The microbiome at the interface between environmental stress and animal health: an example from the most threatened vertebrate group

Abstract

Nitrate pollution and global warming are ubiquitous stressors likely to interact and affect the health and survival of wildlife, particularly aquatic ectotherms. Animal health is largely influenced by its microbiome (commensal/symbiotic microorganisms), which responds to such stressors. We used a crossed experimental design including three nitrate levels and five temperature regimes to investigate their interactive and individual effects on an aquatic ectotherm, the European common frog. We associated health biomarkers in larvae with changes in gut bacteria diversity and composition. Larvae experienced higher stress levels and lower body condition under high temperatures and nitrate exposure. Developmental rate increased with temperature but decreased with nitrate pollution. Alterations in bacteria composition but not diversity are likely to correlate with the observed outcomes in larvae health. Leucine degradation decreased at higher temperatures corroborating accelerated development, nitrate degradation increased with nitrate level corroborating reduced body condition and an increase in lysine biosynthesis may have helped larvae deal with the combined effects of both stressors. These results reinforce the importance of associating traditional health biomarkers with underlying microbiome changes. Therefore, we urge studies to investigate the effects of environmental stressors on microbiome composition and consequences for host health in a world threatened by biodiversity loss.

1. Introduction

Humans have been altering the planet in various ways [1–3], leading to population declines and extinction of many species [3,4]. Therefore, species are often exposed to multiple stressors simultaneously [5], and their possible interactions produce outcomes that are hard to predict, as these can be antagonistic, additive or synergistic [6]. Exposure to a stressor may induce physiological responses that increase tolerance [7,8] or reduce exposure to others [9]. On the other hand, the simultaneous effects of different stressors may add to or amplify negative effects on health [7,10].

Global warming is an ongoing change [11] that makes increased temperatures a ubiquitous stressor, with special relevance for ectotherms [12,13]. Higher temperatures increase the metabolic demand of ectotherms and are likely to reduce their ability to supply the required amount of oxygen to their tissues (decreased aerobic scope [14]) leading to consequences that extend from the individual to the ecosystem [15]. Climatic changes may also increase the toxicity of pollutants and impair the ability of organisms to deal with them [16], because of the dependency of all physiological rates on temperature [17,18]. Among a great variety of chemical pollutants, nitrate is of special relevance as a contaminant in many aquatic habitats worldwide due to its widespread use in fertilizers and elevated concentrations in sewage effluent discharge [19,20]. The decreased aerobic scope imposed on ectotherms by higher temperatures can be aggravated by nitrate pollution because acquired nitrate is endogenously degraded into nitrite, which oxidizes haemoglobin into methaemoglobin, a non-carrier oxygen form that leads to hypoxia [21]. As a result, ectotherms may experience lower locomotory performance [8,22], decreased growth rates [23] and mortality [7]. However, studies addressing the interactive effects of increased temperatures and nitrate pollution on ectotherms have found conflicting results [8,21,24] indicating that responses are stage dependent and species specific.

The effects of stressors on animal populations have been addressed using different biomarkers, such as body condition, production of stress-associated hormones, haematological parameters, growth and survivorship, among others [7,21,22,24,25]. However, the role of their microbiomes (i.e. mutualistic/commensal microorganisms living in their bodies) in the observed results is still relatively little explored despite the broad influence of the microbiome on animal physiology [26]. This knowledge is particularly lacking when it comes to ectothermic organisms [27]. Microbiomes can adapt to environmental conditions faced by the host [28] and are largely influenced by the host’s behaviours and environment [29,30]. Microbial communities are responsive to environmental changes and their compositional changes may improve host performance and ability to deal with varied sources of stress, from parasite infection [31] to heat stress [12] and environmental pollution [32,33]. On the other hand, microbiomes with reduced diversity have been associated with decreased host ability to deal with stress in several instances [34].

Temperature increases have been shown to alter the microbiome of ectotherms with potential benefits [35] or negative effects on the host [34,36]. It has been largely acknowledged that endosymbionts influence thermal tolerance of their hosts [34,35]. Pathways related to amino acid metabolism are enriched in hosts with more diverse microbiomes, which might supply their hosts with bacterial-derived amino acids as important energy sources to be used while facing thermic stress [12]. However, microbiomes leading to improved ability to deal with thermic stress may also bring associated costs, such as reduced population growth under non-stressful conditions, which has been demonstrated for both invertebrate [37] and vertebrate hosts [34]. Pollution is expected to alter microbe composition in the environment, restraining microbiome acquisition by hosts [38]. It also affects microbiome composition and beta diversity with consequent metabolic disturbances to the host [23,39]. Nitrate particularly causes histological alterations in the gut, thus affecting amino acid and fatty acid metabolism as a result of deficient nutrient absorption [23]. On the other hand, nitrate pollution induced alterations in microbiome composition in larvae of the toad Bufo gargarizans resulting in increased nitrate metabolism with upregulation of l-lysine and l-tryptophan levels [23]. As these amino acids are known to be involved in protein synthesis, cell proliferation and growth, their adjusted levels are likely to prevent further damage in larval health, although more studies are needed to determine their real effect [23].

