Abstract
Crop residues are a crucial ecological niche with a major biological impact on agricultural ecosystems. In this study we used a combined diachronic and synchronic field experiment based on wheat-oilseed rape rotations to test the hypothesis that plant is a structuring factor of microbial communities in crop residues, and that this effect decreases over time with their likely progressive degradation and colonization by other microorganisms. We characterized an entire fungal and bacterial community associated with 150 wheat and oilseed rape residue samples at a plurennial scale by metabarcoding. The impact of plant species on the residue microbiota decreased over time and our data revealed turnover, with the replacement of oligotrophs, often plant-specific genera (such as pathogens) by copiotrophs, belonging to more generalist genera. Within a single cropping season, the plant-specific genera and species were gradually replaced by taxa that are likely to originate from the soil. These changes occurred more rapidly for bacteria than for fungi, known to degrade complex compounds. Overall, our findings suggest that crop residues constitute a key fully-fledged microbial ecosystem. Taking into account this ecosystem, that has been neglected for too long, is essential, not only to improve the quantitative management of residues, the presence of which can be detrimental to crop health, but also to identify groups of beneficial micro-organisms. Our findings are of particular importance, because the wheat-oilseed rape rotation, in which no-till practices are frequent, is particularly widespread in the European arable cropping systems.
Background
Crop residues are an essential living element of agricultural soils. Smil [1] stressed that they “should be seen not as wastes but as providers of essential environmental services, assuring the perpetuation of productive agrosystems”. When left in the field in the period between two successive crops, rather than being buried immediately, crop residues contribute to the formation of soil organic carbon, improve soil structure, prevent erosion, filter and retain water, reduce evaporation from the soil surface, and increase the diversity and activity of micro-organisms in the ground [2]. No-till practices are becoming increasingly widespread, as they take advantage of these attributes [3]. However, such practices are often considered likely to increase the risk of disease epidemics [4–6]. Indeed, several leaf-, stem-, head-, and fruit-infecting micro-organisms, classified as “residue-borne” or “stubble-borne” pathogens, are dependent on host residues for survival during the period between successive crops and for the production of inoculum for their next attack [7, 8]. The epidemiological contribution of residues as an effective source of inoculum is well-established but difficult to quantify [e.g. 9] and generalise, because the nature of survival structures depends on the biology of the species. The situation is rendered even more complex by the presence of several species reported to act as crop pathogens in plants as endophytes, without symptom development in the plant, and in the soil and plant residues as saprophytes. Taking into account the inoculum from stubble-borne pathogens and possible competition with other micro-organisms, it appears likely that the expression of a disease is the consequence of an imbalance between a potentially pathogenic species and the rest of the microbial community, rather than the consequence of the mere presence of this species [10].
Residues constitute a crucial ecological niche, not only for pathogenic species, but also for non-pathogenic and beneficial species. Residues can be viewed as both a fully-fledged matrix and a transient compartment, because they originate from the plant (temporal link), are in close contact with the soil (spatial link) and degrade over the following cropping season, at rates depending on the plant species, the cropping practices used [11], and the year (climate effect). It remains unknown whether the succession of microbial communities in residues is driven primarily by plant tissue degradation or edaphic factors [12]. Many studies have investigated the structure of the microbial communities present during the life cycle of the plant [e.g. 13–15], but few have investigated the microbiota associated with plant residues. Several ecological studies have investigated the impact of the residue compartment on the structure of soil microbial communities [2, 16–19], but not the impact of the soil compartment on structure of the residue communities. The detritusphere, defined as the part of the soil attached to residues [ 12, 20, 21], is the most extensive and broad hotspot of microbial life in the soil [22]. The residue compartment and the detritusphere are located in close physical proximity but are considered by microbiologists to be separate trophic and functional niches [23]. A description of the residue communities and the specific changes in these communities over time might, therefore, help agronomists to understand the impact of cropping practices on crop productivity. Fungi and bacteria play important roles in the degradation of plant tissues in debris (cellulose, hemicellulose, lignin), but the interactions between them within the microbial community remain unclear, due to the lack of information about their origins (air-borne, soil-borne or plant-borne), their individual functions and the drivers of community structure in residues.
