Gut microbial community response to herbicide exposure in a ground beetle

Gut microbiota plays a key role in physiological processes of insects, including nutritional metabolism, development, immunity and detoxification. Environmental stressors such as herbicides, used to optimize and improve crop yields, may interfere with the mutualistic relationships causing negative consequences for the host health. Dinitroaniline herbicides, for example pendimethalin, are used worldwide in pre‐emergence application to control grass and some broadleaf weeds. They target microtubules to arrest cell division and inhibit the development of roots and shoots. Effects of a pendimethalin‐based herbicide were assessed on the gut microbial community of Pterostichus melas italicus Dejean, 1828 (Coleoptera, Carabidae). The exposure effect was tested in vivo by using a recommended field rate (4 L per ha, 330 gL−1 of active ingredient) and evaluating the variability of responses in 21 days, corresponding to the half‐life of pendimethalin. The 16S rRNA sequencing data showed that the gut lumen was dominated by Proteobacteria, Firmicutes, Fusobacteria, Tenericutes and Bacteroidetes. The exposure interfered with the bacterial community richness and diversity associated with the gut from 2 days after the treatment. The differential abundance analyses highlighted a shift involving Lactobacillaceae, Streptomycetaceae, Neisseriaceae, Ruminococcaceae and Enterobacteriaceae. An increase in species such as Enterobacter sp., Pseudomonas sp., Pantoea sp and Paracoccus sp. involved in the herbicide degradation was also recorded after 21 days of exposure. Phylogenetic investigation of communities by reconstruction of unobserved states (PICRUSt) analysis indicated that the exposure has effects on the most predicted functional categories of gut microbiota related to metabolic function including carbohydrate, amino acid and lipid metabolism. These results demonstrate that pendimethalin can impact microbial communities associated with generalist predators inhabiting croplands leading to severe implications for the species’ ecological role. Understanding the effects of herbicides such as pendimethalin on ground beetles may help to protect beneficial soil insects that have a crucial role in the ecosystem services.


| INTRODUC TI ON
The gut microbiota is a complex community of obligate or facultative symbiotic bacteria that profoundly influences the host's fitness and life span interacting with its biological (Engel & Moran, 2013) and behavioural (Hosokawa & Fukatsu, 2020;Yuval, 2017) traits.
Beneficial gut bacteria show a wide range in their degree of action on metabolic activities, including nutrition, xenobiotic detoxification (Engel & Moran, 2013;Itoh et al., 2018), and physiological processes of insects (tolerance and resistance to pathogens, modulation of innate immune responses and immune priming) (Dillon & Dillon, 2004;Engel & Moran, 2013;Futo et al. 2016). The diversity of bacterial communities varies according to the host species and its ecological niche (Bonilla-Rosso & Engel, 2018;Yun et al. 2014), developmental stage (Cini et al. 2020;Kim et al. 2017), diet and environmental conditions (Colman et al. 2012;Jones et al. 2018;Kolasa et al. 2019). This is the result of selection pressures that they impose on each other, promoting continuous co-adaptation (Oliver & Martinez, 2014).
The direct or indirect exposure to agrochemicals (insecticides and fungicides), used for pest control in conventional agriculture, can significantly affect the structure and function of the gut microbiome in beneficial insects (Kakumanu et al. 2016;Syromyatnikov et al. 2020) and humans through the trophic web (Yuan et al. 2019).
The alteration of the gut microbial climax community results in physiological disorders and has consequences on survival and disease susceptibility of the host (Botina et al. 2019;Xia et al. 2018;Zeng et al. 2020), compromising its ability to respond to environmental stressors. There is growing evidence that also herbicides have adverse effects on the gut microbiota. Recent toxicological studies have been linked alterations of the gut microbiota to the glyphosate exposure in honey bee Apis mellifera Linneus, 1758, (Blot et al. 2019;Dai et al. 2018;Motta et al. 2018)