Amphibians are the vertebrate group most profoundly affected by the deleterious consequences of human impacts [40], with climate change and environmental pollutants figuring among the main factors affecting mortality and recruitment in amphibian populations [41,42]. Besides easily absorbing pollutants through their permeable skin, amphibians with an aquatic larval stage are especially sensitive to environmental stress because they are exposed to detrimental alterations in both aquatic and terrestrial habitats [43]. Amphibian larvae may have growth rates and time to metamorphosis altered under stressful conditions [44,45], and delayed metamorphosis may impair reproduction and recruitment [41]. In this study, we aimed at investigating the effects of increasing temperatures resulting from global warming and nitrate pollution on larvae of Rana temporaria and their microbiome. We focused specifically on the interactive effect of these ubiquitous stressors on animal health and the potential relation with the induced changes in their gut bacteria.

We raised R. temporaria larvae fed ad libitum under different temperature regimes and nitrate concentrations and assessed their release of corticosterone (CORT, a biomarker for physiological stress) [46], their body condition (as a surrogate for their ability to store energy under the imposed environmental conditions), their developmental rate, their gut bacteria diversity and composition. We hypothesized that CORT release would increase under higher temperatures and nitrate concentrations (i.e. higher stress levels). Considering that the co-evolution of hosts and their microbiomes is expected to result in mutual benefits [28], we expected that exposure of the holobiont to stressful conditions would result in microbiome adjustments likely to mediate the maintenance of host body conditions. We used body condition instead of size (growth) to express larval performance because we previously observed R. temporaria larvae to show compensatory growth under stress [44]. If they exhibited compensatory growth due to stress, that should result in lower energy reserves and body condition. We also assessed developmental rates to determine whether better body condition might be achieved at the cost of slower development (and, consequently, longer exposure to nitrate). Finally, we investigated the interactive effects between increased temperatures and nitrate levels on the microbiome bacteria and related such effects to the outcomes on larvae body condition and developmental rate. For this purpose, we compared microbiome composition among groups of larvae exposed to the different temperature and nitrate treatments to search for taxa of Eubacteria that had their abundances altered, and inferred what metabolic pathways they might be influencing.

2. Material and methods

(a). Study species and field sampling

Five clutches of the European common frog (R. temporaria) were collected at the locality Kleiwiesen (52.328 N, 10.582 E), a site in Braunschweig, Lower Saxony, Germany. The clutches were carefully removed from a pond with individual 12 l buckets (one per clutch) half filled with water from the pond and transported (for about 10 km) to the Zoological Institute of the Technische Universität Braunschweig.

(b). Experimental design

Eggs were kept at ambient temperature (14°C) until hatching in a ventilated room in five 12 l plastic buckets filled with 5 l of fresh pond water to allow for microbiome colonization of amphibian larvae guts until they reached developmental stage 25 (no external gills, free swimming larvae, sensu [47]). See electronic supplementary material for details on animal husbandry.

The experiment was conducted in a climate chamber (Kälte-Klimatechnik-Frauenstein GmbH, Germany) with a 14 : 10 h light:dark cycle and a mean (± s.d.) air temperature of 18 (± 0.1)°C, temperature within the range commonly experienced by R. temporaria larvae during its larval development in this region in Germany [48]. The temperature specification was also done for comparative purposes, as this temperature was used to represent the natural field mean temperature in other related studies with the species from this region [49]. At Gosner stage 25, three larvae from each clutch were randomly selected and allocated to 45 standard 12 l aquaria filled with 9 l of rested tap water. The aquaria were submitted to five different water temperature regimes: 18 ± 0.2°C, 22 ± 0.3°C, 26 ± 0.4°C and 28 ± 0.3°C, and a temperature fluctuation regime 17−26.5°C with a mean temperature of 21.1°C throughout 24 h (refer electronic supplementary material, figure S1 for daily temperature fluctuations) crossed with three levels of nitrate exposure (0 [control], 50 and 100 mg  l⁻¹) in a crossed experimental design using three replicates for each temperature × treatment combination (5 temperatures × 3 treatments × 3 replicates = 45 aquaria). We selected increments of 4°C as that is the limit of the predicted increments for Germany until 2100 in a high-emission warming scenario [50]. The fluctuating temperature regime was included aiming to achieve a mean temperature of 22°C (corresponding to the first 4°C increment in the constant temperatures) for comparative purposes. This treatment was added to include a realistic feature (temperature variability in natural habitats) in our study, which is advisable for more efficient application of the results for conservation purposes [24]. Each aquarium housed 15 larvae (15 larvae × 45 aquaria = 675 larvae in total; larval density: 1.66 larvae  per litre), to contribute to a more realistic scenario, given that the larvae occur in close proximity in natural microhabitats. The water was continuously aerated using air stones connected to aquaria pumps.

Different rearing temperatures in the aquaria were achieved by adjustable heating elements (JBL ProTemp S 25, 25 W, JBL GmbH & Co. KG, Germany) set to the respective rearing temperature. Nitrate concentrations were achieved using sodium nitrate stock solutions (NaNO₃) [51,52] (see electronic supplementary material for details). The selected nitrate concentrations fell within the environmental ranges observed in both surface and ground waters in Germany [53] and aligned with the environmental ranges measured in water bodies where amphibians breed [19,54].

