The effect of pyridine-alkaloid secondary metabolites on sunbirds gut microbiome

Background: Sunbirds ( Cinnyris osea ) feed on nectar of the tobacco tree ( Nicotiana glauca ), which contains the secondary metabolites nicotine and anabasine. These toxic pyridine-alkaloids presumably serve to deter nectar thieves and allow only specific pollinators to feed on nectar. The floral nectar contains low level of proteins and hence nectarivores may have problem balancing their nitrogen requirements. Our aim was to understand whether the gut microbiome of nectarivores contains bacteria that are able to degrade plant secondary metabolites and how the presence of secondary metabolites in nectar shapes the gut microbiome of nectarivores. Results: Sunbirds (n=16) that were held in captivity were randomly divided into control and treatment groups. The control group was fed with artificial nectar while the treatment group was fed with artificial nectar with the addition of nicotine and anabasine. Excreta were collected from each of the 16 birds on day 0 and at the end of weeks 2, 4 and 7. Nonmetric multidimensional scaling analysis (nMDS) demonstrated that the secondary metabolites significantly changed the gut microbiota composition and diversity of those sunbirds that were fed nectar with added pyridine alkaloids. Aeromonas, Enterobacteriaceae , Exiguobacterium, Klebsiella , Lactococcus , Leuconostoc , Rothia and Salmonella were found to contribute to the differences in gut microbiota between the control and treated birds (birds feed with Nicotine and Anabasine). Four weeks after starting to feed the birds with added nicotine and anabasine, we found that genera with the ability to degrade these pyridine-type metabolites proliferated from a total relative abundance of only 0.3% to 52.7% of the gut bacterial community. Conclusion: Sunbirds feed on Nicotiana nectar that contains toxic pyridine-alkaloids. Nonetheless, the symbiotic gut microbiota which is changing dynamically towards enabling the degradation of these secondary metabolites, supply the birds with nitrogen that enables the synthesis of essential amino acids. The current study contributes to our understanding of the complex relationships between plants, birds and their microbiomes. species that are able to degrade these secondary metabolites. The results showed that the composition of the bacterial microbiome changes significantly when the birds were fed nectar with added nicotine and anabasine compared to the control birds; moreover, significant changes were observed over time.