and the Colorado potato beetle
Leptinotarsa decemlineata Say, 1824(Gómez-Gallego et al. 2020. The atrazine had been tested to reduce the gut microbial diversity of house (Culex pipiens) and tiger (Aedes albopictus) mosquitos (Juma et al. 2020).
Pendimethalin is a dinitroaniline herbicide used worldwide to control weeds. It inhibits the development of roots and shoots in seedlings, interrupting the polymerization of microtubules and arresting cell division (mitosis) (Rose et al. 2016;Vighi et al. 2017). Its half-life can vary according to weather conditions and soil pH, from 24-39 to 76-98 days in aerobic soil (Strandberg & Scott-Fordsmand, 2004;Vighi et al. 2017). The residual dose (10%-15%), tested to remain in the soil until 300-400 days after physical, chemical or microbiological transformations, is harmful to non-target species (Strandberg & Scott-Fordsmand, 2004). Previous studies have been highlighted the potential toxicity of the pendimethalin sublethal dose on non-target organisms at different levels of the trophic web in both terrestrial (Belden et al. 2005;Oliver et al. 2009;Traoré et al. 2018) and aquatic (Ahmad & Ahmad, 2016;Tab assum et al. 2015;Traoré et al. 2018;Vighi et al. 2017) environments. Furthermore, the persistence of pendimethalin residues in the environment significantly impacts on the soil bacterial (Strandberg & Scott-Fordsmand, 2004) and fungal (Roca et al. 2009) community richness reducing its growth (Kocárek et al. 2016;Singh et al. 2002) up to 61% starting from 25 days of treatment (Nayak et al. 1994) and consequently affecting the soil biodiversity and fertility.
Despite the ecological and agricultural interest of carabids, little is known about the composition and diversity of bacterial communities associated with their gut systems. Previous metagenomic analysis has been highlighted that the biodiversity of the gut bacterial communities depends on the feeding habits and habitat of the host (Kudo et al. 2019) and facilitates seed consumption in omnivorous species (Lundgren & Lehman, 2010;Schmid et al. 2014).
In this context, knowledge of the herbicide effects on the carabid microbiota is lacking. This study aimed to investigate effects that a commonly used pendimethalin-based commercial formulation have on the structure and potential function of the gut microbiome of beneficial species inhabiting the soil in agroecosystems. We choose as a model Pterostichus melas italicus Dejean, 1828 (Coleoptera, Carabidae), an eurytopic and thermophilous clay soil species. In Central and Southern Europe, this species inhabits pastures, open forests and forest edges, and agricultural lands where it acts as a natural enemy of pests including aphids, lepidopterans and dipterans (Sunderland, 2002). The experiment was designed to test in vivo a recommended field rate and evaluate the variability of responses in K E Y W O R D S agroecosystem, carabid beetles, dinitroaniline, ecotoxicology, microbiota, symbiosis a time of 21 days corresponding to the half-life of the pendimethalin.
We hypothesized exposure effects on the bacterial community associated with the gut of P. melas. We also expected the possible effects of PND on gut microbiota to vary over time. This study will make a significant contribution to optimizing the use of this agrochemical in the agroecosystems to reduce sublethal effects on non-target organisms.

| Species collection and treatments
Adults of P. melas (n = 90) were collected in an organic olive grove (39°59'27.56"N, 16°15'32.64"E, 1,202 m a. s. l. San Marco Argentano, Calabria, Southern Italy) in October 2019 by using in vivo pitfall traps (plastic jars 9 cm in diameter) containing fruit as an attractant. In the laboratory, the beetles were identified using a dichotomous key, separated by gender and kept in 5 L plastic boxes that were filled to a depth of 6 cm with soil from the capture site, held at 60% relative humidity (rh), had a natural photoperiod and were at room temperature. They were fed with mealworms and fruit (organic apples) ad libitum.
To evaluate exposure effects of a pendimethalin-based commercial formulation (PND; Activus EC, product n° HRB00858-39; active ingredient pendimethalin 330gL -1 ), males were exposed for 21 days to the recommended field rate (4 L per ha, for cereal and vegetable cultures) taking into account the pendimethalin half-life ranged from 24.4 to 34.4 days in sandy acid soil (Kocárek et al. 2016;Strandberg & Scott-Fordsmand, 2004). The experimental design included three control and six exposure plastic boxes (180.5 cm 2 ) filled with the clean sandy soil (pH 5 approximately) from the capture site.
The exposure was performed by spraying the PND solution (7.2 µl of Activus in 14 ml of distilled water) with a pipette onto the soil surface of each box of the treated groups to simulate the field exposure by contact with the contaminated soil. Boxes for control groups were sprayed with distilled water. Males (10 for each box) were introduced 15 min after the PND solution has been sprayed.
To remove the digestive tract (foregut, midgut and hindgut), five beetles for control and treated groups were randomly chosen at 2, 7 and 21 days after the initial exposure. Beetles were anaesthetized in a cold chamber at 0°C for 3 min, gently cleaned in 70% ethanol and dissected under a stereo microscope Zeiss using sterile equipment. The guts, removed from beetles, were individually stored in 2 ml microcentrifuge tubes containing absolute ethanol until DNA extraction.