Crop rotation induces changes in the composition of the soil microbial community and usually reduces pathogen pressure [e.g. 18]. For instance, wheat yields benefit from “break crops” such as oilseed rape or other non-host crops to break the life-cycle of wheat-specific pathogens [24]. We focused here on the wheat-oilseed rape rotation, one of the most widely used cropping systems in Europe. The areas under bread wheat and oilseed rape in France were 5.0 × 106 ha and 1.4 × 106 ha in 2017 [25], respectively. As oilseed rape usually recurs every three years in the rotation and is used almost systematically either directly before or directly after wheat, we estimate that this classical rotation is used on almost 4.2 × 106 ha every year. Half the area occupied by these two crops is now grown without tillage, with at least some of the residues of the preceding crop left on the soil [26]. The issue addressed here is thus directly relevant to more than 2 × 106 ha, or about one tenth of the total arable area in France.
In this study, we deliberately focused on crop residues as a neglected, transient, but fully-fledged half-plant/half-soil compartment without describing the soil microbial communities, considering that it has been already performed in several studies [e.g. 27, 28]. We tested the specific hypothesis that plant is a structuring factor of bacterial and fungal communities in residues, and that this effect decreases over time, as contact with the soil induce progressive colonization of residues by other microorganisms. Over the last few years, high-throughput metabarcoding has become an indispensable tool for studying the ecology of such complex microbial communities [29], partly due to the difficulties in isolating fungal and bacterial species and growing them in axenic conditions. We used this approach to describe and compare changes in the microbial community of wheat and oilseed-rape residues left on the soil surface of three cultivated fields during two cropping seasons. We investigated whether the three main determinants (plant species, cropping season, and rotation) of the diversity of fungal and bacterial communities affected the microbiota of crop residues.
Methods
Experimental design
Field plots and rotations
An extensive field experiment based on a wheat (W)-oilseed rape (O) rotation cropping system was carried out during the cropping seasons of 2015-2016 and 2016-2017 at the Grignon experimental station (Yvelines, France; 48°51’N, 1°58’E). This area is characterised by an oceanic climate (temperate, with no dry season and a warm summer). A combined diachronic and synchronic strategy [30] was used to investigate the dynamics of the residue microbial communities both over a two-year period on the same plot and along a chronosequence substituting spatial differences (three plots) for time differences. A first monoculture plot (WWW) was sown with the winter wheat cultivar Soissons. This plot had been cropped with wheat since 2007 and was used in previous epidemiological studies focusing on the impact of wheat debris on the development of Septoria tritici blotch [e.g. 31, 32, 33]. Two other plots were cropped with oilseed rape cv. Alpaga and wheat cv. Soissons in rotation (OWO, adjacent to the WWW plot, and WOW, located 400 m away; Fig. 1). The OWO and WWW plots are characterized by a silty clay loam soil, and plot WOW is characterized by a silty loam soil (Additional file 1: Table S1). The three plots were not tilled during the two cropping seasons. The wheat and oilseed rape residues were left on the soil surface after harvest and partially buried to a depth of 10 cm with a disc harrow 6 weeks later (late September). Crops were managed in a conventional way following local practices (nitrogen fertilization, insecticide and herbicide treatments). No fungicide was sprayed on the leaves during the study.
Residue sampling
Wheat and oilseed rape residues (150 samples) were collected over the two cropping seasons. The changes in the microbial communities during residue degradation were described on the basis of four sampling periods (October, December, February, and May; Additional file 1: Table S2). A supplementary sample was taken in July 2016, and a posteriori in July 2017, to characterise the plant microbiota before the residues came into contact with the soil. For each sampling period, twelve pieces of wheat residue or four pieces of oilseed rape residue were collected from five points in each plot, 20 m apart (Fig. 1).
