Banana (Musa acuminata) is an economically important crop in the Canary Islands (Spain), where it represents a source of income for many small holders, with around 50% of cultivated areas and yields that supply around 70% of national consumption. Dwarf Cavendish (Pequeña Enana, AAA) and derived selections represent the most widespread genotype (>90% of plants) due to its productivity and commercial performance. The crop is cultivated in small parcels, with alkaline volcanic soils often proceeding from different sites, amended with organic matter in part originating from old leaves left rotting on the soil surface. The main adversities of this cultivation include nematode pests, banana weevil (Cosmopolites sordidus), and Fusarium wilt caused by Fusarium oxysporum f. sp. cubense (Foc). The latter disease affects 2–12% of plants, with some farms reaching up to 30% prevalence (Gómez-Lama Cabanás et al., 2021). Pest and Foc management include a conventional approach, based on nematicides and insecticides, pheromone traps or products based on cloripirifos, spirodiclofen, and other active components. Alternative organic approaches for pest and disease management include the use of natural products with nematocidal or fungicidal properties, combined with pheromone traps (Supplementary Table 1).

PPN found in the Canary Islands banana farms include lesion nematodes Pratylenchus spp., spiral nematodes Helicotylenchus spp., and occasionally root-knot nematodes, Meloidogyne spp. These parasites are widespread in banana cultivated areas of the world (Moens et al., 2006; Coyne et al., 2018). Fusarium wilt is the most severe soil-borne vascular disease encountered in the Canary Islands. Its virulence depends on the race as well as on plant tolerance level (Bubici et al., 2019; Gómez-Lama Cabanás et al., 2021). The races reported in the Canary Islands are the subtropical races 4 (SR4), R2, and R1 (vegetative compatibility group 0120), highly virulent on cv. Gros Michel but not on Grand Naine (Domínguez et al., 2001). The low virulence of those races appears to be related to suppressive soil factors, such as the scarcity of iron and high levels of sodium and clay (Domínguez et al., 2001).

Plant and soil microbiota have an important effect on crop productivity. They include species with a varying degree of specialization, underpinning fundamental services such as nutrient recycling and pest or disease regulation. Is it recognized that many species, including BCA, contribute significantly to crop production by sustaining the rhizosphere and plant health (Bulgarelli et al., 2013; Granzow et al., 2017; Jacoby et al., 2017; Berg et al., 2020). The rhizosphere microbiota may hence represent a natural reservoir of BCA and a possible alternative tool for PPN management (Berg et al., 2017). However, due to the complexity of the rhizosphere environment, natural pest/disease regulation in intensive crops is not as frequent as expected. This often occurs in the case of PPN in intensive cropping systems (Topalović and Heuer, 2019; Topalović et al., 2020). The severity of PPN attacks mostly depends on the changes induced in soil by agriculture, such as the environmental pressure of monocultures and/or the low genetic diversity of crops. These conditions occur, and are particularly evident, in banana crops, in which all plants are usually clones of one or few lineages (mostly Cavendish), selected for commercial or technical reasons including resistance to one or more Foc diseases.

Knowledge of soil and rhizosphere microbial diversity and composition is hence fundamental to implement sustainable crop management, as well as to determine the impact of one or more biological/technical factors on the indigenous BCA. Progress has been achieved in determining the links between Fusarium wilt and banana-associated microorganisms (Effendi et al., 2019; Kaushal et al., 2020; Gómez-Lama Cabanás et al., 2021; Ravi et al., 2021). However, the potential of soil microorganisms to manage plant production in a more sustainable way, relying solely on indigenous species, is not yet fully exploited. Intensive banana crop management still remains highly dependent on frequent applications of synthetic pesticides. In particular, it is unknown if and how co-occurring microbial species, as well as their interactions with the BCA present, act on soil pests such as PPN. Soil-inhabiting organisms form a complex food web system, whose final outcomes may range from the natural regulation of noxious organisms to the insurgence of severe pest attacks (Bardgett et al., 1999; Ram et al., 2008).

Soil microbial communities, including BCA with other cryptic co-occurring species, may reveal undetected but useful interactions with the farm and the local agroenvironment (i.e., soil physical properties and pests), or depending on other external variables (i.e., selection of cropping practices and climate). In this regard, belowground links among species and trophic groups are becoming more and more informative as a huge amount of data is made available through deep sequencing and -omics technologies (Berg et al., 2020; Martínez Arbas et al., 2021). A detailed knowledge of soil microbial profiles may allow the setup of information-based crop management practices, exploiting natural pest/disease regulation, reinforcing or sustaining soil, and rhizosphere health.