| DNA isolation and sequencing
Library preparation and sequencing were performed at the DNA sequencing facility of the Life Sciences Department of Trieste University, Italy. Samples were preliminary washed with phosphate buffer (PBS) to remove the storage ethanol. Genomic DNA was extracted using the E.Z.N.A ® Soil DNA kit (Omega Bio-Tek) following the manufacturer's instructions. DNA quality and quantity were assessed with a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific). An extraction blank was performed as a control to monitor for contamination of environmental bacteria DNA. The extracted DNA was used as a template for the amplification of V4 hypervariable region of the 16S rRNA by PCR primers 515F (Caporaso et al. 2011), and a mix of 802R (Claesson et al. 2009) and 806R (Caporaso et al., 2011). Primers were tailed with two different GC rich se-

| Data analyses
The CLC Microbial Genomics Module as a part of the CLC Genomics Workbench 20.0 (QIAGEN Digital Insights, Aarhus, Denmark) was used to analyse alpha and beta diversity, and the composition of the bacterial community. Raw sequencing reads were imported into the CLC environment, and we perform quality control, primers and adapter sequences removal and minimum size cut-off of 150 bp. The OTUs were picked by mapping sequences against the SILVA 16S v132 97% database (Quast et al. 2013) at the same identity percentage to observe OTU at the species level. Next, the OTUs were aligned using multiple sequence comparison by log-expectation and constructing a 'maximum likelihood phylogenetic tree' followed by alpha and beta diversity analyses. We estimated the effect size and significance on beta diversity for grouping variables with pairwise permutational multivariate analysis of variance (PERMANOVA) (Anderson, 2001).
A principal coordinate analysis (PCoA) was carried out of the beta diversity distance matrices to better visualize similarities and differences. We choose the Bray-Curtis distance because it better depicts differences in OTUs representation between the different time points. Differential abundance analysis was performed modelling each OTU as a separate generalized linear model (GLM), where it is assumed that abundances follow a negative binomial distribution.
The Wald test was used to determine the significance of group pairs.
Finally, for the functional analysis of microbiota, we export the OTU table from the CLC Genomics Workbench environment, and we use PICRUST2 (Douglas et al. 2020) to infer the minimal biological pathways (Ye & Doak, 2009). Using STAMP software (Parks et al. 2014), two-sided Welch's t test with Benjamini-Hochberg multiple testing correction was performed to identify the metabolic pathways in the Level-3KEGG (Kyoto Encyclopedia of Genes and Genomes) database that were significantly different (q < 0.05) between groups.

| Sequencing data
The bacterial community in the gut of P. melas was analysed using 16S rRNA gene amplicon sequencing and produced a total of 3,278,520 reads with an average of 109,284.00 ± 12,572.53 reads per sample. Raw sequences (reads) were trimmed, and the remaining sequences were reference-based clustered against SILVA 16S v132 database with a 97% sequence similarity accounting for 1.480 OTUs and 1.167 de novo OTUs from the 30 assayed samples (5 beetles at each time point per control and treated groups, respectively) for a total of 2.647 OTUs. The mean number of reads in OTUs for all the guts was 72,956.00 ± 16,589.67 for control and 73,718.13 ± 10,936.65 for PND exposed samples. Rarefaction curves (Figure 1a,b) calculated for total OTU abundance aggregate at the family level and assessed using the maximum likelihood phylogeny analysis always reached the plateau indicating adequate sequencing deepness to analyse the majority of phylotypes in most of the samples.

