoxamid followed by glyphosate resulted in an intermediate effect. The use of glyphosate alone seems to be the less aggressive one for bacterial communities. Should a combined treatment be needed, the combination of mesotrione and s-metolachlor shows the next best final resilience. Our results show the relevance of comparative rhizobacterial community studies when novel combined herbicide treatments are deemed necessary to control weed growth.
Numerous studies have shown that weeds are able to significantly reduce maize yield (between 35% and 60%), particularly during the so-called critical period of weed control [22 , 38] , thus leading to the widespread use of herbicides to control the growth of competing, undesirable weeds. Herbicide usage has been a common agricultural practice since the middle of the 20 th Century, and over the years has led to the emergence of multiple resistance events in weeds (218 species tolerant to 148 different herbicides as of October 2013 are reported in ). To address this problem, seed companies have resorted to producing new recombinant crops tolerant to broader-range non-selective herbicides such as glyphosate. However, owing to the continued use of these products, new weed tolerance events are increasingly taking place as well [37] , and the development of new strategies for weed control is becoming more relevant, such as combinations of pre-emergence and post-emergence herbicides in the same seeding-harvest cycle, often in conjunction with the cultivation of genetically modified plants.
Herbicides are strong chemical products that not only act on their target weeds but may also display significant toxicity to other organisms [14 , 42] ; among these, the bacteria present in the rhizosphere, which potentially affect plant growth, herbicide degradation capacity of the soil, and the ability of soils to improve crop yield [13 , 25] .
Plant growth-promoting bacteria (PGPB) and rhizobacteria (PGPR) community diversity and evenness have been reported to increase yield by stimulating plant growth, facilitating plant adaptation to adverse or stressing environments, and by reducing susceptibility to pathogen infection by means of a variety of mechanisms. PGPR are plant and soil specific and have been used successfully to facilitate plant seeding, rooting, growth, and yield under aggressive conditions [13] .
The structure of the specific maize rhizobacterial community displays a stronger dependence on soil properties than on cultivar variety. The maize rhizosphere has been shown to be enriched in Proteobacteria, particularly Burkholderiales, which are known to respond to labile carbon sources and are considered r-selected, fast growing, and opportunistic, in contrast to bulk soil, which is enriched in k-selected, slower growing microbia [40] . Actinobacteria play a major role in soil owing to their ability to decompose organic materials and secrete growth-promoting substances, antibiotics, enzymes, and enzyme inhibitors, in addition to many other compounds [27] . Acidobacteria and Verrucomicrobia, although difficult to cultivate and less well known, are ubiquitous, particularly in soils, and are likely to play a relevant role in terrestrial ecosystem processes [33] .
The potential effects of the specific herbicides and their dosage, and of soil properties and maize variants on the rhizobacterial communities have been the subject of several short- and long-term studies since their introduction ( e.g. , [10 , 17 , 19] ). Interestingly, recent studies using Next Generation Sequencing (NGS) and metagenomics have confirmed the frail impact of the Cry1Ab protein [8] or soil properties [7] on the effects of glyphosate, which in turn shows little impact both in cotton [6] and maize [4 , 5] in contrast to the tank mix (combination in a single formulation) of acetochlor plus terbuthylazine [3] . Given its weaker impact, glyphosate is the obvious choice for use in combined treatments.
Existing studies have produced significant insights into the relative impact of each agent and have shown that tank mix use of herbicide combinations may display significant antagonistic, synergistic, or additive effects [43] , and have suggested that time-separated administration might mitigate their combined effects. However, until now there is very little knowledge as to the effects of the combination of pre- and post-emergence applications. Given this scarce knowledge, we have analyzed the potential impact of new strategies on rhizobacteria combining the pre-emergence application of various herbicides with the post-emergence treatment with glyphosate.
Materials and Methods
- Herbicides Tested
We have considered the use of various time-separated commercial herbicide preparations in combination with glyphosate (hence requiring a glyphosate-tolerant maize variant).
Two control demonstration fields were seeded: one field contained solely Bt-maize with no treatment at all, and a second field was treated only in post-emergence with 360 g/l of glyphosate as isopropylamine salt (Roundup Ready, 72 kg/ha).
Three experimental demonstration fields were treated, each one in pre-emergence, with a commercial preparation of either pethoxamid 60% (Successor, 2 l/ha), a combination of mesotrione 4% and s-metolachlor 40% (Camix, 3 l/ha), or a combination of acetochlor 41% plus terbuthylazine 19.5% (Harness GTZ, 3 l/ha), and were subsequently treated in post-emergence with glyphosate, as described.
An additional demonstration field was cultivated using a combined treatment suitable for conventional, non-glyphosatetolerant maize consisting of a combination of aclonifen 50% plus isoxaflutole 7.5% (Lagon, 1 l/ha) applied pre-emergence, followed by mesotrione 10% (Callisto, 1 l/ha) applied post-emergence. All doses used correspond to current standard practices.
- Texture and Chemical Properties of the Soil
Maize samples were grown in experimental demonstration fields located in Ejea de los Caballeros (42˚ 2’ 16’’ N, 1˚ 13’ 47’’ W, Zaragoza, Spain). The surface of each experimental demonstration field measured 40 ㎡, and all of the demonstration fields were annexed to each other and separated by a 4-m-wide path.