DNA extraction
Residues were cut to take off remaining roots, washed to remove the soil and air-dried in laboratory conditions. They were then cut into small pieces, pooled in a 50 mL bowl and crushed with a Retsch™ Mixer Mill MM 400 for 60 seconds at 30 Hz in liquid nitrogen, in a zirconium oxide blender. The crushed powder was stored in 50 mL Falcon tubes at −80°C until DNA extraction. We transferred 40 mg of crushed residues to a 2.0 mL Eppendorf tube, which was stored to −80°C. Total environmental DNA (eDNA) was extracted according to the TriZol® Reagent protocol (Invitrogen, according to the manufacturer’s instructions). Two independent extractions were performed per sample, giving a total of 300 eDNA samples. The two extractions were considered to be technical replicates.
PCR and Illumina sequencing
Fungal and bacterial community profiles were estimated by amplifying ITS1 and the v4 region of the 16S rRNA gene, respectively. Amplifications were performed with the ITS1F/ITS2 [34] and 515f/806r [35] primers. All PCRs were run in a reaction volume of 50 µL, with 1x Qiagen Type-it Multiplex PCR Master Mix (Type-it® Microsatellite PCR kit Cat No./ID: 206243), 0.2 µM of each primer, 1x Q-solution® and 1 µL DNA (approximately 100 ng). The PCR mixture was heated at 95°C for 5 minutes and then subjected to 35 cycles of amplification [95°C (1 min), 60°C (1 min 30 s), 72°C (1 min)] and a final extension step at 72°C (10 min). PCR products were purified with Agencourt® AMPure® XP (Agencourt Bioscience Corp., Beverly, MA). A second round of amplification was performed with 5 µL of purified amplicons and primers containing the Illumina adapters and indexes. PCR mixtures were heated at 94°C for 1 min, and then subjected to 12 cycles of amplification [94°C (1 min), 55°C (1 min), 68°C (1 min)] and a final extension step at 68°C (10 min). PCR products were purified and quantified with Invitrogen QuantITTM PicoGreen®. Purified amplicons were pooled in equimolar concentrations, and the final concentration of the library was determined with the qPCR NGS library quantification kit (Agilent). Libraries were sequenced in five independent runs with MiSeq reagent kit v3 (600 cycles).
Sequence processing
Fastq files were processed with DADA2 v1.6.0 [36], using the parameters described in the workflow for “Big Data: Paired-end” [37]. The only modification made relative to this protocol was a change in the truncLen argument according to the quality of the sequencing run. Taxonomic affiliations for amplicon sequence variants (ASV) generated with DADA2 were assigned with a naive Bayesian classifier on the RDP trainset 14 [38] and the UNITE 7.1 database [39].
Only ASV detected in both technical replicates were conserved for further analyses [40], to ensure robustness. ASV classified as “Chloroplast”, “Anthophyta”, “Arthropoda”, “Cercozoa” or not classified at the phylum level were discarded from the datasets. The remaining ASV were normalised according to the proportion of reads within each sample [41].
Microbial community analyses
Microbial community profiles were obtained for 100 wheat residue samples and 50 oilseed rape residue samples. The diversity of each sample was estimated by calculating the Shannon index with the ggpubr package in R [42]. A Kruskal-Wallis test was performed to assess significant differences in residue diversity with time, between plants within a rotation and between cropping seasons. In cases of significant differences, Wilcoxon pairwise tests were performed to compare sampling periods. A Wilcoxon pairwise test was performed to assess the effects of “plant” and “plant within a rotation” on Shannon index for each cropping season. Divergences were considered significant if p < 0.05.
“Plant” (i.e. crop), “crop within a rotation” and “cropping season” effects on community composition were assessed by multidimensional scaling (MDS) on the Bray-Curtis dissimilarity index with the phyloseq package in R (version 1.22.3 [43]). The effects of plant, cropping season, sampling period and biological sample on community composition were assessed with PERMANOVA, using the Adonis function of the vegan R package (version 2.4-4 [44]). After the aggregation of ASV for each sampling condition “sampling period/cropping year * crop within a rotation”, the betapart R package [45] was used to determine whether temporal changes in community composition were due to turnover (i.e. replacement of ASV between two sampling periods) or nestedness (gain or loss of ASV between two sampling periods). The effect of the plant on the microbial communities associated with residues during degradation was also assessed with PERMANOVA on each sampling period, for each year.