Larvae were fed a diet consisting of 50% high-protein powdered fish food (Sera micron breeding feed for fish and amphibians; Sera, 52 518 Heinsberg, Germany) and 50% spirulina algae, offered in ad libitum rations twice a day to ensure that food was available in abundance. Dead or abnormal larvae were removed from the aquaria. The experiments ran for four weeks, all surviving larvae needed from 12 to 26 days to reach the terminal sampling point at developmental stage 38 [47]. Three larvae were randomly picked per aquarium for this study totalling 135 sample units (larvae), the remaining were used in other studies.

(c). Rana temporaria larvae CORT release and quantification

At stage 38 [47], we assessed the physiological stress levels of larvae through CORT release using the water-borne CORT (WB-CORT) collection method from Gabor et al. [55]. WB-CORT has been validated as an efficient stress biomarker for pro-metamorphic (developmental stages 37–39, sensu [47]) larvae of R. temporaria [46]. See electronic supplementary material for further details on CORT release and extraction.

We measured the CORT levels of samples corresponding to each individual with DetectX Corticosterone ELISA (Enzyme Immunoassay) kits (Arbor Assays: K014-H5, Ann Arbor, MI, USA), previously employed successfully for WB-CORT quantification for R. temporaria larvae [46]. Measurements were performed in duplicate using the 50 µl format for standard preparations and assays and read with a Tecan Spark Microplate Reader (Tecan, Switzerland) at 450 nm. We calculated sample hormone concentration calibrated with standards included in the kit (https://www.myassays.com/arbor-assays-corticosterone-enzyme-immunoassay-kit-improved-sensitivity.assay). The mean coefficient of variation of all samples was 9.76% (range: 0.13−59%). The software included in the kit assay calculates CV, s.d. and s.e. for each replicated sample and highlights samples that are outside the upper or the lower asymptotes. We lost nine samples due to broken tubes (n = 2) and errors in the assays (n = 7) and discarded eight samples with mean coefficient of variation between duplicates above 30%. Therefore, we built the model to explain WB-CORT levels using 118 samples (individual larvae). Lost samples were well distributed among treatments, with either three (fluctuating temperatures, no nitrate addition), two (18°C, no nitrate; 26°C and fluctuating temperatures, 50 mg l⁻¹ nitrate; 28°C, 100 mg l⁻¹ nitrate) or one (18, 22 and 28°C, 50 mg l⁻¹ nitrate; 22°C, no nitrate; 26°C and fluctuating temperatures, 100 mg l⁻¹ nitrate) missed sample per treatment (from a total of nine).

(d). Body condition

For the assessment of body condition, the scaled mass index (SMI) was used [56] as detailed in the supplementary material.

(e). Microbiome analyses

Entire guts and their contents were homogenized with sterile pipette tips inside a 1 ml tube to proceed with DNA extraction using the kit QIAmp Fast DNA Stool Mini Kit from QIAGEN following the manufacturer’s protocol. One negative extraction control was included for each daily continuous session of extractions (n = 10). Duplicate PCRs were performed using the forward primer 515F (5′-GTGCCAGCMGCCGCGGTAA-3′) and reverse primer 806R (5-GGACTACHVGGGTWTCTAAT-3′), which amplify 251 bp at the V4 region of 16S rRNA gene [57]. Unique primer and tag combinations were selected from a stock of 24 forward and 24 reverse primer tags. Each PCR plate (n = 3) had one negative control. PCR products were run on a 2% agarose gel and extracted using the Monarch DNA Gel Extraction Kit (New England BioLabs, GmbH, Germany) following manufacturer’s protocol for purification. Afterwards, DNA was quantified using the QubitTM system (Invitrogen) and sent to the Max-Planck-Institut für Evolutionsbiologie at Plön, Germany, for MiSeq500 Illumina-Sequencing (paired-end 2 × 250 v2 chemistry).

(f). Data processing

(i). Sequence denoising and filtering

QIIME2 [58] was used for sequence denoising, quality filtering and alpha diversity estimation following [59]. Refer electronic supplementary material for further details. A phylogenetic tree was assembled using the Greengenes (16s rRNA) reference database (version gg-13-8 [60]) and a naive Bayes classifier was trained using the sequences, taxonomy and weights relative to animal proximal gut available at https://github.com/BenKaehler/readytowear [61]. This method results in better accuracy due to the use of a reference library of taxa more likely to occur in the environment of interest. The trained classifier was then used to classify the deblurred sequences.

(ii). Microbiome bacteria composition, beta diversity and predicted metabolic pathways

Treatments were compared regarding recorded ASV counts through PERMANOVA using the function adonis2 (‘bray’ method) in the vegan package [62] in R. The function betadisper was used to test for homogeneity of variances among treatments. This function calculates confidence intervals on the differences among distance-to-centroid means of the groups to be compared using the specified (‘bray’) family-wise probability of coverage. We built a NMDS with two dimensions for data visualization using the functions metaMDS, we also built a NMDS with one dimension and saved the scores to represent community composition for subsequent analyses (described below). Stress values were low (rmse = 0.0008 and max. resid. = 0.0045) and variances did not differ between any pair of treatments, so we considered the scores to provide a good representation of microbiome bacteria composition. Indicator ASVs were determined with the package indicspecies [63] to show which ASVs could be associated with specific treatments. In this analysis, ASVs are selected as indicators based on the strength of their association with a specific treatment, calculated by a randomization test that quantifies the probability of association.