Beneficial microorganisms that contribute to plant production include species acting as generalists or, on the contrary, inhabiting specific trophic niches, ranging from endophytes to pathogens or BCA, which affect plant health and thus farm productivity (Bulgarelli et al., 2013; Xue et al., 2015; Granzow et al., 2017; Kaushal et al., 2020; Zhang et al., 2020; Gómez-Lama Cabanás et al., 2021; Ravi et al., 2021). Apart from plant growth promoters, soil harbors endosymbionts or BCA often specifically associated with their hosts. Some isolates are already exploited as active ingredients in a number of commercial products, worldwide. These include, for example, host-specific PPN antagonists such as the bacteria Pasteuria spp. or Bacillus spp. (Tian et al., 2007; Mohan et al., 2020), and ubiquitous fungi such as Pochonia chlamydosporia, a root endophyte and a parasite of nematode eggs (Manzanilla-López et al., 2013). As a root endophyte, P. chlamydosporia also elicits the expression of a wide range of plant defense genes (Larriba et al., 2015; Mingot-Ureta et al., 2020; Tolba et al., 2021; Zhuang et al., 2021).

Other fungi of interest for exploitation as BCA are the members of genera Metarhizium and Trichoderma. The former, phylogenetically close to Pochonia, is characterized by the root endophytism and soil insect parasitism. Within the genus Trichoderma, several BCA and/or endophytes are present, including its teleomorph genus Hypocrea (Sivan and Chet, 1986; Chaverri and Samuels, 2002). However, data are needed on the competence and persistence of these BCA in the soil and rhizosphere, in particular when they are naturally present, as well as on their interactions with associated bacteria.

A further factor impacting the rhizosphere environment, and thus natural pest or disease regulation, originates from anthropic activities including, i.e., practices adopted by producers through organic or conventional farming, and the related aboveground cover, which also affect belowground microbiota profiles (Leff et al., 2018). Few data are available on the effect of organic and low-impact agriculture, soil properties, climate, on the microbiota diversity and, indirectly, on BCA species distribution, prevalence, and co-occurrence.

Considering the important role played by soil microorganisms in crop production and pest regulation, the focus of this study was to evaluate the interactions of the banana crop with the rhizosphere microbiota and the indigenous BCA of nematodes present in the soil, also to evaluate additional variables. In this study, we produced and analyzed metabarcoding data to: (1) determine the effect of cropping on rhizosphere bacteria and fungi in banana farms, in a subtropical environment, (2) investigate the effect of farm properties on rhizosphere microbial profiles, (3) estimate the impact that factors, such as climate or soil properties, and nematodes have on soil microbiota profiles, and (4) evaluate the crop effects on naturally occurring BCA species as well as their potential for PPN regulation.

A total of 38 samples were collected in Tenerife (Canary Islands, Spain) from the rhizosphere of M. acuminata var. Pequeña Enana or Gruesa (Cavendish clones), on farms applying conventional (only synthetic chemicals), integrated pest management (IPM, using chemical pesticides and natural products), or organic cropping techniques (EU regulation n. 848/2018) (Supplementary Table 1). A total of 36 samples were collected in February 2018 (mean month temperature = 18°C, humidity = 65%). Rhizosphere soil samples (ca. 300 ml) were collected at an average depth of 20 cm with banana root fragments. Other local control samples, with mostly grass or weed roots, were collected on the same farms from adjacent, banana root-free sites, within 5–10 m from the sampled banana plants, at the same depth. The three replicate banana samples and corresponding controls (each formed by three or more nearby subsamples, mixed to form a sample) were collected in six farms, located in the northern and southern areas of the island (Supplementary Table 1). Two other samples from banana rhizosphere, collected from a northern farm in September 2017 and stored at 4°C, were also included in this study. A subsample (300 g of mixed soil and roots) from each soil bag was stored at −80°C before processing for subsequent analyses. The remaining soil used for nematode and soil analysis was then stored at room temperature.