functions. One important environment is the gut of animals, which harbours a complex microbial community that is important for physiology, immune system development, digestion and detoxification reactions in their hosts [1][2][3]. The majority of the studies on microbiomes have focused on mammalian hosts -largely ruminants and humans [4]. Studies on gut-microbiome diversity and functions in avian hosts have been largely overlooked [3]. Of those limited number of studies on avian gut microbiomes, most were carried out on domestic birds like chickens and turkeys [3], rather than on wild birds. Furthermore, only a few studies have examined the gut bacteria of passerine birds [5][6][7][8]. As far as we know, only a single study has been carried out on the gut bacteria of nectarivorous birds (hummingbird, Calypte anna), using culture-dependent techniques to investigate nitrogenrecycling bacteria [9].
The importance of specific bacteria for digestive recycling in avian species with large ceca and welldeveloped gastrointestinal microbiotas has been well documented [10]. However, most nectarivorous and frugivorous birds have only vestigial ceca. For example, hummingbirds, arguably the most specialized avian nectarivores, have no ceca and perform extremely fast digestion throughout the entire digestive tract, which may make the colonization by bacteria challenging [3,11,12]. Therefore, it has been assumed that the gastrointestinal tracts of birds that feed on nectar do not have the structures needed to house intensive microbiota, as is presumably required for effective digestive recycling [3]. However, our previous study [13]suggested that, despite lacking large ceca, the frugivorous yellow-vented bulbul (Pycnonotus xanthopygos) possesses a rich gut microflora, including bacterial species that are capable of degrading uric acid.
The widespread existence of 'toxic nectar,' or nectar with secondary metabolites, is puzzling given that the function of nectar is to attract mutualists, such as legitimate pollinators [14]. Nevertheless, plant secondary metabolites may play an adaptive role as mediators of mutual plant-animal interactions, such as pollination and seed dispersal, and thus may increase plant fitness [15][16][17]. The 'direct toxicity hypothesis', suggests that the plant secondary metabolites control or filter out nectar robbers and allow only the appropriate pollinators to feed on nectar [18,19].
Nectarivores may have a problem balancing their nitrogen. Nitrogen can be a limiting resource for nectarivores, which require it for protein, nucleotides and the synthesis of other nitrogenous compounds [20]. Floral nectar, particularly of hummingbird-pollinated plants, contains low levels of proteins [21]. Tsahar et al. [13] showed that nectar-feeding birds have low nitrogen requirements, but the mechanisms that these animals use to conserve nitrogen remain unclear. Studies on the gut bacteria of insects demonstrated that these bacteria compensate for dietary deficiencies in essential amino acids by using various mechanisms, such as the recycling of nitrogenous waste, to nutritionally upgrade the diet of their host [22]. A similar mechanism was reported in hummingbirds by Preest et al. [9], who found the decomposition of nitrogenous compounds by intestinal bacteria. A metagenome analysis of cecal samples from sage-grouse (Centrocercus urophasianus) by Khol et al. [23], discovered rich genes associated with the metabolism of plant secondary metabolites. They suggested that the gut microbiota of specialist avian herbivores play a similar role to that of the gut microbiota of mammals in degrading plant secondary metabolites.
We hypothesized that plant secondary metabolites are mediators of the interactions between plants and nectar consumers through the gut microbiome of the latter. Thus, our aim was to understand (i) whether the gut microbiome of nectarivores contains bacteria that are able to degrade plant secondary metabolites; (ii) how the presence of secondary metabolites in nectar shapes the gut microbiome of nectarivores; and (iii) how the time elapsed from initial exposure to secondary metabolites in the birds' diet affects the dynamics of the composition of the gut microbiome. To accomplish these goals, we based our study on the natural relationships occurring between the tobacco tree (Nicotiana glauca), whose nectar contains pyridine alkaloids (mainly anabasine and nicotine), and its avian nectar consumers in the Old World (orange-tufted sunbirds, Cinnyris osea). In an experimental setup with captive birds fed with artificial nectar and with or without the addition pyridine alkaloids, we were able to demonstrate that sunbirds' gut microbiota include bacterial species that are able to degrade these secondary metabolites. The results showed that the composition of the bacterial microbiome changes significantly when the birds were fed nectar with added nicotine and anabasine compared to the control birds; moreover, significant changes were observed over time.

Sunbirds feeding experiment
Sixteen birds were adapted to the laboratory conditions were randomly divided into two groups. The control group was fed the artificial nectar without any additional nutrients while the treatment group was fed the same artificial nectar with the addition of nicotine and anabasine (0.5 ppm and 5 ppm, respectively) (Fig. 1). Excreta were collected from each of the 16 birds on day 0 and at the end of weeks 2, 4 and 7 (Fig. 1). The excreta samples were used for culture-dependent and cultureindependent bacterial analyses (more details can be found in the Methods).

Culture-independent approach
A total of 3 638 528 quality reads were generated from 66 cloacal samples from sunbirds, with an average count of 55 129 reads per sample (after size filtering, quality control, and chimera removal).
Overall, 812 operational taxonomic units (OTUs) were identified based on 97% nucleotide sequence identity between reads. After removing rare OTUs, 331 were left. The rarefaction curves of each sample reached an asymptotic level (Fig. S1), suggesting that our sampling efforts were sufficient to obtain a full estimate of OTU richness.