Information on taxonomy, sequence abundance and treatments applied to each sample was combined with phyloseq [64] to be used in ANCOM-BC [65] to show which taxa showed significantly different abundances between at least two treatments. Microbiome bacteria of larvae raised at 18°C were considered as the reference level and compared to those of larvae raised at the other temperatures (22°C, 26°C, 28°C and fluctuating). Microbiome bacteria of larvae raised without nitrate addition were used as the reference level for comparison with those from larvae exposed to intermediate (50 mg of NaNO₃ l⁻¹) and high (100 mg of NaNO₃ l⁻¹) nitrate concentrations. Briefly, ASVs with different abundance between at least two groups across three or more different groups were identified and selected based on corrected p-values, and the log fold changes between the reference treatment and additional treatments were calculated for each significant ASV.

Functional roles of R. temporaria larvae gut bacteria were predicted based on deblurred sequences using PICRUSt2 [66]. PICRUSt2 provides high-level predictions of metabolic pathways related to genes contained in 16S rRNA gene sequence data based on the MetaCyc database [67]. The predictions are based on the placement of ASVs into a reference phylogeny containing 20 000 genes from 41 926 genomes of Archaea and Bacteria from the Integrated Microbial Genomes (IMG) database [68]. Despite limitations represented by the dependency on the reference databases and the restriction of the analysed sequences to the 16S rRNA gene amplicon, such predictions are still a useful tool to gain insights into functional microbial ecology [66]. Predicted metabolic pathways were annotated and compared among groups using ggpicrust2 [69], employing the ALDEx2 method [70] for the analyses of functional profiles. Relative abundances of annotated metabolic pathways were compared among nitrate and temperature treatments to identify significant (p < 0.05) log2 fold changes between treatments.

(iii). Statistical analyses

Amphibian larvae stress level (WB-CORT), body condition (SMI), developmental rate (given by the number of Gosner’s developmental stages advanced by the larvae divided by the number of days from hatching until the end of the experiment), microbiome bacteria diversity (Shannon entropy) and composition (represented by the scores of the NMDS conducted with just one dimension) were considered as dependent variables and their responses to different temperature regimes and nitrate levels (explanatory variables) were tested, while controlling for the effect of aquarium (random variable), using generalized linear mixed models in afex [71] in R [72]. We used aquarium as a random variable to account for the facts that larvae within the same aquarium can share microbes and also be exposed to CORT released by each other (for the effect on CORT release, refer [46]). We conducted post hoc tests to obtain pairwise significant differences for significant variables using emmeans [73] in R. As the results for bacteria composition were equivalent to those of the PERMANOVA, we present the results of the GLMM, which accounted for aquaria as a random variable.

3. Results

(a). Variation on R. temporaria larvae CORT release, SMI, developmental rate, gut bacteria diversity and composition among treatments

Increasing temperature (from 18 up to 28°C) caused a stepwise increase in CORT, with only the 28°C treatment being significantly different from the control (18°C). Larvae in the fluctuating temperature treatment did not differ from the control group (table 1; figure 1a ; electronic supplementary material, S2). Similarly, the overall effect of nitrate showed a stepwise increase in CORT with increasing nitrate concentration, but only the high nitrate (100 mg l⁻¹) treatment was significantly different from the control group. The interaction between temperature and nitrate was not significant (table 1).

Table 1.

Generalized linear mixed models built to explain the effect of temperature, nitrate concentration and their interaction on amphibian larvae stress levels (WB-CORT), body condition (SMI) and developmental rate given by the number of Gosner’s developmental stages advanced by the larvae divided by the number of days from hatching until the end of the experiment, gut bacteria alpha diversity (Shannon entropy index) and composition (1 − axis NMDS scores), controlling for the effect of aquarium. Degrees of freedom are calculated using the Kenward–Roger method.

model		d.f.	F	p
logCORT ~Temp*Nitrate+ (1|Aquarium)	temperature	4; 29.52	3.97	0.010
	nitrate	2; 29.62	6.77	0.004
	temperature: nitrate	8; 29.43	1.79	0.119
SMI ~Temp*Nitrate+ (1|Aquarium)	temperature	4; 29.88	3.49	0.019
	nitrate	2; 29.90	7.40	0.002
	temperature: nitrate	8; 29.86	1.76	0.124
dev_rate ~Temp*Nitrate+ (1|Aquarium) a	temperature	4; 29.81	4.9705e+30	<0.001
	nitrate	2; 29.85	6.8178e+29	<0.001
	temperature: nitrate	8; 29.80	6.9342e+28	<0.001
diversity ~Temp*Nitrate+ (1|Aquarium)	temperature	4; 29.92	4.15	0.009
	nitrate	2; 29.94	3.46	0.045
	temperature:nitrate	8; 29.91	4.41	0.001
composition ~Temp*Nitrate+ (1|Aquarium)	temperature	4; 29.85	62.85	<0.001
	nitrate	2; 29.88	13.48	<0.001
	temperature:nitrate	8; 29.83	6.53	<0.001

Significant p-values (<0.05) are shown in boldface.