Samples were classified by latitude (northern or southern coast farm location), farm of origin and cultivation method, soil texture (measured by decanting and weighting of the three soil fractions), pH (Kettler et al., 2001; Schoeneberger et al., 2012) and number of PPNs, free-living, and predatory nematodes. The soil sieving and decanting technique was used for the extraction of nematodes by suspending a 200-ml soil subsample in tap water, followed by filtering and decanting, with a set of 500- and 75-μm sieves. The filtered suspension was then examined for nematode identification and density measurement with light microscopy, using a Hawksley counting chamber at 50 ×, in three replicates. PPN were identified at the species level using available taxonomic keys (Boag and Jairajpuri, 1985; Handoo and Golden, 1989; Uzma et al., 2015; see Table 1 for variables considered) with a Leitz Orthoplan light microscope, at 312–500 ×. Hand-picked nematode specimens were placed on slides in temporary water mounts. We classified the free-living (Rhabditida and Aphelenchoides) and predatory nematodes (Mononchida and non-plant parasitic Dorylaimida) during counts using a Hawksley counting chamber at 40-100 × and available nematode descriptions (Goodey, 1963). Free-living (bacterial and fungal feeders) and predatory nematodes (mononchids and dorylaims) were counted as groups. The remaining soil was then used to measure soil texture and pH, for each sample. Spearman's correlations among nematode and soil variables were calculated using PAST (Hammer et al., 2001).

Metabarcoding data showed a clear separation of microbial and fungal samples on the PCA and MDS planes, with clusterings associated with the presence/absence of banana roots, and latitude (Figures 2–4, 6). The representation of the samples hence appears indicative of both anthropogenic effects due to cultivation and, to a lesser extent, farm local climate, which affects a number of microbial taxa. The local climate depends on latitude, differing between northern areas (more humid and cooler, due to exposure to trade winds and ocean currents) and southern zones (arid and more exposed to Sahara heatwaves).

Both climate and aboveground cover have been consistently recognized as key factors influencing soil microbiota profiles. The effect of the aboveground cover on belowground microorganisms has been reported under different agroecological conditions, ranging from pot trials to natural ecosystems. Cropping affects soil microorganisms through many factors including cultivation regimes, plant cover and age, root exudates, and soil types (Berg and Smalla, 2009; Reinhold-Hurek et al., 2015; Compant et al., 2019; Delitte et al., 2021). Experimental data from assays carried out under controlled conditions (with faba bean and wheat plants) for example, showed an effect of the plant species on the soil fungal communities profiles, with differences related to the sample origin (bulk soil or rhizosphere). The tested cropping regimes (monocultures vs. intercroppings) also showed an effect of rhizosphere soil on bacterial diversity and richness (Granzow et al., 2017).

The differences observed between the banana and control samples were also reflected in the different fungal families (Wallemiaceae, Microascaceae, and Agaricaceae) and genera (unclassified member of Claroideoglomeraceae), which were more represented in the banana samples (Supplementary Table 6). These changes, including organic matter decomposers and an AMF, appear to be consistent with the introduction and cultivation of the plants. The mechanisms underpinning such an effect are likely related to the farming practices applied, including an enriched organic matter content in the first soil layers, due to the practice of covering the soil surface with banana decaying leaves, adopted locally by farmers. Fertilization, organic matter decomposition, root exudates, and irrigation are likely factors responsible for the observed changes in microbiota structures. Metabarcoding data on banana root endophytes, proceeding from the same environment, showed significant differences between mother plants and suckers, with Pseudomonas spp. as the most prevalent endophytic bacteria (Gómez-Lama Cabanás et al., 2021). The microbial community structure originally present in the soil is hence dependent on microenvironmental changes, shifting composition profiles from soil to rhizosphere, up to plants tissues, suggesting a progressive adaptation to the changes that characterize the colonized microhabitats.

