Habitat-host microbial associations across a gradient in land use 1 intensification in Southern Amazon 2

The conversion of natural habitats to agriculture results in the structural, physical and chemical 40 degradation of ecosystems; and microbial communities respond to change with shifts in diversity 41 and environmental function. Because the structure of amphibian-associated microbiome depends 42 partially on the dynamics of microorganisms in the habitat, we hypothesized that land use would 43 affect the tadpole skin microbiota structure of populations inhabiting water bodies in agricultural 44 lands. To study this, we sampled microbial communities of water bodies and skin of larval 45 amphibian across a gradient of land use intensification represented by cerrado, pastures and 46 soybean fields. We used 16S rRNA high-throughput gene sequencing to characterize the 47 microbial communities. Land use had a strong effect on both host and habitat bacterial 48 communities. Bacterial ASVs richness and diversity in water bodies decreased from pristine 49 habitat to soybean plantations, with the cerrado community differing from pasture and soybean 50 fields. The aquatic microbial community composition and structure were different across the 51 gradient, showing a robust effect of land use on this habitat. The richness and diversity of 52 amphibian-associated bacterial community was lowest in cerrado and highest in pasture 53 populations. The soybean plantation exhibited the most distinct composition and structure of 54 amphibian microbiota while the pasture and cerrado communities were similar. Bacterial ASVs, 55 candidates for biomarkers of the land use effect on both host and water bodies communities, were 56 indicated. Our results highlight the effects of land use intensification as a driver for amphibian 57 microbiome and offer information on the functionality of agro-industrial environments.


82
Many amphibians have characteristics that make them particularly sensitive to 83 environmental changes, which is why they are often proposed as indicators of habitat quality [1].

84
These vertebrates are characterized by a humid skin in which they shelter a microbial community 85 in a commensal relationship. Hosted microorganisms seem to benefit from the moisture and 86 nutrition produced by the host's dermal glands and help prevent colonization by foreign 87 microorganisms [2]. A general profile of amphibian skin microbiota is given by the host's 88 taxonomic position, but the environment also contributes to composition and maintenance [3].

95
Considerations about environmental influences on skin microbiota must include the 96 impact of ecosystem degradation because this process can modify pathogen-host interactions.

97
This may happen because the habitat degradation causes the loss of regulation of population 98 abundances, which is made by the local biological diversity [7].

99
For example, the conversion of native habitats into agricultural areas involves transforming areas 100 of diverse vegetation physiognomies, such as savanna and forest, into homogeneous herbaceous 101 counterparts. Land use alters the physical (abiotic) conditions of the environment to which 102 individual amphibians are exposed, expectedly influencing both host ecological performance, the 103 dominant microbial communities, and the interaction among these two components. Relevant 104 variables include, among others, incident solar radiation, temperature, dissolved oxygen, pH, 105 conductivity, water hardness, turbidity, nutrients, and pesticides [8][9][10]. Soil bacterial 106 communities clearly respond to changes in land use with shifts in composition and abundance 107 [11]. An important consequence of changes in the structure of soil microbial communities is the 108 alteration of their ecological functions, as carbon and nitrogen cycling [12]. However, little is 109 known about the consequences of such shifts in the environmental microbial communities on 110 vertebrates.

111
Soybean plantations and cattle pastures are widespread agro-industrial activities in Brazil, 112 and involve not only drastic land management but also agrochemical use and soil degradation [9].

113
Pasture has a lower environmental impact than soybean fields and the conversion of pastures to 114 soybean fields usually involves fire, tilling and liming [9]. Following conversion, yearly land 115 management involves sowing, applying fertilizers and pesticides. Although the response of the 116 environmental microbiome to the impact of forest-pasture and forest-soybean conversions is 117 4 poorly understood, it does seem that the implementation of pasture tends to homogenize soil 118 microbial communities [13]. The conversion of Amazon rainforest into large-scale soybean 119 plantation promotes changes in the abundance and composition of soil bacterial community [14].