^a The model built to explain developmental rate failed to converge because all larvae within treatments achieved developmental stage 38 in the same day, generating a negative eigenvalue. Residual distributions are available in the electronic supplementary material, figures S2–S6.

Figure 1.

Comparison among model estimates for R. temporaria larvae (a) stress levels (WB-CORT), (b) body condition (SMI), (c) developmental rate given by the number of Gosner’s developmental stages advanced by the larvae divided by the number of days from hatching until the end of the experiment, (d) gut microbiome bacteria alpha diversity (Shannon entropy index) and (e) bacteria composition (1 − axis NMDS scores), controlling for the effect of aquarium. Larvae were submitted to the combination of three levels of nitrate exposure (0 [control], 50 and 100 mg l⁻¹) and five temperature regimes (18°C [control], 22°C, 26°C, 28°C and fluctuating temperature). Underlying models are presented in table 1. Different capital letters above each graph indicate differences among nitrate treatments (Tukey method adjusted p < 0.05). Different capital letters within the boxes corresponding to temperature regimes indicate differences among them for each graph (Tukey method adjusted p < 0.05). For models with significant interactive effects, different small cap letters (a–c) in colours corresponding to nitrate exposure indicate nitrate treatments that differed within temperature regimes. Different letters (d–r) in colours corresponding to temperature regimes indicate temperature treatments that differed within nitrate treatments (Tukey method adjusted p < 0.05).

Body condition (SMI) showed a stepwise decrease at increasing temperatures at intermediate or high nitrate concentrations (table 1; figure 1b ; electronic supplementary material, S3). The effect of nitrate varied among temperatures, although the interaction was not significant. In general, however, nitrate treatments resulted in lower body condition compared with the control (figure 1b ).

Developmental rate increased with temperature (especially at 28°C; table 1; figure 1c ; electronic supplementary material, S4) and decreased with nitrate level (table 1; figure 1c ; electronic supplementary material, S4). For developmental rate, the interaction was significant, and the increase in developmental rate with temperature was less pronounced as nitrate levels increased (table 1; figure 1c ; electronic supplementary material, S4). The model built to explain developmental rate showed convergence warnings due to a negative eigenvalue. However, this problem resulted from the lack of variability within treatments and could not be fixed. All larvae started the experiment in developmental stage 25 and reached stage 38 on the exact same day within treatments, but the number of days varied among treatments. Thus, although the data did not meet model assumptions, the detected difference was real.

Larvae gut microbiome bacteria diversity varied with temperature and nitrate concentration, with interactive effects. Bacteria diversity increased from intermediate to high nitrate concentrations (although neither nitrate treatment differed from the control) and showed different patterns within nitrate treatments, being higher at 22°C and intermediate nitrate concentrations, and at fluctuating temperatures and high nitrate concentrations (table 1; figure 1d ; electronic supplementary material, S5). Microbiome bacteria composition also varied with temperature and nitrate concentration, with interactive effects. Without nitrate addition, it was similar at all temperatures but 18°C. At intermediate nitrate concentrations, it showed a stepwise increase in the values represented in the NMDS axis (the corresponding variation in microbiome composition is presented in the next section) with increasing temperatures, and at high nitrate concentrations the changes in composition showed increasing NMDS axis values from 18°C to 26°C and decreasing values from 26°C to 28°C (table 1; figure 1e ; electronic supplementary material, S6).

(b). Compositional and predicted functional changes in R. temporaria larvae gut bacteria among treatments

Our filtered sequences contained 260 ASVs assigned to 12 phyla and 48 identified families of bacteria from the gut microbiome of R. temporaria larvae. The most abundant phyla were Proteobacteria, Bacterioidetes, Firmicutes, Fusobacteria and Cyanobacteria, in that order (figure 2a ). Bacterial communities were mostly dominated by Proteobacteria, except for larvae raised at 18°C and high nitrate concentrations, whose microbiomes were dominated by Bacterioidetes. These larvae also had a marked increase in the abundance of Cyanobacteria and the most diverging microbial communities among all treatments (figure 2a,b ). Firmicutes had higher abundances in larvae raised at 26°C without nitrate. Fusobacteria had higher abundances in larvae exposed to 28°C and fluctuating temperatures (figure 2a ).

Figure 2.

(a) Relative abundance of phyla and (b) NMDS distribution (Procrustes: rmse = 0.0001, max. resid. = 0.0007) of ASVs of bacteria in the guts of R. temporaria larvae submitted to the combination of three levels of nitrate exposure (0 [control], 50 and 100 mg l⁻¹) and five temperature regimes (18°C [control], 22°C, 26°C, 28°C and fluctuating temperature).