The effect of treatment and time on gut-microbiome composition and diversity
nMDS analysis demonstrated dynamic changes between the gut microbiota composition of sunbirds from each feeding group and over time (Fig. 2). Significant differences (ANOSIM) were found between the composition of the sunbirds' gut bacterial communities in the control and treatment groups. The maximal difference was observed after four weeks (R = 0.353, P = 0.013), with the average of 49.9% Bray-Curtis dissimilarity between the control and treatment groups. Fewer differences were observed after two weeks (ANOSIM R = 0.29, P = 0.006) and after seven weeks (R = 0.329, P = 0.002), with the averages of 32.9% and 31.8% Bray-Curtis dissimilarities, respectively ( Fig. 2; Tables 1, S1).
Significant differences in the composition of the sunbirds' gut microbiota were also observed over time, both in the control and treatment groups (Table 1). Moreover, the changes over time were more pronounced in the control group than in the treatment group. The differences in the bacterial community composition groups between the samples at day 0 and after two weeks in the control group were larger than in the treatment group (ANOSIM control groups -R = 0.517, P = 0.001; treatment groups -R = 0.488, P = 0.001). The same pattern was observed between the samples after two and four weeks (control, R = 0.755, P = 0.001, and R = 0.393, P = 0.001, respectively) and between four and seven weeks (R = 0.691, P = 0.001 and R = 0.202, P = 0.009, respectively) ( Table   1).
The Shannon H diversity index of the control group decreased over time, whereas in the treatment group, it increased until the fourth week and then decreased drastically (Fig. 3). After the beginning of the feeding experiment, in week 2, the Shannon H index was higher in the treatment group than in the control group for each sampling time (Fig. 3). The highest difference between the two groups in the Shannon index was observed after four weeks and the lowest difference occurred after seven weeks (Fig. 3). A mixed model ANOVA showed that time had a significant effect on Shannon diversity (all birds pooled, F 3,42 = 3.88, P = 0.016). The interaction of time X treatment was also significant (F 3,42 = 5.66, P = 0.002), indicating that the two groups (control and treatment) demonstrated different diversity patterns over time (Fig. 3).

Bacterial composition in the sunbirds' gut
Proteobacteria, Firmicutes and Actinobacteria were the three dominant phyla detected in the experiments conducted at different time intervals (Fig. 4a). In the naïve birds (at day 0), the Proteobacteria were the relatively dominant phyla (49% in 0WC and 56% in 0WT), but they decreased thereafter with time. After two weeks Firmicutes became dominant and were more abundant in the control (63% in 2WC) than in the treatment group (53% in 2WT). At the end of the fourth week, Proteobacteria (28%) and Firmicutes (48%) abundances were greater in the treatment group (4WT) compared to the control group (4WC) 14% and 33%, respectively. By contrast, Actinobacteria, which were less abundant in the previous time intervals, showed the highest prevalence at week four and were more abundant in the control than in the treatment group (52% in 4WC and 21% in 4WT). After seven weeks, Firmicutes was the most dominant phylum in both the control (61% in 7WC) and the treatment groups (65% in 7WT), followed by Proteobacteria, with a relative abundance of 32% in 7WC and 24% in 7WT (Fig. 4a).
A total of eight genera displayed mean relative abundances of over 1% across all samples, and the genera with the highest relative abundances are represented in Fig. 4b. Acinetobacter was one of the dominant genera in the initial naïve birds, with a relative abundance of 25% in 0WC and 18% in 0WT; however, its abundance decreased over time, until it reached less than 1% in both groups. After two weeks, the dominant genera in both the treatment and control groups were Lactococcus (36% in 2WC and 35% in 2WT), unclassified Enterobacteriaceae (28% in 2WC and 32% in 2WT), and Leuconostoc (18% in 2WC and 12% in 2WT). Rothia became the most abundant genus in the fourth week and was more abundant in the control (51% in 4WC) than in the treatment group (21% in 4WT). After seven weeks, Lactococcus (35% in 7WC and 49% in 7WT) and unclassified Enterobacteriaceae (29% in 7WC and 19% in 7WT) became the most abundant genera (Fig. 4b).
Ten OTUs were found to contribute to most of the significant differences between the bacterial community compositions in the control and treatment groups at various time intervals ( Differences in Lactococcus relative abundances contributed most to the dissimilarities between the microbial community structures of the control and treatment groups (8.46% of the total dissimilarity, on average), followed by Rothia (7.81% of the dissimilarity, on average) (Table S1). Leuconostoc had an average contribution of 6.69% of the dissimilarities and unclassified Enterobacteriaceae contributed another 5.96%. Other contributing genera were Klebsiella (1.10%) and Salmonella (1.75%) (Fig. 5).