The agricultural soil texture and composition were uniform across all of the demonstration fields and composed of clay loam consisting of 43% sand, 29% clay, and 28% silt, with 3.27% oxidable organic matter (Wakey-Black), a pH of 8.17, a cationic exchange capacity of 15.20 meq/100 g, and an electrical conductivity of 242 μS/cm at 20℃, as determined by the independent contractor Labs & Technological Services AGQ, S. L.
Since all soils share the same properties, we used typical values of DT 50 (a measure of contaminant degradation time) and K FOC (based on the Freundlich isotherm and normalized for organic carbon values) to discuss the differences in herbicide leaching and soil sorption properties ( Table 1 ), as recommended by FOCUS ( ).
Herbicide leaching propertiesa.
PPT Slide
Lager Image
aValues for relevant herbicide leaching properties were obtained from the Pesticide Properties Database website ( bSolubility in water at 20℃ is expressed as mg/l. cTypical 50% dissipation time (DT50) values for each herbicide, expressed in days.
- Plant Material and Sampling
Previous studies [4 , 5 , 40] have confirmed that regional conditions exert the largest impact on bacterial biodiversity, and that, within each location, differences due to the herbicide treatment are the most relevant, indicating that comparisons should be based on relative differences within each location [7] . It has also been shown that maize production of the Cry1Ab protein exerted a minor or undetectable effect [8] . To reduce any possible confounding variables in the study, all experiments and samplings were conducted simultaneously, using the same maize variant, in demonstration fields with the same soil, hence comprising the same texture, NPK composition, and properties, and subject to the same environmental and climatic factors as well as the same agricultural practices.
Equivalent numbers of plants were used in each of the experimental fields: Glyphosate-tolerant Bt-maize, event NK603, MON810, variety DKC6451YG (Monsanto Agricultura, Spain), expressing the Cry1Ab Bt toxin that confers resistance to the European corn borer was grown using current agricultural practices throughout the cultivation period.
In order to deal with local soil micro-heterogeneity, and given the small size of the demonstration fields, these were divided into nine randomized subplots, and three subsamples from each subplot were taken at each sampling time. Equal amounts of soil (10 g) from the 27 extractions from each maize field at each sampling time were pooled for the subsequent analysis of each soil to obtain an average figure of the rhizobacterial community in each sample. No appreciable differences in plant weight, length, above-ground weight, number of leaves, root system, or yield were detected among the plants grown in the different fields at any given sampling time. Hence, the differences in the results are expected to be due to the differences in the herbicide treatments on the rhizosphere bacterial community.
We use the following terminology for fields according to the corresponding treatment applied: C (control field), G (glyphosate only), PG (pethoxamid and glyphosate), MSG (mesotrione plus smetolachlor and glyphosate), ATG (acetochlor plus terbuthylazine and glyphosate), and AIM (aclonifen plus isoxaflutole and mesotrione). Samples are termed according to the field, followed by a number (1 or 2 depending on the sampling time), hence C1 stands for “control field, first sampling time.”
- DNA Extraction, PCR Amplification, and Pyrosequencing
During the course of 2012, samples were harvested from each field at two different stages: soon after emergence when the plants were shorter than 10 cm and had 2-3 leaves (first sampling time), and just before crop harvesting, approximately six months after seeding (final sampling time). The root system and adhered soil measuring approximately 2 mm or less in diameter were shaken gently to separate bulk soil. Soil adhered to these roots was carefully extracted to separate the rhizosphere from the plant roots.
The rhizospheres from each field at each collection time were subjected to three independent DNA extractions using PowerMax Soil DNA kits (MO Bio Laboratories Inc., Carlsbad, USA) following the protocol provided by the manufacturer. DNA obtained from each of these three independent extractions was used as template for PCR amplification of the V6 region of the 16S rRNA gene. The oligonucleotide design included 454 Life Science’s Titanium A (5’- CGTATCGCCTCCCTCGCGCCATCAG-3’) or B (5’-CTATGCGCCTTGCCAGCCCGCTCAG-3’) pyrosequencing adaptors fused to the 5’ end of the template-specific primers 967F (CAACGCGAAGAACCTTACC) or 1046R (CGACAGCCATGCANCACCT), and a sample-specific MID (multiplex identifier) immediately preceding the V6-specific primer ( Table 2 ). Soil DNA was used as template for PCR amplification with incubation at 95℃ for 5 min, followed by 30 cycles of incubation at 95℃ (30 sec), 63℃ (45 sec), and 72℃ (1 min), with a final extension cycle of 5 min at 72℃ [3] . The resulting DNAs obtained from the three independent PCRs were pooled and cleaned using an Illustra GFX PCR DNA Amplification kit (GE Healthcare), checked with Bioanalyzer 2100 (Agilent Technologies), quantified with Quant-IT-picogreen (Invitrogen), and used to make the single strands on beads, as required for 454 Titanium pyrosequencing [35] . Pyrosequencing was conducted using a 454 Life Sciences Titanium pyrosequencer (Life Sequencing S. L. and the Research Centre for Public Health, CSISP, Valencia, Spain). Bar coding allowed us to analyze two samples per single 454 available lane ( Table 2 ).
Samples and bar codes used.
PPT Slide
Lager Image
aExperimental fields have been labeled according to treatment (C: control; G: glyphosate; PG: pethoxamid and glyphosate; MSG: mesotrione plus s-metolachlor and glyphosate; ATG: acetochlor plus terbuthylazine and glyphosate; AIM: aclonifen plus isoxaflutole and mesotrione). bThe “Time” column indicates sampling time: 1 (soon after emergence) or 2 (late before harvest). cPre-emergence and dpost-emergence herbicide treatments applied to each field. eMultiplex identification tags are the oligonucleotides used to separate the sequences obtained from each sample. fReads indicate the number of sequences recovered from each sample.