The genus composition of fungal and bacterial communities was assessed with a cladogram based on genus names. Only genera observed in three biological samples harvested on the same plot were incorporated into the cladogram. A cladogram representing the number of ASV for each genus, read percentage, occurrence and distribution for each sample, was constructed with the Interactive Tree Of Live (iTOL [46]) online tool for phylogenetic trees.
To illustrate taxonomic changes over time, especially between plant-derived communities and communities involved later in the colonization of the residues, we focused on seasonal shifts (increase, decrease or stability) in the relative abundance of a selection of some fungal and bacterial genera and tested their statistical significance (Wilcoxon tests between sampling periods).
Results
The bacterial and fungal communities associated with wheat (W) and oilseed rape (O) crop residues were characterised on three plots: a wheat monoculture (WWW), and two oilseed rape-wheat rotation plots (WOW and OWO) (Fig. 1). We assessed the composition of these microbial communities four times per year, during two consecutive cropping seasons (in October, December, February and May). An additional time point (in July) was also included for identification of the micro-organisms present on the plant before contact with the soil (Additional file 1: Table S2). An analysis of raw sequence datasets for the 150 samples of wheat and oilseed rape residues collected over the two cropping seasons resulted in the grouping of 14,287,970 bacterial and 9,898,487 fungal reads into 2,726 bacterial and 1,189 fungal amplicon sequence variants (ASV). ASV not detected in both technical replicates (5.4% of bacterial reads and 1.5% of fungal reads; Additional file 1: Table S3) were removed from the datasets.
Alpha diversity
Fungal and bacterial diversity was influenced by cropping season
Diversity dynamics, assessed by calculating the Shannon index, differed between the two cropping seasons and between fungi and bacteria. It was influenced only slightly by the type (or the absence) of rotation (Fig. 2). Fungal diversity increased over time during the first cropping season, whereas the differences between the samples in the second year did not reflect a gradual increase. Bacterial diversity did not increase during the first cropping season, except for wheat residues in rotation (WOW). During the second year, diversity increased from December to May, for all conditions. The impact of climatic conditions during residue degradation (Additional file 1: Table S4) or differences in initial diversity on the plant before harvest may explain the less marked trends observed between the two cropping seasons.
Fungal and bacterial diversity are influenced by plant species and rotation
Oilseed rape residues supported less fungal diversity than wheat residues in 2015-2016, but not in 2016-2017 (Additional file 1: Table S6). The opposite trend was observed for bacteria: bacterial diversity in oilseed rape was significantly lower than that in wheat in 2016-2017, but there was no difference in bacterial diversity between the two crops in 2015-2016. In addition, the Shannon index was significantly higher in wheat grown in monoculture than in wheat grown in rotation for both years for fungi and in 2015-2016 for bacteria.
Comparison of microbial communities associated with residues
We analysed the effects of plant species, rotation, cropping season and sampling period on communities, using the Bray-Curtis index and PERMANOVA. Differences between sample replicates collected from the same plot during the same sampling period were not significant for bacterial or fungal communities (Table 1). Thus, there was remarkably little heterogeneity between the samples from the same plot, and the number of biological samples was, therefore, sufficient to assess differences due to the variables of interest (i.e. plant species, rotation, cropping season and sampling period).
The structure of bacterial and fungal communities is influenced by plant species and rotation
Oilseed rape and wheat residues presented different sets of ASV, for both bacterial and fungal communities (Fig. 3). Plant species was the main factor explaining differences between the communities, accounting for 22.7% of the variance for bacteria and 32.4% for fungi (Table 1). The effect of plant species on fungal community structure decreased over time, while the effect of plant species on bacterial community structure tended to increase between October and December (Additional file S1: Table S5). For wheat, the type of rotation (i.e. rotation or monoculture) accounted for 10.5% of the variance for fungal community composition and 6.6% of the variance for bacterial community composition (Table 1).