Culture-dependent approach
A total of 146 nicotine/anabasinedegrading strains were isolated from the sunbird excreta. At day 0, 28 were isolated from all 16 birds. After four weeks, 118 isolates were isolated (Table 2), 51 from the control birds and 67 from the treated birds. The isolates belonged to 6 bacterial classes and 14 genera. The majority of the isolates at day 0 belonged to Gammaproteobacteria class and to four species in the genus Enterobacter. After four weeks, five bacterial classes were identified from the control group and six classes from the treatment group. The genera Methylorubrum and Roseomonas were isolated only from the nicotine/anabasine-treated birds. Interestingly, all isolates were able to degrade both nicotine and anabasine, although they were enriched and isolated on only one kind of medium (M9 with nicotine or M9 with anabasine) ( Table 2).

The proliferation of specific genera during treatment
All statistical parameters indicated that the most significant differences were observed in the bacterial communities in the excreta of the treatment group compared to those of the control group.
(Figs. 2, 3, 5; Tables 1, 2, S1). Thus, we followed the changes in the Illumina results for the OTUs of the gut bacterial community between day 0 (0WT) and the fourth week (4WT). All the OTUs with increased relative abundances between 0WT and 4WT were recorded. We found that the abundances of 45 genera increased from time 0 to the fourth week (Illumina results at the OTU level, data not shown). Of these 45 genera, 19 were either reported as nicotine-degrading bacteria in the literature or were isolated by us as nicotine-and anabasine-degrading species in the culturable part of the current study (Tables 2, 3). Interestingly, these 19 genera comprised only a very minor part of the bacterial communities in the excreta at the beginning of the experiment (0WT, 0.30% abundance out of the OTUs) and proliferated up to 52.72% after four weeks of treatment (4WT) with the addition of the pyridine-alkaloid food additives ( Table 3). The two most dominant genera at week four, whose representative species were also isolated in the current study, were Lactococcus and Enterobacter, which were below detection in the Illumina sequencing results at day 0 and reached the high abundances of 29.54% and 18.38%, respectively, in the fourth sampling week.

The effect of diet on gut microbiome
The study of birds' gut microbiomes is increasingly important for ecological and evolutionary perspectives because of its relevance to host fitness, longevity, disease resistance and adaptation [36]. Here we explored the effect of secondary metabolites on the gut microbiota of sunbirds, whereupon we fed the control birds with artificial nectar and added nicotine and anabasine to the nectar of the treated birds. nMDS analyses of the Illumina 16S rRNA gene sequencing demonstrated significant differences between the bacterial communities of control and treated sunbirds' guts ( Fig.   2; Table 1). These differences reveal that diet influenced the composition of the gut microbiome.
A similar observation has been reported by Baurhoo et al. [37] regarding the impact of different types of diet on the composition of the gut microbial community of broiler chickens. They fed the chickens with two types of mannan-oligosaccharides and two types of antibiotics. The mannanoligosaccharides diet significantly affected the composition of the gut microbiota of broilers compared to the control and antibiotic diets [37]. Borda Molina et al. [38] studied the effects of diets supplemented with the minerals phosphorous, calcium and phytase on the gut microbiome of broiler chickens and found that the phosphorous-supplemented diet significantly influenced the microbial community compared to the calcium-supplemented diet.
Diet affects not only the microbiome composition but also its diversity. Lower Shannon diversity was observed in chickens that were fed the calcium diet, compared to chickens that were fed the control diet [38]. Our results showed that the gut-microbiome diversity of sunbirds that were fed with added nicotine and anabasine was higher than the control group (Fig. 3). Pyridine alkaloids may stimulate both positive and negative responses on various bacteria. Hence, the higher diversity of the gut microbiome of the treatment group may reflect a transitional state of the gut microbiome, in which the suppressed bacteria are still present alongside the appearance of new bacterial species.