- Metagenomic Sequence Analysis
The sequence data were quality-filtered at various levels: first, allowing 3’ trimming (trim back valley filter), followed by a second step using QIIME [11] with a quality score window of 50, allowing for no ambiguity and truncating sequences at the reverse primers. After splitting the sequences according to their bar code, chimeras were detected with Otupipe’s UCHIME [18] using de novo and Gold database-based reference filtering and removed. The remaining sequences were then compared using BLAST [1] against the SILVA database release 92 [41] for taxonomical classification. BLAST output files were used to generate the corresponding taxonomic trees by the Lowest Common Ancestor (LCA) method using MEGAN [28] and NCBI taxonomy. Observed differences in the relative proportions of taxa among samples were statistically validated, normalizing the data and using the Directed Homogeneity test with both the Bonferroni and Holm-Bonferroni corrections for multiple samples using MEGAN. Only differences among major taxa with p -values < 0.05 were considered for discussion.
An additional taxonomy-based filtering step was applied next: reads identified as contaminants by taxonomy (Eukaryota and Archaea), typically less than 0.5%, were removed from the analysis. From the remaining reads, sequences that could not be assigned to specific bacterial taxa were separated and clustered using Otupipe [18] to identify and remove singletons, which are most likely considered errors [44] . The remaining unassigned reads were pooled back with the other bacterial sequences for subsequent analysis.
To perform the alpha diversity analysis, QIIME was used to identify observed OTUs (Operational Taxonomic Units) and to compute the Chao1 estimates and the Shannon-Wiener Index as well as their respective rarefaction curves at 97% similarity (conventionally considered equivalent to species-level). Given the impossibility of comparing values beyond the minimal read depth, an amount of 4,000 reads from the first sampling time and of 30,000 reads from the final sampling time were selected as a balance, and a bootstrap stratified by treatment and sampling time using 50 randomly selected subsets from each sample was used to calculate the average and 95% confidence interval of the mean for each value.
For the beta-diversity analyses (comparison among populations), we compared the filtered sequences against Greengenes [15] using BLAST to obtain a mapping against the Greengenes reference phylogenetic tree to be used in the calculation of continuous weighted UniFrac metrics [24] . The matrix of UniFrac values obtained was used to carry out a Principal Components Analysis (PCA). The statistical significance of the relationships observed between groups was assessed by hierarchical clustering and determination of statistical support for each clustering step (branches in the sample grouping hierarchy) using jack-knifing with 1,000 permutations, keeping 2,100 sequences and using normalized abundance weights to account for uneven sample size.
The sequences obtained have been deposited in the EBI SRA archive (study accession number PRJEB4333).
- Taxonomical Analysis
The taxonomic breakdown of the more relevant phyla for each sample is shown in Fig. 1 and Table S1. On average, 32% of the reads could not be assigned to specific taxa below the Bacteria level in each sample. The assigned reads were used to compare the relative amounts of the different taxa among the samples. Only changes with highly significant p -values in the corrected Directed Homogeneity test were considered ( p -values have been omitted for simplicity).
PPT Slide
Lager Image
Taxonomical breakdown resulting from pyrosequencing region V6 of the 16S rRNA of the rhizobacterial communities of glyphosate-tolerant Bt-maize subject to various herbicide combinations. Rhizobacterial community samples were collected at two different sampling times from soils subject to different herbicide treatments (C: control; G: glyphosate; PG: pethoxamid and glyphosate; MSG: mesotrione plus s-metolachlor and glyphosate; ATG: acetochlor plus terbuthylazine and glyphosate; AIM: aclonifen plus isoxaflutole and mesotrione). The taxonomical breakdown at the phylum level of each sample rhizobacterial community at the initial (A) and final (B) sampling times is displayed as a pie chart. The most relevant phyla (presence above 1%) are charted with percentages rounded to the closest units displayed next to them.
The changes between C1 and C2 reflect the influence of plant growth on the relative composition of the rhizobacterial community, showing that most major phyla experienced an increase, with the exception of Acidobacteria whose relative presence decreased, thus providing the baseline for the rhizosphere changes induced by plant growth.
Treatment with glyphosate alone (G) provided a second reference to identify the additional effects of the combination treatments. Sample G1 showed a reduction in Proteobacteria, compensated by minor increases in Verrucomicrobia and Actinobacteria with respect to the untreated soil. In G2, Acidobacteria experienced a major reduction, compensated by increases in Verrucomicrobia, Actinobacteria, and Proteobacteria that brought the community structure closer to that of C2.
In the ATG field, the herbicide combination resulted initially (ATG1) in a decrease in Proteobacteria and Acidobacteria, counterbalanced by an increase in Verrucomicrobia and other changes in minor phyla with respect to the untreated soil. After plant growth (ATG2), Proteobacteria and Acidobacteria continued decreasing, while the presence of Verrucomicrobia and Actinobacteria increased. A comparison between ATG2 and C2 indicated that Proteobacteria had experienced a large reduction, opening a gap that was compensated by increases in other groups, mainly Verrucomicrobia. Compared with G2, the addition of acetochlor and terbuthylazine resulted in a 2-fold reduction of Proteobacteria, accompanied by a reduction of Actinobacteria, confirming that the effects were worse than the treatment with glyphosate alone.