Community structures change over time
Cropping season was the main temporal factor underlying changes in community structure, accounting for 16.4% of the variance for bacteria and 12.5% of the variance for fungi (Table 1). Sampling period also had a significant impact on community composition, accounting for 17.2% of the variance for bacteria and 7.2% of the variance for fungi. Theoretically, changes in ASV composition result from turnover (replacement of ASV between two sampling periods) and nestedness (gain or loss of ASV between two sampling periods [45]). We found that the dissimilarity between sampling periods was smaller for bacterial than for fungal ASV structure. By breaking down the dissimilarity between sampling periods, we found that most of the changes in fungal and bacterial ASV structure were due to turnover (Additional file 1: Table S7). Furthermore, we found that nestedness had a greater impact on bacterial communities than on fungal communities.
Changes in communities, by genus
Community succession across the different sampling dates was explained largely by the turnover of ASV. We characterised potential taxonomic differences in communities over time by analysing wheat and oilseed rape residues separately. ASV were aggregated together at genus level, resulting in 84 fungal (Fig. 4) and 184 bacterial genera (Additional file 1: Fig. S1, S2) for wheat, and 63 fungal (Fig. 5) and 186 bacterial genera (Additional file 1: Fig. S3, S4) for oilseed rape. For both plant species, we identified genera that disappeared or displayed a significant decrease in relative abundance over time (Additional file: Fig. S5). Among these genera, some are known to be associated with plants, such as Alternaria, Acremomium [14, 47, 48], Cryptococcus [49], Sarocladium [50] and Cladosporium [13, 47–50].
Some of the fungal species detected on wheat, such as Oculimacula yallundae (all ASV of Oculimacula genera), Zymoseptoria tritici and Pyrenophora tritici-repentis, are known to be pathogenic. Some of the species detected on oilseed rape, such as Verticillium spp., Leptosphaeria maculans (= Plenodomus maculans) and Leptosphaeria biglobosa (= Plenodomus biglobosa), are also known to be pathogenic. Strikingly, L. maculans and L. biglobosa predominated over the other taxa. Verticillium longisporum, V. dahlia and V. albo-atrum were mostly detected during the second sampling year. As samples were collected in two different fields, it was not possible to determine whether the occurrence of Verticillium spp., a soil-borne pathogen complex causing Verticillium wilt [51], was affected more by year or by the soil contamination. Acremomium, Clonostachys and Alternaria genera, which have also been described as associated with plants [52], were detected in the early sampling periods (Additional file: Fig. S5). Their relative abundances decreased over time. Most of the genera that were not present at early sampling points and with relative abundances increasing over time (e.g. Coprinellus, Psathyrella, Torula, Tetracladium, and Exophiala) were common to wheat and oilseed rape residues. These genera can thus be considered as probably derived primarily from the surrounding soil.
For bacteria, the difference in the genera detected between the two plants species was less marked than for fungi, as 146 genera were common to wheat and oilseed rape residues. These 146 genera corresponded to the 98.7% most prevalent reads for wheat and 97.5% most prevalent genus reads for oilseed rape. Proteobacteria was the predominant phylum the first year. The most prevalent proteobacterial subgroup was Alphaproteobacteria, with a high prevalence of Rhizobiales and Sphingomonadales. Rhizobium and Neorhizobium, two major genera from Rhizobiales, decreased in abundance between October and May in both wheat and oilseed rape. Sphingomonadales genera were much more abundant on wheat than on oilseed rape, especially Sphingomonas. Bacteroidetes genera, including Pedobacter in particular, were frequently detected and their prevalences tended to be stable for oilseed rape residues, and to decrease for wheat residues. In parallel, an increase in Actinobacteria, particularly Nocarioides, was observed. Major differences between July and October were observed for oilseed rape, consistent with the beta-diversity analysis, in which the percentage dissimilarity between July and October was high, due to both species extinction and turnover. Gammaproteobacteria were highly abundant on oilseed rape in July. Their frequency then decreased rapidly from October to May, due largely to the decrease in Pseudomonas. In parallel, we observed an increase in the levels of Alphaproteobacteria, especially Rhizobium and Sphingomonas, between July and October. A small decrease in levels of Gammaproteobacteria was observed between July and October for wheat in rotation, whereas the percentage of reads associated with this class increased between July and December for wheat in monoculture, due largely to the decrease in Pantoea and Enterobacteria. The abundance of Bacteroidetes, especially Pedobacter and Flavobacterium, also increased between July and October.