The effect of time on the gut microbiome
Our study demonstrated that not only diet but also time significantly affected the gut-microbiome composition and diversity (Figs. 2, 3; Table 1). A large-scale study based on 16S rRNA genesequencing, which was conducted on faecal samples of newborn piglets at various time intervals (days 1, 7, 14, 21, 28, 35, 120 and 180), showed relatively apparent temporal and spatial changes in the gut microbiome [39]. In another study, caecal and faecal microbiomes of broiler chickens, which were sampled at various time points between one and six weeks, showed significant changes over time. Significant differences were observed both in richness and diversity in the six-week-old birds compared to three-and one-week-old birds [40].
A similar observation was reported by Ji et al. [41]. Four healthy human individuals and six individual housed mice were fed two different diets (low-fat, high-plant polysaccharide, LFPP; and high-fat, highsugar, HFHS). The diets, especially the LFPP diet, significantly affected the short-term dynamics of the gut microbiota, whereas in the long term, the dynamics of the microbiota decreased and was stabilized [41]. We observed similar dynamics when the diversity of the gut microbiota of birds that were fed with nicotine and anabasine increased between the second and the fourth week and then decreased after seven weeks (Fig. 3). Furthermore, the pyridine-alkaloid diet showed a marked effect on the composition of the gut microbiome after four weeks, but after seven weeks, the composition of the gut microbiome was more closely resembled to that at the second week (Fig. 2).
Proteobacteria, Firmicutes and Actinobacteria were relatively abundant phyla in the sunbirds' gut microbiota in both the control and the treatment groups at different time intervals (Fig. 5a). This pattern bears high resemblance to the microbiota composition of other neotropical birds [42]and also to that of Darwin's finch [43]. Proteobacteria and Firmicutes were the dominant phyla in the sunbirds treated with nicotine and anabasine after four weeks. Similar results were observed in gull species that were treated with glucocorticoids [44].