The PG soil initially (PG1) showed a reduction of Actinobacteria, compensated by an increase in Verrucomicrobia and an alteration in the formula of Proteobacteria, with a decrease in Deltaproteobacteria matched by an increase in Alphaproteobacteria. After plant growth (PG2), Acidobacteria decreased significantly and were counterbalanced by the growth of Actinobacteria, Firmicutes, and Verrucomicrobia. Compared with C2, Acidobacteria also showed a significant reduction, confirming them as the most affected group, while Verrucomicrobia and Actinobacteria increased and Proteobacteria were less affected. When compared with G2, Verrucomicrobia and Acidobacteria decreased and Proteobacteria increased. In summary, the combination of pethoxamid and glyphosate mainly affected the Acidobacteria population.
The MSG field experienced changes between MSG1 and MSG2, with a decrease in Proteobacteria, compensated by various increases in other major phyla. The comparison between MSG1 and C1 revealed that the early effects led to a decrease in Acidobacteria and an increase in Verrucomicrobia and Proteobacteria, suggesting that this combined action mainly affects Acidobacteria. A comparison between MSG2 and C2 showed a decrease in Proteobacteria, with Deltaproteobacteria and Gammaproteobacteria being most affected, counterbalanced mainly by increases in Verrucomicrobia. Compared with the G field, the main difference between MSG1 and G1 consisted of a greater reduction in Acidobacteria, whereas at the final sampling time Acidobacteria showed a heightened presence.
The use of a conventional treatment for non-glyphosatetolerant maize (field AIM) shows that between the first sampling time (AIM1) and after plant growth (AIM2) there was a very large decrease in Acidobacteria. A comparison with C1 revealed decreases in Actinobacteria, Proteobacteria, and other minor phyla, counterbalanced by an increase in Verrucomicrobia and Acidobacteria. At the final time (AIM2 vs. C2), Acidobacteria continued to experience a moderation accompanied by reductions in Actinobacteria, with Verrucomicrobia filling most of the gap.
- Phylogenetic Analyses
The results of the PCA are indicated in Fig. 2 . As expected, since the environmental properties were the same for all fields, the first principal component (PC1), explaining a 32.95% variation, is clearly associated with sampling time, with samples grouped separately from those collected at the final time; the second component (PC2), explaining a 19.08% variation, separates the samples in two parts: at the first sampling time one part groups together soils PG1 and ATG1, and the other part groups all the remaining soils, while at the final sampling time, the soils are spread across the second principal component, with C2 at one extreme and the remaining soils progressively separating from it in the following order of increasing distance: MSG2, G2, ATG2, PG2, and AIM2.
PPT Slide
Lager Image
UniFrac Principal Components Analysis of rhizobacterial communities from glyphosate-tolerant Bt-maize subject to different herbicide combinations. Abbreviations are used to identify the treatments applied to each sample (C: control; G: glyphosate; PG: pethoxamid and glyphosate; MSG: mesotrione plus s-metolachlor and glyphosate; ATG: acetochlor plus terbuthylazine and glyphosate; AIM: aclonifen plus isoxaflutole and mesotrione), followed by a number identifying the sampling time (1: soon after emergence; 2: late before harvest). Samples with similar bacterial communities are localized in similar positions in the diagram. Dispersion across the first component, P1 (explaining 32.95% of the variation), clearly groups samples by collection time, while dispersion across the second component, P2 (19.08% of the variation), shows how samples group at each sampling time in relation to the control samples C1 and C2. Proximity to C2 reflects the degree of recovery of the communities at the end of the experiment.
- Hierarchical Clustering Analysis
The hierarchical clustering tree obtained using the Unifrac metric with jackknife validation is shown in Fig. 3 , with nodes labeled according to their statistical support. Jackknifing values represent a combination of support for the branching event and the maximum-likelihood distance. The statistical validation further supports the above observations by separating samples according to collection time (statistical support of 100%) in agreement with PC1, thereby helping to classify the impact of the treatments according to their proximity to the control (C). At the first sampling time, the PG1 and ATG1 grouping gained a support of 100%, as did its clustering to the group formed by C1 and G1, and the grouping of their supercluster ((PG1, ATG), (C1, G1)) to MSG1. Hence, the low jackknife fraction for AIM1 was likely due to the uncertainty about the corresponding phylogenetic distance.
PPT Slide
Lager Image
Hierarchical clustering tree of the rhizobacterial communities of glyphosate-tolerant Bt-maize subject to different herbicide combination treatments. The communities recovered from soils treated with various herbicides (C: control; G: glyphosate; PG: pethoxamid and glyphosate; MSG: mesotrione plus s-metolachlor and glyphosate; ATG: acetochlor plus terbuthylazine and glyphosate; AIM: aclonifen plus isoxaflutole and mesotrione) at two different sampling times (1: soon after emergence; 2: late before harvest) were clustered using UniFrac and subjected to jackknife statistical analysis. Branching events in the clustering tree have been tagged with their jackknife significance values according to the scale shown. Clustering separates samples first according to sampling time, and then according to the treatment applied. Community resilience to treatment is reflected in the proximity to the control at the second sampling time (C2).
Samples collected at the second sampling time were further divided into two major groups (statistical support of 98%), one grouping C2 with the soils treated with G2 and MSG2, and the other clustering the rest. ATG2 and AIM2 were clustered together in 100% of the permutations. PG2 maintained an intermediate position, with the high support for the other two groupings strongly sustaining its location in the tree.