The metabarcoding data also showed the presence of several known BCA species of nematodes or fungi. Microbial antagonists of nematodes have a potential to naturally regulate or even suppress PPN numbers, as well as to induce a defense reaction in plants (Oka, 2010; Liang et al., 2019; Topalović and Heuer, 2019). Although correlations may not be considered as a sufficient determinant of an antagonistic or cooperative/exclusion interaction, our data showed specific links of taxa with the variables considered. However, only a limited number of known nematophagous species was directly linked to PPN in the banana samples. Surprisingly, the co-occurrence analysis showed that, apart from the nematode trapping fungus A. oligospora, the bacterial and fungal taxa correlated to nematode densities did not include known nematophagous BCA species (Supplementary Table 8). The data instead indicated an effect of the nematode guilds as each considered group (two plant parasites, a free-living and a predatory species group) showed significant correlations with a specific subset of unique bacteria and fungi. In particular, the densities of both P. goodey and Helicotylenchus spp. showed positive correlations with different microbial species. The latter was correlated with a broader range of taxa, different from that of P. goodeyi, including, among others, the plant pathogenic fungus T. cucumeris (teleomorph of R. solani), with other unclassified and poorly studied bacterial lineages (i.e., C111, WD2101, SAR202, and Ellin6529). Moreover, a few significant negative correlations with fungi or bacteria were found for both PPN groups, that included Spiromastix spp., a genus of fungi producing antibacterial compounds (Niu et al., 2014), also inversely correlated to free-living nematodes (Supplementary Table 8). Negative correlations were found more frequently for free-living, microbiovorous nematodes, likely related to the short life cycle of microbiovorous species, to their direct exposure to fungi or bacteria and possibly to the release of microbial toxic compounds. In this case, a microbial antagonistic activity affects hosts and antagonists almost at the same time, yielding inverse relationships. As the PPN spend part of their cycle in the roots a delay usually occurs between the pathogen and nematode density shifts, affecting correlations. This may be also due to the PPN longer life cycle (4–5 weeks), and to the more complex dispersal and infection strategies of their antagonists.

Data thus indicate that the links between the nematophagous BCA species found and their target PPN hosts are more complex than expected, characterized also by a microbial trophic range and adaptive metabolic capability likely wider than known. An example of this complexity is given by OTUs classified in the genus Beauveria, a fungus usually associated with insects, whose higher frequency in banana samples and higher representation with the highest densities of free-living nematodes (together with other BCA of fungi such as Paraconiothyrium) (Supplementary Tables 4, 6) is worth further investigation. Despite suggestive of an association, direct assays are needed to establish a dependence of Beauveria spp. on nematodes, as both may be correlated to a further, unknown factor (i.e., increased invertebrate densities in the rhizosphere due to higher fertility levels or enhanced freeder roots development). Further assays are hence needed to clarify this aspect. Few data are, moreover, available in the literature for a Beauveria nematophagous activity, apart from an isolate of Beauveria bassiana obtained from cysts of Heterodera filipjevi and successfully tested as a BCA vs. the nematode juvenile stages, in controlled conditions (Zhang et al., 2020).

The data also suggest a probable association of the entomopathogenic fungus M. anisopliae with insects other than the banana weevil C. sordidus, which may represent a secondary host, as the fungus was more represented, with similar frequency levels, in the control samples. An inverse relationship of this fungus with soil pH was also found (Supplementary Table 4). A phylogenetically close species, the endophyte P. chlamydosporia, was also found in the banana rhizosphere samples, although with a low frequency (Supplementary Table 4). As an endophyte, this species is capable to colonize banana roots, inducing plant growth promotion (Mingot-Ureta et al., 2020). The fungus is also a nematode egg parasite, and is known to activate defense-related genes in endophytically colonized roots (Manzanilla-López et al., 2013), an effect also observed in banana aboveground tissues (Tolba et al., 2021). This fungus may also produce secondary metabolites with insecticidal activities (Lacatena et al., 2019).

The nematode BCA data also showed a number of Trichoderma spp., which were found with a higher frequency in the banana samples (Supplementary Table 4). Several Trichoderma spp. have been reported as an efficient BCA of nematodes, including Trichoderma harzianum (Sharon et al., 2001; Sahebani and Hadavi, 2008; Fan et al., 2020). However, no correlation was found for Trichoderma spp. with PPN among the samples examined (Supplementary Table 8). Moreover, an effective role of Trichoderma spp. against plant pathogens, i.e., Fusarium spp., has been reported for many hosts, including Musa spp. (Sivan and Chet, 1986; Thangavelu and Gopi, 2015; Chaves et al., 2016; Bunbury-Blanchette and Walker, 2019).