| Effects of the PND exposure
To investigate potential changes in microbial community diversity between control and PND-treated groups, we used the relative abundance profiles obtained by 16S  and Tatumella (p < .05) was recorded in treated group at 2 days of exposure. We also checked changes in the relative abundance of the   Pendimethalin has been well known to reduce the microbial biochemical activity in the soil, acting on carbon dioxide evolution and dehydrogenase activity of enzymes responsible for the oxidation of organic compounds with effects on the pH level (Strandberg & Scott-Fordsmand, 2004). Our findings first showed that PND could affect the gut microbiota in insects. Adults of P. meals exposed to a recommended field rate of PND for 21 days hold a lower diversity and species richness of the bacterial community, while a positive shift of the alpha-diversity was observed in the control group over time. Moreover, variations of the OTU abundance at the family level were recorded between PND-treated and control groups at different time points. The relative abundances of Neisseriaceae (neutrophiles and aerobics) and Ruminococcaceae (acidophiles and anaerobics) reduce after 2d and 21d from the initial exposure to PND, respectively.

| D ISCUSS I ON
The exposure to PND for 7d positively acted on the abundances of facultative aerobic Lactobacillaceae and Streptomycetaceae. In Lactobacillaceae, the increase is related mainly to the abundance of Lactobacillus, involved in the sugar fermentation leading to lactic and acetic acid and CO 2 production, as observed in Apis mellifera (Vásquez et al. 2012). Streptomycetaceae secretes a variety of enzymes that hydrolyse complex macromolecules (i.e. degradationproducts of chitin), and the resulting compounds can frequently serve as carbon or nitrogen sources (Glaeser & Kämpfer, 2016). They have also been found to produce antifungal compounds in a mutualistic association with termites (Chouvenc et al. 2018 The genus Pragia decreased in PND-treated beetles after 7 days of exposure. This genus belongs to a relatively small group of hydrogen sulphide-producing enterobacteria, including also Budvicia spp. and Leminorella spp., and it contains only one species, P. fontium, a free-living bacterium isolated from an environment that under anaerobic conditions utilizes monosaccharides and their derivatives but lacks fatty acid degradation pathways (Snopková et al. 2017). The genus Budvicia has been identified in the red palm weevil (Tagliavia et al. 2014), while Pragia was previously found to be associated with the gut of Poecilus chalcites (Lehman et al. 2009). Other shifts were observed for genera involved in different metabolic pathways such as pyrimidine (Alkanindiges and Empedobacter), aromatic compound (Comamonas) and amino acid (Pseudoxanthomonas) metabolism, hydrogen production (Clostridium sensu stricto 13) (Rosenberg et al. 2013) and detoxification of NO by its conversion to nitrate (Vitreoscilla) (Stark et al. 2012).
Pendimethalin, acting as a microtubule polymerization inhibitor, has no direct adverse effects on bacteria where the tubulin system is absent. Nevertheless, our results showed a variation of the microbiota structure in beetles exposed to the herbicide likely related to the physiochemical changes that occur during the treatment. In insects, the digestive system is part of a network of reactions and physiological processes that includes respiratory, circulatory, excretory and metabolic functions, so alterations in any of the network components cause changes in the other ones (Chapman, 2012). Previous studies have been indicated PND to have detrimental effects on physiological processes of metazoan organisms. In vertebrates, PND sublethal effects have been measured in the zebrafish Danio rerio Hamilton, 1882 (Park et al. 2021) and the teleost Clarias batrachus (Linnaeus 1758)(Gupta & F I G U R E 2 Box plots of microbiome alpha diversity metrics at the family level presented for P. melas males from control (a) and pendimethalin-treated (b) groups at days 2, 7, 21 from the initial exposure. An asterisk indicates significant differences based on Kruskal-Wallis test (p < .05) [Colour figure can be viewed at wileyonlinelibrary.com] F I G U R E 3 Similarity analysis of microbial communities. All principal coordinates analysis (PCoA) were based on weighted UniFrac distances (Bray-Curtis distance) showing the distribution of the bacterial community composition in P. melas males exposed to pendimethalinbased herbicide and control ones at days 2 (a), 7 (b) and 21 (c) after the start of treatment. Percentages on the axes represent the proportion of explained variation of each component of the PCoA [Colour figure can be viewed at wileyonlinelibrary.com] Verma, 2020). It has been demonstrated to cause mortality in the wasp Tiphia vernalis Rohwer 1924 (Oliver et al. 2009), disruption of the springtail Folsomia candida Willem 1902 reproduction, reduction in the earthworm Eisenia fetida (Savigny, 1826) growth (Belden et al. 2005) and negatively affect the humoral and cellular immune response in the ground beetle Harpalus rufipes (De Geer,