- Taxonomy-Independent Alpha Diversity Analyses
The values of observed OTUs, Chao1, and Shannon index are indicated in Table 3 , and the rarefaction curves of OTUs and Chao1 at the final sampling time are provided as Fig. S1.
OTU, Chao1, and Shannon index values of rhizobacterial communities subject to various herbicide treatments.
PPT Slide
Lager Image
Richness estimation in the rhizobacterial communities of Bt-maize subject to different herbicide treatments. aSamples are labeled, using the initials of the treatments applied (C: control; G: glyphosate; PG: pethoxamid and glyphosate; MSG: mesotrione plus smetolachlor and glyphosate; ATG: acetochlor plus terbuthylazine and glyphosate; AIM: aclonifen plus isoxaflutole and mesotrione) followed by the sampling time (1: soon after emergence; 2: late before harvest). bOTU stands for observed taxonomic units. cChao1 provides an estimation of the actual number of OTUs in the sample. dThe Shannon index quantifies the entropy of the distribution and depends on the evenness of the proportional abundances of OTUs in the population. Values for the average and 95% confidence interval of the mean were computed using QIIME at a 97% similarity level and calculated from 50 replicas of 4,000 (first sampling time) and 30,000 (second sampling time) randomly selected sequence reads from each sample. Given the different sampling sizes, values are comparable within each sampling time, but not between different collection times.
Owing to the uneven number of reads recovered, we had to perform analyses of alpha diversity at the first and last sampling times separately, and therefore the values reported are only comparable within each sampling time but not among different sampling times. At the first sampling time, G1 and AIM1 show a decrease in their OTUs and Chao1 predicted values, while ATG1 appears significantly affected at its Chao1 predicted values.
Analysis of the values and rarefaction curves at the final sampling time obtained using more information shows that the treated soils display different degrees of resilience, as evidenced by the changes in the distance of their rarefaction curves with respect to the control.
More interestingly, evenness in the community’s diversity has been suggested to be a major factor in maintaining the effectiveness of a healthy rhizobacterial community. Diversity may be evaluated using the Shannon-Wiener index, which also considers relative species abundance. Owing to the uneven sample sizes, a limited number of reads was considered, and comparisons had to be made with caution. Here, only the most extreme differences are considered. Overall, most fields showed some convergence with the control at the end of the experiment, except for ATG2, whose Shannon index remained smaller than the control over time, pointing to a loss of rare clades, and for PG2 and AIM2, which showed the opposite trend (Shannon index larger than the control at the final time), pointing to a lower weight of abundant species.
As already mentioned, the use of maize cultivars producing Cry1Ab protein or tolerant to glyphosate has been shown to exert a minor or undetectable effect on the rhizobacterial community compared with conventional maize [4 , 5 , 7 , 8] . The potential impact of other confounding factors was addressed by keeping all variable factors constant and pooling subsamples to address local microheterogeneity, as described. No appreciable differences in microbial biomass, plant growth, conditions, or yield that could induce additional rhizobacterial community changes were observed in this experiment. Microbial biomass studies usually find difficulty spotting significant differences among various herbicide treatments [30] , implying that changes are compensated by differential species growth. Therefore, changes to the taxonomical composition observed in the sampling times were used to examine the progress of the rhizobacterial community.
The differences between the changes observed in the untreated soil (C) and those found in the previous studies were expected, as indicative of the different environmental conditions. Proteobacteria have been reported to be enriched normally in the maize rhizosphere [40] , which is in agreement with our findings.
The soil treated solely with glyphosate (G) provides an additional baseline to contrast the combination treatments. Glyphosate [21] is a widely-used, broad-spectrum, and low-toxicity herbicide often coupled to glyphosate-resistant crops. Glyphosate is non-mobile, highly soluble, and nonpersistent, which may explain why this soil revealed a reduction in alpha diversity mainly affecting Proteobacteria at the first sampling time with a full recovery of richness and diversity at the final sampling time, in agreement with previous observations in maize [17] , glyphosate-tolerant maize [3 , 7] , and cotton [6] .
The use of acetochlor plus terbuthylazine combined with the post-emergence treatment with glyphosate (ATG) after a comparison with the C and G samples confirms this treatment as being more distant from the control, hence exerting a greater impact that affects Proteobacteria to a wide extent. Acetochlor is moderately soluble and nonpersistent, and is suspected to have endocrine and carcinogenic effects in humans, amphibians, and fish [20] . Like racemic metolachlor, it has been reported to affect antioxidant enzymes in soil bacteria [36] . Residual acetochlor effects on the rhizosphere have recently been reported to be detectable soon after application (until day 15), but is very low afterwards [2] . Terbuthylazine has algaecide, herbicide, and microbiocide properties and is primarily adsorbed by organic matter in the soil, with the bacterial community playing a leading role in its degradation. It has been reported to significantly alter soil bacterial composition by increasing denitrifying bacteria in soils, possibly reducing the efficiency of fertilizers [31] . The reduced effect of glyphosate and acetochlor when used alone suggests that the toxic degradation products, low solubility, and moderate persistence of terbuthylazine could be responsible for the deep and lasting effect of this combination on the rhizobacterial community.
The combination of pethoxamid and glyphosate (PG) exerted a significant early impact and a relevant late impact on the community, although it seemed to be less aggressive than the treatment with acetochlor and terbuthylazine plus glyphosate. Pethoxamid was reported as a selective systemic herbicide [32] for maize and soya bean that is absorbed by the roots and young shoots. Since glyphosate is nonpersistent and pethoxamid is a moderately soluble nonpersistent herbicide, the presence of lasting effects suggests a synergy where the early effects observed may be dramatic enough to outlast the herbicide presence in the soil.