Discussion
Most studies on crop residues have focused on their impact on soil microbial communities [16], and the rare studies investigating the impact of soil on residue communities focused exclusively on bacteria [27, 28] or fungi [53]. Most of these studies were conducted on residues from a single year. Bastian et al. [12] established an extensive description of the species present in the soil, detritusphere and wheat residues, using sterilised residues and soil in a microcosm. In this study, we showed, under natural conditions, that three main factors (plant species, cropping season, rotation) simultaneously influence the composition of both fungal and bacterial communities present on residues. This study is the first to investigate the total fungal and bacterial communities associated with wheat and oilseed rape residues by a metabarcoding approach over two consecutive years. The very low variability of the communities for the five replicates is remarkable and shows that our strategy would be appropriate for comparing the effects of different treatments on microbial communities.
Crop residues should be viewed as a shifting platform for microbial meeting strongly affected by plant species
Oilseed rape and wheat residues contained different sets of micro-organisms before soil contact and during the firsts sampling dates after harvest. Similar results were previously obtained for the bacterial communities of buried crop residues [28]. Consistent with the findings of this previous study, the divergence between wheat and oilseed rape bacterial communities was probably due to differences in the chemical compounds present in the plants. The rapid change in the community observed at early stages of residue degradation for oilseed rape may be explained by the modification of simple compounds (sugars, starch, etc.), whereas wheat is composed of more complex compounds (lignin) and is, therefore, broken down less quickly, resulting in a slower change in the microbial community [28]. Overall, the change in bacterial community composition highlights turnover between copiotrophs and oligotrophs. Although copiotrophy and oligotrophy are physiological traits, several attempts have been made to classify microorganisms as oligotrophs and copiotrophs based on phylogeny [54]. According to this generalization, bacterial and fungal taxa whose relative abundances are significantly decreased during succession belong mainly to copiotroph. These taxa include for instance Alternaria, Cladosporium, Massilia and Pseudomonas (Additional file: Fig. S5). In contrast, the relative abundances of oligotrophic taxa such as Coprinellus or Nocardiodes increased during residues degradation, which could be indicative of the superior abilities of these micro-organisms to degrade complex polymers.
The initial fungal communities were structured mostly by the presence of species originating from the plant, several of which were highly specialised on the host plant. These species were gradually replaced by more generalist species, which colonised the residues of both plants. Most of these generalists, such as Exophiala, Coprinellus and Torula, are known to be soil-born [55, 56], or involved in degradation, such as Coprionopsis [57]. The host-specific fungi identified in our study included a large number of ascomycetes known to be foliar pathogens (O. yallundae, Fusarium sp. and Gibberella sp., Z. tritici, P. tritici-repentis, Parastagonospora nodorum, Monographella nivalis, L. biglobosa and L. maculans). The lifestyles of some pathogens are well-documented, as for Z. tritici, P. tritici-repentis and L. maculans. The decrease with time in levels of Z. tritici and other pathogens in wheat residues contrasts with the persistence of L. maculans and L. biglobosa in oilseed rape residues. These three pathogens are all known to reproduce sexually on the residues of their host plant [31, 58], but the life cycle of L. maculans is characterised by systemic host colonisation through intracellular growth in xylem vessels [59], whereas the development of Z. tritici is localised and exclusively extracellular [60]. Oilseed rape residues thus provide L. maculans with greater protection than is provided to Z. tritici by wheat residues. This likely explains differences in the persistence of the two pathogens and in the temporal dynamics of ascospore release: over up to two years for L. maculans [61, 62] but only a few months for Z. tritici [31, 63]. The predominance of L. maculans on oilseed rape residues was not surprising given that the oilseed rape cultivar Alpaga is known to be susceptible to L. maculans, but the high abundance of L. biglobosa was much more remarkable. One surprising finding of our study was the constant association of L. maculans with L. biglobosa on residues. Indeed, L. biglobosa is known to be more associated with upper-stem lesions [64], and its presence in large amounts on residues has never before been reported.