120
The responses of amphibians to such changes could be anticipated as drastic; indeed, many 121 species and populations are intolerant of human-dominated landscapes. However, there are also 122 resilient species, able to thrive in harshly modified environments [15]. The Amazonian Arc of

123
Deforestation poses yet one additional question because it is rainfed. In other words, most land 124 management practices coincide with amphibian reproduction so that larval phases occur in 125 severely affected water bodies. Pesticides can affect larval amphibian physiology, behaviour, 126 development and survivorship [16].

127
The skin microbiota of amphibians is part of the environment-organism interface and a 128 recent review argues that the disruption of microbial diversity associated with vertebrates is a 129 "serious threat to wild populations" and should be recognized as an "essential component" of 130   premisses, this study investigates how a context of land use intensification influences microbial 139 communities associated to amphibians. We hypothesize a link between land use effects and 140 amphibian skin microbiota structure, and assume that such a link would be corroborated by effects 141 on the skin microbiota along a gradient of environmental degradation. We focused on land use in 142 soybean fields and pastures within the Cerrado domain, the original habitat [9] , and compared 143 the skin microbiota of tadpoles in water bodies along a land-use gradient. This is the first study 144 to test the effect of environmental changes on the amphibian host microbiome in a gradient of 145 land use intensification; it also provides information on the functionality of agro-industrial 146 environments linked to microbial communities associated with amphibians.

197
We decided to sample two pasturesat Canarana and Água Boabecause of the 198 biogeographic scenario of our study region, which is at the edge of the Amazon Forest. S.

199
fuscovarius is an open-area species that does not occur in closed-canopy forests. Therefore, the 200 appropriate reference condition for S. fuscovarius and associated microbiota is the cerrado, whose 201 northernmost limit starts ~20km south. Due to extensive land cover change in the region, a 202 preserved cerrado patch was found in Água Boa, which is ~100km south. Pastures are a land use 203 type that is common to both localities and therefore were used as a control for regional effects on

214
Sterile Whirl-Pak® bags (Nasco) were used to collect 1L of pond water at a depth of 0-215 10, immediately after the capture of tadpoles. Pond water was transported to the laboratory and 216 filtered a 0.20 μm pore cellulose acetate filter (Sartorius Stedim Biotech) using a vacuum pump.

217
Membranes were individually preserved in a GTE buffer solution and stored in a freezer at -20 ° 218 C. The time between the collection of water and its total filtration was no longer than three hours.

272
The Random Forest algorithm was used to identify the best predictors, i.e., ASVs, 273 explaining differences observed between groups (i.e., between habitat and host, and among land  (Table 1). By contrast, no differences among land uses were 289 observed in the phylogenetic diversity metric (P = 0.072; ANOVA).  (Table 2).

294
When analysing Canarana and Água Boa pastures separately (Table S1, Supplementary material), 295 microbial communities in cerrado and pasture water bodies were globally similar to each other 296 but distinct from those in soybean fields; bacterial communities from the two pastures were 297 different from each other in taxonomic (i.e., CoDA) but similar in phylogenetic (i.e., PhILR) 298 community composition and structure.

299
The random forest analysis indicated 7 bacterial phyla, from 20 ASVs, as the most 300 important predictors to distinguish the water body microbial communities among land uses. Those 301 with highest abundance in cerrado the cerrado were Acidobacteriota, Actinobacteriota, 9 Bacteroidota and Proteobacteria. The total abundance of these phyla was 4 and 2-fold their 303 abundance in pastures and the soybean field, respectively. Among them, Acidobacteriota was 304 absent in soybean fields, as predictor. By contrast, phyla Bdellovibrionota, Cyanobacteria and 305 Verrucomicrobiota were more abundant in soybean fields, corresponding to a total abundance 10 306 and 3-fold their abundances in pastures and the cerrado. Cyanobacteria was absent in cerrado and 307 pasture, as predictor (Figure 2a).