The effect of secondary metabolites on the gut microbiome
In a recent study, Aizenberg et al. [45] studied the influence of nicotine on the bacterial community composition of nectar of the wild-type N. glauca and the wild-type and transgenic N. attenuata. They [45] found that the elimination of nicotine from the transgenic N. attenuata (in which the nicotine biosynthesis was silenced), significantly affected the bacterial community of the nectar compared to the wild-type plants. Although the bacterial community composition in N. glauca nectar was different from that of N. attenuata, both in its wild-type and transgenic forms, the two wild-type plants that contained nicotine in their nectar were more similar compared to the transgenic N. attenuata plant that was manipulated to not produce nicotine. Thus, nicotine manipulation in Nicotiana plants affects the microbiome of the floral nectar [45].
Nicotine can be toxic even in a relatively low concentration and, at the same time, this metabolite can be the source of nutrients if enzymes are present that can degrade this molecule. Here we demonstrated that pyridine alkaloids that are added to artificial nectar in concentrations that naturally occur in the floral nectar from which sunbirds feed affect the sunbirds' microbiome composition and diversity and shape the gut microbiome of these nectarivorous birds.
In nature, the birds consume plant products and invertebrates that contain secondary metabolites that are toxic when absorbed [46]. According to Dearing et al. [47] birds may harbour bacteria that can detoxify these toxic metabolites. The plant saponins was found to be degraded by bacteria in the crop of Opisthocomus hoazin [48], and several mycotoxins were found to be metabolized by microbial communities from the intestines of chickens [49]. A caecal metagenome analysis of the Centrocercus urophasianus revealed that its microbiota was enriched with genes related to the metabolic pathways that can degrade phenols to pyruvates as well as genes responsible for xenobiotic degradation and the metabolism of terpenoids [23]. Kohl et al. [23]also found genes that were responsible for the biosynthesis of essential amino acids. They suggested that these bacterial genes can assist the host in maintaining nitrogen balance. Similarly, we suggest that the gut microbiome of sunbirds degrades pyridine alkaloids and thus promotes nitrogen balance in sunbirds.
There are various reports of nicotine degradation by different bacterial genera; for example; Delftia, Klebsiella, Stenotrophomonas Pseudomonas and Lactobacillus [28][29][30]32,35]. Here we have also isolated various bacterial genera ( Table 2) that showed the ability to grow on nicotine and anabasine as sole carbon and nitrogen sources and, thus, to degrade these compounds. The following bacterial genera that were isolated and identified by us in the current study have, as far as we know, never been reported as nicotine and or anabasine-degrading bacteria: Chryseobacterium, Enterobacter, Exiguobacterium, Lactococcus, Methylobacterium and Kocuria (Table 2).
Similarly using caffeine as a sole source of carbon and nitrogen, Ceja-Navarro et al. [50] isolated bacteria from the gut of the insect Hypothenemus hampei that had the ability to degrade the toxic alkaloid caffeine. The majority of the caffeine-degrading isolates belong to the genus Pseudomonas.
In our study, Pseudomonas isolates were also the prominent genus among nicotine-and anabasine degrading isolates (Table 2). Ceja-Navarro et al [50] showed that the gut bacterial community of H.
hampei metabolized and detoxified caffeine and thus promoted the reproduction and fitness of the host. In our study, we found a significant change in the bacterial community composition, in which a number of genera with the ability to degrade the toxic metabolites nicotine and anabasine (Table 3) proliferated from 0.3% of the total bacterial community to 52.7% after four weeks of feeding sunbirds with these secondary metabolites, in the same concentrations as they occur in their natural food, N. glauca nectar. These results clearly indicate that these bacteria play an important role in the detoxification and degradation of nicotine and anabasine.

Conclusion
The current study contributes to our understanding of the complex relationships between plants, birds and their microbiomes. Sunbirds feed on Nicotiana nectar that contains toxic pyridine-alkaloid secondary metabolites, which presumably serve to deter nectar thieves. Nonetheless, when sunbirds feed on this nectar, their symbiotic gut microbiota detoxify these toxic secondary metabolites and provide sunbirds with valuable nutrients like nitrogen. The nectar is very poor in protein content and the bacterial community, which is changing dynamically towards enabling the degradation of the secondary metabolites supplies the birds with nitrogen and enables the synthesis of amino acids, which are essential for protein synthesis. For further verification of this theory, a metagenomic study should be performed on a similar set of experiments like the current one.

Studied Organisms
The orange-tufted sunbird (Cinnyris osea) is a small passerine bird weighing 6-7 g. It has a long, slender, decurved bill (1.4-2.0 cm in length) with a long tongue, which allows it to feed mainly on floral nectar and arthropods [51]. Like other nectarivorous birds, the sunbirds feed on high carbohydrate foods with low-protein content and have high sugar-absorption efficiencies despite the rapid speed that food moves through their gut [52].
Nicotiana glauca is native to Argentina and Bolivia and is also found in other parts of South America, Hawaii, Australia and in the east Mediterranean region, including Israel. N. glauca is a small tree with long, tubular yellow flowers that bloom almost throughout the year in warm climates. The nectar of N. glauca is rich in sugar with a mean sugar equivalent concentration of 20% ± 0.3%. The nectar contains the toxic pyridine alkaloids nicotine and anabasine in the concentrations of 0.50 ± 0.12 ppm and 5.0 ± 0.8 ppm, respectively [53]. The pollination of N. glauca depends on pollinating vectors because its stamens are shorter than the stigma [54]. N. glauca has a relatively long corolla and, hence, mainly depends on birds with long bills, such as sunbirds and hummingbirds, for pollination.
The sunbirds in Israel are the main pollinators of N. glauca (60% of them are legitimate visitors, where they feed on nectar from the front of the flower) [55].