Our findings are consistent with current epidemiological knowledge of emblematic wheat and oilseed rape diseases, but they highlight our lack of knowledge concerning the lifestyles of many other fungal pathogens present on residues. A key point to be taken into account is that the trophic status of many species known to be principally pathogenic or non-pathogenic is not definitive [65]. For instance, Alternaria infectoria is sometimes described as a pathogen of wheat [13, 66], sometimes as an endophyte [67], and has even been tested as a potential biocontrol agent against Fusarium pseudograminearum on wheat [68]. Crop residues, half-plant/half-soil, should be the focus of future studies aiming to disentangle the succession of microbial species with different lifestyles and to characterise their relative impacts on the development of currently minor, but potentially threatening diseases.
The residue microbiota should be analysed in a dynamic manner, both within and between years
The results of our study highlight the importance of conducting multi-year studies focusing on ecological dynamics both within and between years in natural conditions. Year had a strong effect on both bacterial and fungal communities. Fluctuations of climatic conditions (temperature, rainfall, wind) have a major impact on pathogenesis (disease triangle concept [69]) and on the saprophytic survival of plant pathogens during interepidemic periods [70]. The two years of our study were marked by similar means of 10-day mean temperatures, but large differences in rainfall: mean 10-day cumulative rainfall in the first year was almost twice that in the second (Additional file 1: Table S7). The colonisation of residues by late colonisers may be affected by such climatic differences: in wheat, most prevalent degrading fungi (like Coprinellus, Psathyrella, Coprinopsis) were almost absent in the second year of the study. There was also considerable dissimilarity between the bacterial communities associated with each of the two years. For example, genus Enterobacter, which was highly abundant in the second year, was barely detectable in the first year.
Crop rotation has little impact on residue microbial communities
Oilseed rape is never grown in monoculture, so the effect of crop rotation was assessed only for wheat. The effect of rotation on residue microbial communities was much smaller than the effect of year (cropping season). It was more marked for fungi, for which diversity was greater in monoculture than in rotation. The use of a rotation may prevent the most strongly specialised species, in this case fungi, from becoming established, regardless of their pathogenicity. This finding is consistent with the greater development of some diseases in monoculture conditions, which promote the maintenance of pathogens through the local presence of primary inoculum. For instance, the presence of P. tritici-repentis, agent of tan spot disease, in the wheat monoculture plot and its absence from wheat-oilseed rape plots is consistent with epidemiological knowledge indicating that this disease can be controlled by leaving a sufficient interval between consecutive wheat crops in the same field [71].
Lesson to be learned from the residue microbial communities for the sustainable management of debris-borne diseases: a delicate balance between pathogenic and beneficial micro-organisms
The maintenance of crop residues at the surface of the cultivated soil increases the microbial diversity of the soil and, in some ways, helps to maintain good functional homeostasis [72]. However, conservation practices tend to increase the risk of foliar diseases [4–6]. Most disease management strategies focus on epidemic periods, during which the pathogen and its host are in direct contact. Interepidemic periods are also crucial for pathogen development, although during these periods the primary inoculum is not directly in contact with the new crop whilst not present in the field. Indeed, by carrying the sexual reproduction of several fungal pathogens, residues contribute to the generation and transmission of new virulent isolates potentially overcoming resistance genes, during monocyclic epidemics, as described for oilseed rape canker caused by L. maculans [73], but also polycyclic epidemics, as described for Septoria tritici blotch caused by Z. tritici [74].