Pre-emergence treatment with s-metolachlor and mesotrione followed by post-emergence treatment with glyphosate (MSG) exerted a considerable early impact that was largely restored at the final sampling time. Mesotrione was developed to control broad leaf weeds in maize crops [29] and is reportedly degraded rapidly by soil microorganisms; maize is naturally tolerant to mesotrione with no yield penalty. Mesotrione has moderate solubility and persistence and can be degraded by soil bacteria ( Bacillus sp. ) [9] . The effect of mesotrione has been studied using polymerase chain reaction temporal temperature gradient gel electrophoresis 16S rDNA fingerprints [12] , showing no response from the microbial communities at field rate. The biological activity of s-metolachlor was discovered in 1970, and it is still widely used on account of its high efficiency [45] , although it is becoming less common owing to its bioaccumulation and potential carcinogenic, cytotoxic, and genotoxic effects. S-Metolachlor has been reported to influence Pseudomonas sp. metabolism and to exert high toxicity on microalgae species [45] , but is less persistent. The combined synergistic effects of mesotrione and s-metolachlor on rhizobacteria have been the subject of recent analysis by polymerase chain reaction denaturing gradient gel electrophoresis [30] , showing a synergistic effect equivalent to a 10-fold field rate. Our finer-grained analyses indicate that despite a strong initial effect, this combination shows good resilience by the convergence of its rhizobacterial community composition with that of the control at the end of the study.
The combination of aclonifen with isoxaflutole has been reported to exert no depressive effect on maize yield [39] . We have tested the impact of aclonifen plus isoxaflutole followed by mesotrione (AIM) on rhizobacteria. The observed effects might be related to the low solubility, moderate mobility, and low persistence of isoxaflutole and to the persistence and low solubility of aclonifen, which is known to affect carbon mineralization microorganisms in sand and loam and to affect bean yield via its effects on Rhizobium phaseolii [23] . Isoxaflutole is known to increase nutrient level reduction in soils [34] , and the effect of its combination with atrazine has been compared to that of glyphosate alone [26] without any major differences. Aclonifen is used in conjunction with isoxaflutole (an isoxazole) in pre-emergence application to inhibit the growth of grasses and broad leaf weeds.
Our findings in the treatment with a different combination (isoxafutole plus aclonifen followed by mesotrione) further suggest that the effect may be due mainly to the presence of aclonifen, and possibly, to a much lower extent, mesotrione and isoxaflutole.
Verrucomicrobia seems to be the phylum that is less affected, usually filling the void left by other phyla, and suggesting that it may have a relevant opportunistic or compensatory role in culture soil. Proteobacteria are mostly affected at the end of the ATG treatment and Acidobacteria at the end of the PG treatment. Changes observed at lower taxonomical levels, although potentially relevant [16] , are difficult to assess owing to the uneven sample sizes.
The extent to which the rhizobacterial composition from each sample differs at the final time is indicative of the long-term effects of each treatment on rhizobacterial composition. All the analyses performed (taxonomical, phylogenetic, and alpha and beta diversities) support the separation of samples into two main groups at the end of the experiment; one composed of samples C2, G2, and MSG2, and a second one consisting of samples PG2, AIM2, and ATG2, with PG2 taking an intermediate position. This separation is strongly supported by the statistical analyses.
In conclusion, from all the treatments considered, the use of glyphosate alone to treat glyphosate-tolerant maize seems the less aggressive one for the rhizobacterial community. The combination of the pre-emergence treatment with mesotrione and s-metolachlor followed by the postemergence treatment with glyphosate shows a good final resilience. The effect of pethoxamide and glyphosate displays lasting effects on rhizobacteria and seems less aggressive than using a treatment suitable for conventional non-glyphosate-tolerant maize with aclonifen and isoxaflutole followed by mesotrione. The treatment with acetochlor plus terbuthylazine in pre-emergence, followed by glyphosate in post-emergence, exerts a great impact, which is in agreement with previous observations. We must emphasize, however, that this study has been conducted in a single location over a one-year cultivation period. Our observations support the need for more in-depth analyses of soil microbial properties (such as functional microarrays) and for spreading the analyses to other locations and over several years of cultivation, so as to accurately measure the cumulative effects of the repeated application of each treatment, leading to sounder recommendations for longterm agricultural practices.