F I G U R E 4
Composition of gut microbiota in males of P. melas. The relative abundance of major taxa at the family (a) and genus (b) level in control and pendimethalin-treated beetles at different time points. Taxa with sequence abundance <1% of total sequences were pooled together as 'Other' in all the taxonomic ranks and reported in Table S1  are well known to be a good marker in ecotoxicological studies . Although we did not evaluate this in our study, we speculated that the exposure to PND interferes with the network of acid-base reactions, which together contribute to the organism's homeostasis (Harrison, 2001), modifying functional and morphological conditions of the gut and affecting the microbiota in P. melas. Indeed, using level 3 KEGG predictions, differences in the functional potentials of the bacterial communities were also observed mainly after 2d from the initial exposure.
Gut bacteria such as Enterobacter spp., Pseudomonas spp., Pantoea spp. are also found to assist in the detoxification processes of potentially toxic compounds in insect exposed to pesticides (Douglas, 2015;Itoh et al. 2018;Kucuk, 2020). The nitroreduction has been indicated as the initial degradation and detoxification step for pendimethalin (Ni et al. 2016). Numerous pendimethalin-degrading microorganisms have been isolated in the environment (Elsayed & El-Nady, 2013;Strandberg & Scott-Fordsmand, 2004)  Moreover, a previous study on the ground beetle Harpalus rufipes has been revealed a reduction in circulating phenoloxidase levels after exposure to PND field rate ). In addition, gut microbiota takes part in endocrine system regulation, being involved in development and growth processes, as observed for honey bees and red palm weevil Zheng et al. 2017). Thus, further studies are needed to investigate the effects of this herbicide on gut microbiota related to immune responses and the regulation of hormones controlling moulting, pupation, metabolism and reproduction in insects.
The association and interaction between bacteria have a crucial role in host homeostasis due to the non-pathogenic insect-associated microorganisms involved in a range of functions such as nutritional processes (digestion, provisioning and assimilation), development and pathogen resistance (Douglas, 2015;Engel & Moran, 2013 abundance of other bacteria in the gut of crickets (Douglas, 2015).
Moreover, modifications observed in the microbiota structure of P.melas exposed to PND likely cause a variation of the antagonistic interaction among different bacteria with different metabolic requirements, favouring the growth of facultative pathogens such as Pantoea and Empedobacter at 2d and 7d, respectively. Thus, we assume that the alterations in the microbiota community could lead to dysbiosis, compromising other microbiota-dependent life traits.
There is a growing need for studies that contribute to our understanding of the herbicide impact on the gut microbiome helpful to protect beneficial soil insects that have a crucial role in the ecosystem services. In this study, the changes in the microbiome structure may alter the community functions, and thus, fallouts can occur for beetles' physiology and behaviour. In generalist predators such as P. melas, bacteria enter the gut by horizontal transmission from the surrounding environment (Kolasa et al. 2019;Kudo et al. 2019).
The reduction in soil microbiota is well known in pendimethalin field application, and the exposure of this carabid to PND by contact or ingesting contaminated food caused modifications of microbiota structure and related functions. The PND exposure may indirectly affect microbiota because of alterations of gut physical, chemical or structural conditions. Such modifications could bring out a niche competition among different bacterial species for nutrient sources, changing their colonization ability. This interference with consolidated symbiotic relationships may have effects on different lifehistory traits, compromising the ecological role of P. melas as pest control, such as feeding behaviour, reproduction as well as the capability to withstand the colonization of the gut by non-indigenous species, including pathogens and therefore prevent infections. Thus, our results contribute to evaluating the risk assessment of herbicides such as pendimethalin on soil invertebrates in agroecosystems.

ACK N OWLED G EM ENTS
Authors thank Dr Giampiero Ventura for allowing the access to the sampled fields. The Italian Ministry of Education, University and Research (MIUR) (grant n ° UA.00.2014.EX60) supported this research.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N
AG, VML and AP conceived research. AG and VML conduced field sampling and exposure assays. FG conducted sequencing experiments. AP conduced statistical analyses. AG and MLV wrote the manuscript. AG secured funding. All authors read and approved the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The raw reads from 16S rRNA gene sequencing were deposited into the Zenodo public repository at http://doi.org/10.5281/zenodo.4663948 . The data that support the findings of this study are available in the supplementary material of this article (Table S1).