Bacteria composition in R. temporaria gut microbiome showed large alterations at intermediate and high nitrate concentrations compared with the control, reflecting on their large number of indicator ASVs in common (29 ASVs) in comparison to the smaller overlap of each nitrate treatment and the control (5 ASVs; electronic supplementary materials, figure S7 and table S1). Larvae reared at 18°C, followed by those reared at 22°C, had the most divergent bacteria composition compared to those from other rearing temperatures (figure 1e ). Indeed, the number of unique indicator ASVs (14 and 12, respectively) was the largest for these two rearing temperatures (electronic supplementary material, figure S7 and table S2). The overlap in species composition was the largest among 26°C, 28°C and fluctuating rearing temperatures, with 23 indicator ASVs unique to them and 44 more also shared with the 22°C treatment (figure 1e ; electronic supplementary material, figure S7). On the other hand, these three treatments only shared five indicator ASVs with the 18°C treatment (electronic supplementary material, figure S7).

Seven families of bacteria and four additional ASVs that could not be assigned to the families represented in the database showed significant increase or decrease in abundance between at least two treatments (electronic supplementary material, figure S8). Among the identified families, Victivallaceae (Lentisphaerae), Veillonellaceae (Firmicutes) and Rikenellaceae (Bacteroidetes) decreased their abundance in the presence of nitrate and tended to increase their abundance with increasing temperatures. Rhizobiaceae, Rhodocyclaceae, Neisseriaceae (Proteobacteria) and Weeksellaceae (Bacteroidetes) showed the opposite pattern. These families increased their abundance with increasing nitrate levels and showed reduced abundance at temperatures above 18°C (electronic supplementary material, figure S8).

Significant increase or decrease in abundance between at least two treatments was recorded for 21 identified genera and 11 additional ASVs which could not be assigned to genus based on the reference database (electronic supplementary material, figure S9). Among the identified genera, the most remarkable decreases in abundance occurred for PW3 and AF12 (Rikenellaceae), followed by Plesiomonas (Enterobacteriaceae) with nitrate addition in comparison to the control, and for Flavobacterium (Flavobacteriaceae) and Chryseobacterium (Weeksellaceae), followed by Pseudomonas (Pseudomonadaceae) and Arthrospira (Phormidiaceae) at higher temperatures compared with the control. Most identified genera showed increases in abundance with nitrate addition. Plesiomonas showed the most remarkable increase in abundance at temperatures above 18°C, followed by PW3(electronic supplementary material, figure S9).

Exposure to nitrate pollution increased the predicted expression of bacterial genes related to denitrification and decreased several pathways of amino acid biosynthesis (electronic supplementary material, figure S10). Metabolic pathways predicted to decrease with increasing temperatures included many degradation pathways (e.g. glucose, amino acids), as well as some biosynthesis pathways (electronic supplementary material, figure S11).

4. Discussion

(a). Physiological responses of R. temporaria larvae to increased temperatures and nitrate pollution

We demonstrated under controlled experimental conditions that increased temperatures and nitrate pollution influenced WB-CORT release rates, body condition, developmental rate, as well as gut bacteria diversity and composition in R. temporaria larvae, with many interactive effects. Larvae exposed to the highest temperatures developed faster and revealed a lower body condition under nitrate pollution, indicating energy allocation to development in the detriment of storage.

As ectotherms, amphibians show increased metabolic and developmental rates at higher temperatures, resulting in greater energy demands [74]. Temperature is a major determinant of amphibian larvae developmental rate [75,76]. Thus, it is not surprising that larvae developed synchronously within our treatments, as both temperature and other experimental conditions were standardized and they were fed ad libitum. Additional energy demands imposed by higher temperatures result from the activation of the hypothalamic–pituitary–adrenal/interrenal (HPA/I) axis in response to stressors [77,78]. The consequent secretion of glucocorticoid hormones (GC) from the adrenal (cortisol; predominant GC in most primates and teleost fish) or interrenal (CORT; predominant GC in most amphibians and sauropsids) glands (rev. in Ruthsatz et al. [46]) is followed by mobilization of glucose and lipid stores to regulate metabolism and nutrient homeostasis [77,79]. Thus, the combined energy demands of increased temperatures and nitrate pollution likely decreased larvae body condition, culminating in a significant effect at 28°C. At lower temperatures, larvae may have been able to adjust to the increasing energy demands by eating more and converting food into mass (possibly helped by microbiome-mediated metabolic adjustments, as discussed below). Under natural conditions in a warming scenario, however, growth restriction may occur at lower temperatures because food is not expected to be available ad libitum [80] as in our experiment.

In accordance with our results, in amphibian larvae, elevated CORT levels in response to stress exposure accelerate development, thereby facilitating a faster emergence from suboptimal conditions in the larval habitat through a rapid metamorphosis [81,82]. However, the opposing effect of nitrate pollution, decreasing developmental rate, is likely to increase the exposure time to this pollutant, ultimately decreasing survivorship [7,44]. Our results corroborate a trade-off between investment of the available energetic storages on development versus detoxification processes and these conflicting energy demands may be especially challenging for amphibian larvae exposed to both high temperatures and levels of nitrate. It is also important to note that the developmental rate was faster at the fluctuating temperature treatment compared with 22°C and even 26°C treatments, although the mean temperature was 21.1°C, indicating that it is desirable to incorporate natural variability (e.g. in temperature regimes) in laboratory experiments to help extrapolate their results to natural habitats [24].