Ethical statement
All methods were performed in accordance with the relevant guidelines and regulations. Sunbirds After the experiment ended, the birds were set free.

Sunbirds feeding experiment
After capture, the 16 birds were adapted to the laboratory conditions that included a room temperature of 25 °C, 12h:12h light:dark and fed with artificial food (Sunbird nectar special formula for Nectariniidae; Aves & Avian, Lot nr IS240718; Reg.nr. NL113333, Raalte, Netherlands) and water, for a period of four weeks.
After four weeks of captivity, the birds were randomly divided into two groups by assigning eight birds to each group (that is, each bird was assigned randomly, but the cages stayed in the same spot and were not moved). The control group (eight birds) was fed the artificial nectar mentioned above, without any additional nutrients. The treatment group (eight birds) was fed the same artificial nectar with the addition of nicotine and anabasine (Sigma Aldrich, Rehovot, Israel), in concentrations that naturally occur in N. glauca (0.5 ppm and 5 ppm, respectively) (Fig. 1). The concentrations of the pyridine alkaloids were based on Tadmor-Melamed et al. [53]. Fresh artificial nectar and water were supplied every day to both groups. Excreta (cloacal fluid and faeces) were collected from each of the 16 birds on day 0 (before adding nicotine and anabasine to their feed) and at the end of weeks 2, 4 and 7 (Fig. 1). The excreta collection procedure was as follows: a new, clean piece of baking paper was spread on the bottom of each cage so that the cloacal fluids/faeces would not get contaminated with the cage surface; once the bird left its excreta it was immediately collected using a sterile tip into a sterile Eppendorf tube.
The excreta samples were used for DNA extraction for microbiome analyses (culture-independent) and for isolating nicotine/anabasine-degrading bacteria, as described below.

Culture-independent approach
Cloacal fluids/faeces (excreta) samples for culture-independent analyses were kept at -20 ºC until DNA was extracted using DNeasy Blood and Tissue isolation kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. University of Illinois, Chicago. The sequencing protocol was followed exactly as described by Aizenberg et al [45]. Before sequencing the samples, a second PCR amplification was performed in a 10 μl reaction in 96 wells plate. The master mix used for the reaction was made using the 2X AccuPrime SuperMix II (Thermo Fisher Scientific, Massachusetts, United States). A final concentration of 400 nM of each primer was used, and each respective well in the 96 wells plate received a separate primer set with a unique 10-base barcode (Fluidigm, South San Francisco, CA, USA; item #100-4876).
The unique barcodes in separate reactions were used for the positive control and a second notemplate control reaction with only Access Array Barcode library primers were used. The amplification conditions were 95 °C for 5 min, followed by 8 cycles at 95 °C for 30s, 60 °C for 30s and 68 °C for Sequence analysis. Bioinformatics was performed using the DADA2 pipeline. Data analysis was performed on unrarefied data. Amplicon sequence variants (ASV) were selected using DADA2 [57].
ASVs were filtered when they were not assigned to the Domain Bacteria or were assigned to the family Mitochondria, the class of Chloroplast or the phylum of Cyanobacteria/Chloroplast. Taxonomy was assigned using the Ribosomal Database Project 16S rRNA gene sequence database [58]. All subsequent ecological analyses were completed using PhyloSeq and other R packages [59].
Sequences were submitted to the National Center for Biotechnology Information (NCBI) Sequence Read Archive (https://www.ncbi.nlm.nih.gov/bioproject/) under the BioProject accession number PRJNA548382.
Statistical analysis. The sample rarefaction analysis was performed using the iNEXT R package [60]. The non-metric multidimensional scaling analysis (nMDS), based on the Bray-Curtis dissimilarity matrix, was analysed in PRIMER7 in order to test whether the bacterial community differs between the treatment and control groups. An analysis of similarity (ANOSIM) was performed between the treatment and control groups at various time intervals. The Shannon H′ diversity index was analysed in PAST. Mixed model ANOVA was carried out to determine the effects of time and treatment on the Shannon H diversity index of the microbiome using SPSS23. No violations of data sphericity were detected (Mauchly's test of sphericity: p > 0.05). Similarity percentages routine (SIMPER) was applied using the PRIMER7 software. The average Bray-Curtis dissimilarity in the structure of the microbial community between the control and treatment groups was calculated for each time point separately.
The contributions from each bacterial genus to the overall dissimilarity between the groups (with a 90% cut-off) are presented in the results.