The densities of P. goodeyi and Helicotylenchus spp. in soil varied largely among samples. The endoparasite P. goodeyi was only found in the northern banana fields, with a maximum density of 1,100 nematodes × 100 cc soil⁻¹. Helicotylenchus spp. were found in 90% of the northern samples and <10% in the southern samples, with a maximum of 2,300 individuals × 100 cc soil⁻¹ (Supplementary Table 2; Supplementary Figure 1). Both nematodes are severe parasites of Musa spp., and may reach high densities in roots (Ssango et al., 2004; Roderick et al., 2012). Due to the low amounts of roots collected and required for metabarcoding analyses, nematode densities were preferably assessed in the soil. Density levels appeared, however, sufficient to sustain a severe root infestation, and compatible with the field data reported on other Musa cvs from other regions (Talwana et al., 2000; Aguirre et al., 2016). The presence of several BCA and the observed PPN densities appear indicative of complex rhizosphere interactions, based on the BCA host preference, polyphagy, and other interactions that may affect nematode regulation.

The metabarcoding analysis also showed the occurrence of antagonists of fungi, including A. luteoalbus and unclassified species of phyla Rozellomycota and Zygomycota (Supplementary Figures 2A,B). The former has been reported as an antagonist of Alternaria, Fusarium, and Phytophthora spp. (Lv et al., 2019). It was reported as a causal agent of ginger rhizome rot (Moreira et al., 2013), as an antagonist of fungi and a mushroom pathogen (He et al., 2010; Zhang and Tang, 2015), and as an endophytic plant growth promoter (Khalmuratova et al., 2021). Rozellomycota include mostly unclassified zoosporic species characterized by thick-walled resting spores and obligate parasitism on protozoa and fungi (Doweld, 2013). They were found almost uniquely in the banana rhizosphere (Figures 7, 8). Although the biology of these taxa is still poorly investigated, a natural regulatory role may be assigned to members of these clades, which include endoparasitic species in fungi and other eukaryotes. Their density and frequency in banana samples may have been favored by the organic matter applied to plants and by irrigation, providing beneficial services in the rhizosphere such as carbon and nutrients recycling, as well as the regulation of fungal root pathogens.

Finally, the effect of cropping regimes on microbial community structures is considered to be more effective over a longer cultivation period, resulting more evident in older crops or plants (Granzow et al., 2017). Farm age, however, did not show significant differences in the sequence representations (%) of fungi, apart from Wallemia, a basidiomycete genus including species inhabiting highly osmotic environments including dry and hypersaline substrates (Zajc and Gunde-Cimerman, 2018), more represented in farms >40 years old (Supplementary Table 6).

The comparison of microbiota composition between banana rhizosphere samples and close, adjacent controls deprived of banana roots showed differences affecting both bacterial and fungal profiles, indicating an effect of cropping. When considering other variables, sample clusterings also reflected latitude effects for both bacteria and fungi. Metabarcoding data showed the occurrence of taxa reported as BCA of nematodes, as well as other endophytes, mycorrhizal species, and obligate endoparasitic taxa (i.e., Rozellomycota), almost only present in the banana samples. However, apart from a Dactylella spp., the nematophagous fungi did not show a strict association or correlation with the two PPN species found, P. goodeyi and Helicotylenchus spp. Instead, differences were found among the nematode guilds as each phytoparasitic, free-living, and predatory nematode group showed correlations with a specific and different subset of bacteria and fungi. Other factors considered, such as crop cultivation method and soil texture, showed differences in fungal sequence representations as a function of the different variables examined. They were mostly related to trophic specialization and specific biotic requirements or adaptations to a range of decomposers, beneficial endophytes, mycorrhizae, or BCA, as well as plant pathogens. In conclusion, the belowground bacterial and fungal microbiota profiles were affected by plants and latitude, and showed different links to the nematode taxa present. As the impact of microbial species depends not only on their relative but also on their absolute abundance, direct, and specific quantification measurements of the rhizosphere microbial loads for most differential taxa may result informative to exploit the nematode and BCA interactions detected.

The datasets produced in this study can be found in the following online repository: https://www.ncbi.nlm.nih.gov/, SRA BioProject PRJNA540248.

AC, MC, and LR planned and designed the research work. JL-C, MC, and LR performed local samplings. MC performed RNA extraction, bioinformatic work, and the production of sequence data sets. AC and MC interpreted and analyzed the data and wrote this manuscript. All authors critically reviewed, revised, and approved the final version of this manuscript.

Research funded by EU H2020 Project MUSA, Microbial uptakes for sustainable management of major banana pests and diseases, GA n. 727624.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.