However, the results of our study suggest that residues should not only be considered as a substrate for pathogens and a potential source of inoculum. Indeed, we detected several fungi identified as beneficial or even biocontrol agents in previous studies, such as Clonostachys rosea, Aureobasidium pullulans, Chaetomium globosum and Cryptococcus spp.. C. rosea, which was detected in both oilseed rape and wheat residues, has been reported to limit the sexual and asexual reproduction of Didymella rabiei on chickpea residues by mycoparasitism [75]. It has also been reported to be effective against Fusarium culmorum on wheat plants, through antibiosis during the epidemic period [76], and on wheat residues, through antagonism during the interepidemic period [77]. Cladosporium spp., which were abundant in our study, have also been reported to inhibit the development of P. tritici-repentis on wheat plants [78] and of Fusarium spp. on wheat residues [77]. The presence of these fungal species on wheat and oilseed rape residues is of potential interest for future analyses of interactions. Due to the use of a low-resolution marker for bacterial characterisation, we were unable to identify similarly the bacteria potentially interacting with pathogenic fungi. For instance, the presence of Pseudomonas spp. suggests possible interactions both with other microbial species and with the host plant [79], but the nature of the potential interactions is indeterminate: species of the Pseudomonas fluorescens group are known to be beneficial to plants, whereas Pseudomonas syringae and Pseudomonas aeruginosa are known to be pathogens of plants and even humans.
Although our study reveals the presence of genera or species reported in the literature as biocontrol agents, it has not yet shown any interaction between them and the pathogens. This experimental study (sampling effort, residue treatments, etc.) was not designed to characterize such interactions. A strategy involving the inference of microbial interaction networks from metabarcoding datasets might help to identify the species beneficial against pathogens, through competition, antagonism or parasitism. This however requires a more analytical, comparative experimental approach, that goes beyond the only description of shifts in natural communities composition: for example, using different “treatments” in a broad sense (e.g. artificial inoculation with a species or a group of species, change of biotic or abiotic environmental conditions, etc.) in order to modify interaction networks and so highlight the impact of some groups of micro-organisms on the whole community or a given species.
Conclusion
This study shows that crop residues, which can be seen as half-plant/half-soil transient compartment, constitute a pivotal fully-fledged microbial ecosystem that has received much less attention than the phyllosphere and rhizosphere to date. This study therefore fills a gap in knowledge of the communities present on crop residues under natural conditions. It confirms that the microbiote of crop residues should be taken into account in the management of residue-borne diseases. Taking into account this ecosystem is essential, not only to improve the quantitative management of crop residues, but also to identify groups of beneficial micro-organisms naturally present. The beneficial elements of the microbial community should be preserved, or even selected, characterised and used as biological control agents against the pathogens that complete their life cycle on the residues. These results are particularly important in that wheat-oilseed rape rotations are among the most widespread arable cropping systems in France and Europe.
Funding
This study was supported by a grant from the European Union Horizon Framework 2020 Program (EMPHASIS Project, Grant Agreement no. 634179) covering the 2015-2019 period.
Availability of data and materials
The raw sequencing data is available from the European Nucleotide Archive (ENA) under the study accession PRJEB27255 (Sample SAMEA4723701 to SAMEA4724326). We provide the command-line script for data analysis and all necessary input files as Additional File 2.
Authors’ contributions
LK, FS, VL, MHB, MB conceived the study, participated in its design, and wrote the manuscript. LK conducted the experiments and analysed the data. FS and VL supervised the project. All authors read and approved the final manuscript.
Ethics approval and consent to participate
Not applicable
Consent for publication
Not applicable
Competing interests
The authors declare that they have no competing interests.
Acknowledgements
This study was performed in collaboration with the GeT core facility, Toulouse, France (http://get.genotoul.fr) and was supported by France Génomique National Infrastructure, funded as part of “Investissement d’avenir” program managed by Agence Nationale pour la Recherche (contract ANR-10-INBS-09). We thank Martial Briand (INRA, UMR IRHS) and Dr. Gautier Richard (INRA, UMR IGEPP) for assistance with bioinformatic analyses, and Dr. Thierry Rouxel (INRA, UMR BIOGER) for improving and clarifying this manuscript. We thank Julie Sappa for her help correcting our English.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