The Spanish Ministry of the Environment and Rural and Marine affairs has commissioned and supported this research (Grant No. EGO22008). Additional support was provided by the Ibero-American Programme for Science, Technology and Development (CYTED: 510RT0391) and the EU 7FP COST Action (BM1006). The permission to carry out the field studies was given by the Spanish National Commission on Biosafety. The owner of the land granted permission to conduct the studies on the selected site. The funders played no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Altschul SF , Madden TL , Schäffer AA , Zhang J , Zhang Z , Miller W , Lipman DJ 1997 Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25 3389 - 3402    DOI : 10.1093/nar/25.17.3389
Bai Z , Xu HJ , He HB , Zheng LC , Zhang XD 2013 Alterations of microbial populations and composition in the rhizosphere and bulk soil as affected by residual acetochlor. Environ. Sci. Pollut. Res. 20 369 - 379    DOI : 10.1007/s11356-012-1061-3
Barriuso J , Marín S , Mellado RP 2010 Effect of the herbicide glyphosate on glyphosate-tolerant maize rhizobacterial communities: a comparison with pre-emergency applied herbicide consisting of a combination of acetochlor and terbuthylazine. Environ. Microbiol. 12 1021 - 1030    DOI : 10.1111/j.1462-2920.2009.02146.x
Barriuso J , Marín S , Mellado RP 2011 Potential accumulative effect of the herbicide glyphosate on glyphosate-tolerant maize rhizobacterial communities over a three-year cultivation period. PLoS One 6 e27558 -    DOI : 10.1371/journal.pone.0027558
Barriuso J , Valverde JR , Mellado RP 2011 Effect of the herbicide glyphosate on the culturable fraction of glyphosatetolerant maize rhizobacterial communities using two different growth media. Microbes Environ. 26 332 - 338    DOI : 10.1264/jsme2.ME11137
Barriuso J , Mellado RP 2012 Glyphosate affects the rhizobacterial communities in glyphosate-tolerant cotton. Appl. Soil Ecol. 55 30 - 36    DOI : 10.1016/j.apsoil.2011.12.010
Barriuso J , Mellado RP 2012 Relative effect of glyphosate on glyphosate-tolerant maize rhizobacterial communities is not altered by soil properties. J. Microbiol. Biotechnol. 22 159 - 165    DOI : 10.4014/jmb.1107.07036
Barriuso J , Valverde JR , Mellado RP 2012 Effect of Cry1Ab protein on rhizobacterial communities of Bt-maize over a four-year cultivation period. PLoS One 7 e35481 -    DOI : 10.1371/journal.pone.0035481
Batisson I , Crouzet O , Besse-Hoggan P , Sancelme M , Mangot JF , Mallet C , Bohatier J 2009 Isolation and characterization of mesotrione-degradingBacillussp. from soil. Environ. Pollut. 157 1195 - 1201    DOI : 10.1016/j.envpol.2008.12.009
Blackshaw RE , O’Donovan JT , Harker KN , Clayton GW , Stougaart RN 2006 Reduced herbicide doses in field crops: a review. Weed Biol. Manage. 6 10 - 17    DOI : 10.1111/j.1445-6664.2006.00190.x
Caporaso JG , Kuczynskim J , Stombaugh J , Bittinger K , Bushman FD , Costello EK 2010 QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7 335 - 336    DOI : 10.1038/nmeth.f.303
Crouzet O , Batisson I , Besse-Hoggan P , Bonnemoy F , Bardot C , Poly F 2010 Response of soil microbial communities to the herbicide mesotrione: a dose-effect microcosm approach. Soil Biol. Biochem. 42 193 - 202    DOI : 10.1016/j.soilbio.2009.10.016
de-Bashan LE , Hernandez JP , Bashan Y 2012 The potential contribution of plant growth-promoting bacteria to reduce environmental degradation – A comprehensive evaluation. Appl. Soil Ecol. 61 171 - 189    DOI : 10.1016/j.apsoil.2011.09.003
DeLorenzo ME , Scott GI , Ross PE 2001 Toxicity of pesticides to aquatic microorganisms: a review. Environ. Toxicol. Chem. 20 84 - 98    DOI : 10.1002/etc.5620200108
DeSantis TZ , Hugenholtz P , Larsen N , Rojas M , Brodie EL , Keller K 2006 Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl. Environ. Microbiol. 72 5069 - 5072    DOI : 10.1128/AEM.03006-05
Dohrmann AB , Küting M , Jünemann D , Jaenicke S , Schlüter A , Tebbe CC 2013 Importance of rare taxa for bacterial diversity in the rhizosphere of Bt- and conventional maize varieties. ISME J. 7 37 - 49    DOI : 10.1038/ismej.2012.77
Duke SO , Lydon J , Koskinen WC , Moorman TB , Chaney RL , Hammerschmidt R 2012 Glyphosate effects on plant mineral nutrition, crop rhizosphere microbiota, and plant disease in glyphosate-resistant crops. J. Agric. Food Chem. 60 10375 - 10397    DOI : 10.1021/jf302436u
Edgar RC 2010 Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26 2460 - 2461    DOI : 10.1093/bioinformatics/btq461
El Fantroussi S , Verschuere L , Verstraete W , Top EM 1999 Effect of phenylurea herbicides on soil microbial communities estimated by analysis of 16S rRNA gene fingerprints and community-level physiological profiles. Appl. Environ. Microbiol. 65 982 - 988
Foley M , Sigler V , Gruden C 2008 A multiphasic characterization of the impact of the herbicide acetochlor on freshwater bacterial communities. ISME J. 2 56 - 66    DOI : 10.1038/ismej.2007.99
Franz JE , Mao MK , Silkorski JA 1997 Glyphosate: a unique global herbicide. ACS Monograph ISBN: 0-8412-3458-2.