(b). Gut bacteria hosted by R. temporaria larvae exposed to increased temperatures and nitrate pollution

Environmental challenges may bring microbiomes to a new equilibrium state with benefits or detriment to the host, and beneficial new equilibrium states are not necessarily more diverse, as suggested by the contradictory results regarding the diversity of the microbiomes of ectotherms exposed to increasing temperatures [34,36] (and this study). Temperatures 2–3°C above natural temperatures resulted in microbiomes with lower richness in the lizard Zootoca vivipara in mesocosms, and microbiome richness was positively associated with survivorship [36]. Microbial community diversity of R. clamitans larvae raised at 14°C, 22°C and 28°C was also shown to decrease at higher temperatures [34]. For R. temporaria, however, a temperature 4°C higher than natural temperatures was associated with an increased microbiome bacteria diversity that was comparable to the bacteria diversity of larvae reared at the other temperature regimes (26°C, 28°C and fluctuating temperatures). Despite similar bacteria diversity, larvae reared at 28°C showed a marked decrease in body condition compared with the control. Microbiome bacteria composition, however, differed among 18°C, 22°C and 26°C, and such changes may have contributed to the maintenance of body condition at these temperatures. On the other hand, an additional temperature increase to 28°C did not change bacteria diversity and composition in comparison with 26°C, indicating a threshold of the influence of temperature on bacteria community composition changes and consequent effects on host health.

Although correlative, our results indicate that microbiome bacteria composition and its associated metabolic pathways are more likely to explain outcomes in host health than bacteria diversity. Many metabolic pathways predicted to decrease with increasing temperatures in R. temporaria larval gut bacteria are present in Pseudomonas (MetaCyc database [67]), whose abundance decreased in the same direction. For example, the observed reduction in leucine degradation may increase leucine availability to the host and corroborate the accelerated development of larvae with increasing temperatures. Leucine regulates protein metabolism preventing protein degradation [83] and has been demonstrated to boost muscle growth and protein deposition in fish [84]. The likely increased levels of l-tyrosine achieved through reduced l-leucine degradation in larvae reared at higher temperatures can be beneficial to the host immune response [85], whereas decrease in glucose oxidation may result in high levels of glucose, which may help attend increased metabolic demands resulting from higher temperatures [77,79]. On the other hand, some alterations in metabolic pathways observed at higher temperatures may be detrimental to the larvae. For example, decreased octane oxidation and toluene degradation may lead to increased levels of these hydrocarbons. Because hydrocarbons are associated with oxidative stress, the stressful effects of increased temperatures may be aggravated in polluted environments with high hydrocarbon concentrations [86].

Exposure of amphibian larvae to nitrate pollution is likely to increase nitrate metabolism and alter amino acid metabolism, resulting in decreased body condition[23] (and this study). For R. temporaria larvae, nitrate reduction (denitrification) increased with nitrate exposure, and Pseudomonas, known to have this pathway (MetaCyc database [67]), increased in abundance in the same direction. As also observed for B. gargarizans, increased nitrate degradation can bring negative results to amphibian larvae [23] mediated by increased levels of nitrite impairing oxygen transport [21]. Alterations in nitrate metabolic pathways have already been observed in nitrate-polluted environments. For example, the microbiomes of oysters (Crassostrea virginica) regulated genes associated with specific nitrate and phosphorus metabolic pathways in response to eutrophication in natural habitats [33]. In this case, an increased expression of genes involved in stress response and phosphorus metabolism occurred under eutrophic conditions, whereas genes involved in nitrogen metabolism were upregulated in less eutrophic sites. However, the influence of oxygen levels in the water was assumed to play an important role, corroborating a complex, multifactorial response of microbiomes to environmental conditions [33]. Our experimental conditions were aimed to standardize sources of environmental variation other than nitrate exposure and temperature, which means that amphibian larvae microbiomes may show different responses to the same nitrate levels and temperature conditions tested here due to interactive effects among these and other variables in natural habitats.

Besides the responses likely related to increased temperatures or nitrate pollution, some alterations in the microbiome bacteria and their influence on metabolic pathways may have resulted from the interaction between both stressors. The l-lysine biosynthesis and fermentation to acetate and butanoate were predicted to increase at 22°C, 28°C and fluctuating temperatures compared with the control (18°C), together with an increase in the abundance of Clostridium, a genus presenting this pathway (MetaCyc database [67]). These pathways are related to lysine metabolism and the regulation of lysine production has already been detected as a mechanism to deal with stress caused by nitrate pollution in amphibian larvae [23]. Adjustments in this pathway might have helped the R. temporaria larvae to deal with the combination of increased temperatures and nitrate exposure, as Clostridium was also abundant at high nitrate concentrations.