326
The 20 most important ASVs explaining the observed effects of land use on the S. 327 fuscovarius skin microbiota belonged to 6 bacterial phyla. Among these, Proteobacteria 328 predominated in the host microbiota of the cerrado population, with a 4 and 23-fold increase in 329 abundance relative to pastures and soybean fields, respectively. Phyla Bacteroidota,

330
Desulfobacterota and Cyanobacteria were more abundant in the soybean field. Together, these 331 phyla showed an 8 and 90-fold increase in abundance in soybean fields relative to pastures and 332 cerrado, respectively. Elusimicrobiota and Firmicutes were more abundant in the pasture, with 333 total abundances 9 and 80-fold higher than in the soybean field and the cerrado (Figure 2b). host microbiomes. We also report drastic results for water bodies, which suggests that impacts 348 may be stronger in aquatic microbial communities than host associated communities.

349
The literature reports that land transformation into agricultural systems often increases 350 the bacterial alpha diversity and changes the composition of soil microbiome [11]. We show a 351 reduction in water bacterial alpha diversity, in ponds and puddles of agricultural lands.

352
Importantly, the water bodies samples lacked sediments, so that results may be context specific.

353
Independently of this, we assume that changes in the diversity, composition and structure of 354 microbiota in water bodies relate to changes in the physicochemical and biological conditions 355 across environmental gradient. A previous study indicated that biological water bodies properties 356 as conductivity and turbidityand the communities of algae, tadpoles, predators and fish are 357 consistently affected by intensity of land use [10]. We observed differences in environmental 358 variables; soybean plantations water bodies had a distinct pattern compared to the water bodies 359 from pastures and cerrado. The latter two were similar for the same variables (Table S2,

361
The skin microbiota of S. fuscovarius larvae may be shaped by variables intrinsic to 362 individuals, including physiological traits affecting the skin milieu [40]. However, these traits 363 may modulate skin microbiota within the major microbiome shifts that seem induced by habitat 364 modification due to agricultural practices. Pastures display higher local diversity and richness of 365 amphibian bacterial community compared to soybean fields, which maintain the richness and 366 diversity typical of preserved Cerrado. However, regarding the composition and structure, this 367 pattern changes dramatically. We interpret these results as evidencing that soybean plantation 368 produces the greatest changes in the microbiota of larval S. fuscovarius. It is not yet possible to 369 generalize these findings, not even to compare with previous results, given the nature of our study 370 design. Yet, the study by Lammel et al. [41] shows that the intensity of land usebased on pH, 371 C, litter degradability, pesticides, and nutrient levelsdetermines the structure of soil bacterial 372 community, in which the soil microbial community of soybean plantation presents the most 373 distinct composition, compared to pasture and cerrado soil communities.

374
Exposure of the frog Lithobates pipiens to a broadspectrum sulfonamide antibiotic, 375 commonly used in livestock, and deposited in pasture soils and aquatic systems, does not alter 376 11 alpha diversity, but does change the composition of the skin microbiota [42]. Also, and as already 377 mentioned, the sediment of water bodies may be important. In aquatic systems, sediments act as 378 repositories for materials from anthropogenic activities [43] and are a likely source for tadpole 379 skin microbiota. Which could explain the parallel effects observed in soil [41] and amphibian 380 microbiota in this study.

381
A collateral finding in this study involves a set of bacterial ASVs that may constitute

402
In summary, the intensification of land use impacts environmental microbial 403 communities, as observed in this and some previous studies, and these findings seem extendable 404 to microbial communities associated with amphibians. In this context, the forest-soybean 405 conversion, relative to forest-pasture counterparts, has greater local impact for amphibian 406 microbial communities. We suggest that changes in habitat may directly affect the biology of 407 amphibian populations; and shifts in host physiology could also modulates its microbiome. We 408 do not address the consequences of observed changes in the amphibian skin microbiota, but we 409 assume functional changes and interactions with host, rooting from shifts in the microbial 410 communities. Therefore, environmental impact may change critical aspects of amphibian biology.

411
Also, the study of microbial communities can lead to valuable indicators of the viability of 412 ecosystems, agricultural productivity and human and animal health.       Obs.: Bonferroni-adjusted P-values for multiple comparisons.