Gantoli G , Rueda Ayala V , Gerhards R 2013 Determination of the critical period for weed control in corn. Weed Technol. 27 63 - 71    DOI : 10.1614/WT-D-12-00059.1
Gürsoy O , Padem H 2012 Influence of aclonifen on the growth ofRhizobium phaseoliiand the yield of green beans (Phaseolus vulgarisL.). Int. J. Plant Res. 2 195 - 198    DOI : 10.5923/j.plant.20120206.04
Hamady M , Lozupone C , Knight R 2009 Fast UniFrac: facilitating high-throughput phylogenetic analyses of microbial communities including analysis of pyrosequencing and PhyloChip data. ISME J. 4 17 - 27    DOI : 10.1038/ismej.2009.97
Hardoim P , Nissinen R , van Elsas JD 2012 Ecology of bacterial endophytes in sustainable agriculture,In: Bacteria in Agrobiology: Plant Probiotics. Springer Berlin Heidelberg 97 - 126
Hart MM , Powell JR , Gulden RH , Dunfield KE , Pauls KP , Swanton CJ 2009 Separating the effect of crop from herbicide on soil microbial communities in glyphosateresistant corn. Pedobiologia 52 253 - 262    DOI : 10.1016/j.pedobi.2008.10.005
Hodgson DA 2000 Primary metabolism and its control in streptomycetes: a most unusual group of bacteria. Adv. Microb. Physiol. 42 47 - 238
Huson DH , Mitra S , Ruscheweyh HJ , Weber N , Schuster SC 2011 Integrative analysis of environmental sequences using MEGAN4. Genome Res. 21 1552 - 1560    DOI : 10.1101/gr.120618.111
Johnson BC , Young BG , Matthews JL 2002 Effect of postemergence application rate and timing of mesotrione on corn (Zea mays) response and weed control. Weed Technol. 16 414 - 420    DOI : 10.1614/0890-037X(2002)016[0414:EOPARA]2.0.CO;2
Joly P , Besse-Hoggan P , Bonnemoy F , Batisson I , Bohatier J , Mallet C 2012 Impact of maize formulated herbicides mesotrione and s-metolachlor, applied alone and in mixture, on soil microbial communities. ISRN Ecol. Article ID 329898 2012
Kara EE , Arli M , Uygur V 2004 Effects of the herbicide Topogard on soil respiration, nitrification, and denitrification in potato-cultivated soils differing in pH. Biol. Fertil. Soils 39 474 - 478    DOI : 10.1007/s00374-004-0729-3
Kato S , Kitajima T , Okamoto H , Kobutani T Pethoxamid, a novel selective herbicide for maize and soybean In: The British Crop Protection Council Conference: Weeds United Kingdom 2001 23 - 28
Kielak AM , van Veen JA , Kowalchuk GA 2010 Comparative analysis of acidobacterial genomic fragments from terrestrial and aquatic metagenomic libraries, with emphasis on Acidobacteria Subdivision 6. Appl. Environ. Microbiol. 76 6769 - 6777    DOI : 10.1128/AEM.00343-10
Lin CH , Lerch RN , Garrett HE , Jordan D , George MF 2007 Ability of forage grasses exposed to atrazine and isoxaflutole to reduce nutrient levels in soils and shallow groundwater. Commun. Soil Sci. Plant Anal. 38 1119 - 1136    DOI : 10.1080/00103620701327976
Margulies M , Egholm M , Altman WE , Attiya S , Barder JS , Bemben LA 2005 Genome sequencing in microfabricated high-density picolitre reactors. Nature 437 376 - 380
Martins PF , Carvalho G , Gratao PL , Dourado MN , Pileggi M , Araujo WL , Azevedo RA 2011 Effects of the herbicides acetochlor and metolachlor on antioxidant enzymes in soil bacteria. Process Biochem. 46 1186 - 1195    DOI : 10.1016/j.procbio.2011.02.014
Nandula VK , Reddy KN , Duke SO , Poston DH 2005 Glyphosate-resistant weeds: current status and future outlook. Outlooks Pest Manage. 16 183 - 187    DOI : 10.1564/16aug11
Page ER , Cerrudo D , Westra P , Loux M , Smith K , Foresman C 2012 Why early season weed control is important in maize. Weed Sci. 60 423 - 430    DOI : 10.1614/WS-D-11-00183.1
Pocanam YA Journées Internationales sur la Lutte contre les Mauvaises Herbes 20ème Conférence du COLUMA Dijon, France 2007 699 - 705
Peiffer JA , Spor A , Koren O , Jin Z , Tringe SG , Dangl JL 2013 Diversity and heritability of the maize rhizosphere microbiome under field conditions. Proc. Natl. Acad. Sci. USA 110 6548 - 6553    DOI : 10.1073/pnas.1302837110
Pruesse E , Quast C , Knittel K , Fuchs BM , Ludwig W , Peplies J , Glöckner FO 2007 SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res. 35 7188 - 7196    DOI : 10.1093/nar/gkm864
Talbot AR , Shiaw MH , Huang JS , Yang SF , Goo TS , Wang SH 1991 Acute poisoning with a glyphosate-surfactant herbicide (‘Roundup’): a review of 93 cases. Hum. Exp. Toxicol. 10 1 - 8    DOI : 10.1177/096032719101000101
Tornisielo VL , Botelho RG , de Toledo Alves PA , Bonfleur EJ , Monteiro SH 2013 Pesticide tank mixes: an environmental point of view,InPrice AJ, Kelton JA (eds.).Herbicides - Current Research and Case Studies in Use. InTech 473 - 487
Valverde JR , Mellado RP 2013 Analysis of metagenomic data containing high biodiversity levels. PLoS One 8 e58118 -    DOI : 10.1371/journal.pone.0058118
Viegas CA , Costa C , André S , Viana P , Ribeiro R , Moreira-Santos M 2012 Does s-metolachlor affect the performance ofPseudomonassp. strain ADP as bioaugmentation bacterium for atrazine-contaminated soils? PLoS One 7 e37140 -    DOI : 10.1371/journal.pone.0037140