An important remark is that many altered bacterial metabolic pathways detected in R. temporaria larvae gut microbiomes are not reported in the MetaCyc database for any microbial taxa with altered abundances among our treatments but are recognized for well-known bacteria such as Escherichia coli. Considering the relative scarcity of studies on microbiomes of ectotherms [27], the presence of such pathways in bacteria associated with guts of ectotherms is likely to represent a knowledge gap. It is therefore important to investigate the microbiome composition of ectotherms as well as gene expression through transcriptomes in order to acquire more information on microbiome–host metabolic pathways and how they respond to environmental conditions [33].

(c). Microbiome responses to environmental stress and conservation implications

Increasing evidence has been highlighting the ability of symbiotic microbial communities to adjust to environmental conditions with benefits to their hosts. Ectothermic hosts have been shown in many instances to have an improved ability to deal with heat stress after colonization by specific groups of microorganisms (reviewed by Fontaine & Kohl [12]). According to the adaptive microbiome hypothesis, hosts exposed to pathogens tend to have lower microbiome diversity with higher representativeness of bacteria that boost useful immune functions to resist infection [28]. The order Rickettsiales (Alphaproteobacteria) showed increased abundance and dominance in staghorn corals (Acropora cervicorni) under increased nutrient concentrations achieved with fertilizer addition and hosts were able to remain healthy under the resulting pollution conditions, although the influence of microbial community changes on host health remained to be investigated [32]. Proteobacteria, Bacterioidetes, Firmicutes and Fusobacteria are dominant phyla in guts of amphibian larvae [38,39] (and this study). The gut microbiome of B. gargarizans larvae showed increased abundances of Proteobacteria and Firmicutes under Cu and Pb pollution [39], and a reduced abundance of Bacterioidetes and Fusobacteria under nitrate pollution [23]. Here, instead, we observed a relatively high abundance of Bacterioidetes in the guts of R. temporaria larvae at high nitrate concentrations at 18°C. Such contrasting results indicate that it is important to investigate the composition of microbiome components at a deeper taxonomic level and also to investigate gene expression through transcriptomes to be able to understand how the adjustments of microbiomes to environmental conditions can bring consequences to the host’s health.

5. Conclusion

In this study, we demonstrated that increased temperatures and nitrate levels resulted in higher levels of physiological stress and reduced body condition on a sensitive ectotherm model organism represented by R. temporaria larvae. Larvae accelerated development at higher temperatures, however, this mechanism was counteracted by nitrate pollution, forcing larvae to remain for a longer time in a stressful aquatic environment. Pollution, a remarkable cause of population declines affecting survivorship, reproductive success and recruitment [41], is expected to alter microbiome composition and beta diversity, leading to consequent metabolic disturbances [38,39]. Although some observed alterations in the gut bacteria and associated metabolic pathways appeared to counteract the detrimental effects of increasing temperatures and nitrate levels to a certain degree, the combination of the highest levels of both stressors resulted in decreased body condition. Our study emphasizes the potential interaction between nitrate pollution and global warming on ectotherm microbiome shaping that might be of great relevance for population persistence. Although the effects of pollution [23,32,33] and temperature [12] have already been addressed, this is the first study to show an interactive effect of both stressors on the gut microbiome of an ectotherm. As species frequently endure multiple stressors simultaneously [16,87,88], it is possible that even if their microbiome can adjust to some of them, the addition of others may hamper the adaptive flexibility of the holobiont, for example, restraining the availability of colonizing microorganisms in the host’s environment [38] or the ability of the microbiome to achieve a new equilibrium state that maintains host health [89]. We therefore encourage additional research on the role of microbiomes on host ability to deal with environmental stressors, as well as its underlying mechanisms, to make possible the development of conservation strategies that take the capacity of the holobiont to adapt into account.

Ethics

Field work was conducted in accordance with permits of the Stadt Braunschweig (Fachbereich Umwelt und Naturschutz, Richard-Wagner-Straße 1, 38106, Braunschweig; 38106 Braunschweig Gz. 68.11-11.8-3.3) and the experiments were performed in accordance with permits from the Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit, Germany (Gz. 33.19-42502-04-20/3590). The authors have no ethical conflicts to disclose.

Data accessibility

Data associated with this are available in the Figshare data repository [90]. The genomic data for this study have been deposited in the European Nucleotide Archive (ENA) at EMBL-EBI under accession number PRJEB72767 [91].

Supplementary material is available online [92].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors’ contributions

P.C.E.: data curation, formal analysis, investigation, methodology, validation, visualization, writing—original draft, writing—review and editing; R.S.: investigation, writing—review and editing; J.S.-P.: validation, writing—review and editing; C.Y.: software, writing—review and editing; S.K.: resources, writing—review and editing; K.R.: conceptualization, funding acquisition, investigation, methodology, project administration, supervision, validation, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

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.

Funding

The German Research Foundation (DFG) project (459850971; A new perspective on amphibians and global change: Detecting sublethal effects of environmental stress as agents of silent population declines) supported K.R. and P.C.E. R.S. was supported by the DFG priority program SPP 1991 Taxon—Omics (grant number VE247/20-1). European Union's Horizon 2020 under the Marie Skłodowska-Curie grant agreement ID: 101028000 supported J.S.-P. Funding sources have no role in data collection and analyses, manuscript writing or decision to submit for publication.