Culture-dependent approach
Isolation of nicotine/anabasine-degrading bacteria. The nicotine/anabasine-degrading bacteria were isolated at two different sampling time points: (i) cloacal excreta were collected from the naïve birds, which were fed with artificial nectar at day 0 (0 weeks control group, 0WC; 0 weeks treatment group, 0WT; n = 8 at each group; hereafter, weeks control and weeks treatment will be mentioned as WC and WT, respectively) ( Fig. 1); (ii) four weeks after starting to add nicotine and anabasine to the special food in the treatment group (4WC and 4WT) (Fig. 1). The excreta were collected as described above. Collected samples were immediately cultured. For enriching nicotine-or anabasine-degrading bacteria, we used the M9 minimal medium (Na 2 HPO 4 .7H 2 O -64g, KH 2 PO 4 -15g, NaCl -2.5g, H 2 O -1L), sterilized by autoclaving. To 200 ml of this mixture, we added 700 ml water, 2 ml sterile 1M MgSO 4 and 100 µl 1M CaCl 2 ; and the whole solution was adjusted to 1 l with H 2 O, with the addition of either 0.1% nicotine or 0.1% anabasine as the only carbon and nitrogen sources. Samples of 100 µl of the collected faeces were incubated in 500 µl M9 minimal medium at 37 ºC for 3 to 6 hours. After this enrichment incubation, samples were inoculated on agar plates with the same medium but with the addition of 2% agar. Plates were incubated at 37 ºC for 5 to 7 days.
Nicotine-and anabasine-degrading bacterial colonies were picked and streaked five times on Luria-Bertani (LB, HiMedia Laboratories, Mumbai, India) agar plates. Their ability to grow on nicotine or anabasine as the only carbon and nitrogen sources was verified again by growing them on M9 agar plates with nicotine or anabasine as carbon or nitrogen sources. Pure bacterial isolates were kept in LB broth with 30% glycerol at -80 ºC.
The bacterial isolates were identified by amplifying and sequencing a 1501-bp internal fragment of 16S rRNA gene, in accordance with Senderovich et al. [61]. Purified PCR products were sequenced at MCLAB (South San Francisco, CA, USA) and analyses of all sequences was carried out using the EzTaxon website (http://eztaxon-e.ezbiocloud.net/) [62]. The sequences were deposited in the GenBank database under the accession numbers: MK348690-MK348835.

Declarations
The authors declare that they have no conflicts of interest. Tables   Table 1. Analysis of similarities (ANOSIM) for the pairwise test between all the treatments (See also Fig. 2). Samples were taken at 0, 2, 4 and 7 weeks; W, Week of sampling; C, control; T, treatment. The number before the letter W -indicates the week of sampling.  Table 3. A list of the genera from the sunbird's excreta Illumina sequencing results that their abundances increased from time zero (0WT) to the 4 th week sampling time (4WT) and that have the ability to degrade nicotine or anabasine (evidence for the degradation abilities are either from the current study or from the literature). The table shows the relative genera abundances in the birds' excreta at 0WT and 4WT sampling times. In parallel to the Illumina analyses, we have isolated at the same dates, bacterial strains that were able to use nicotine and or anabasine as a sole carbon and nitrogen sources (these genera are marked in bold; see also Table 2). After 4 weeks of nicotine and anabasine addition to the birds' food, these genera proliferated from a relative total abundance of only 0.3% to 52.7% out of the bacterial excreta community, demonstrating the effect of the edition of nicotine and anabasine to the birds' food. Figure 1 The setup of the feeding experiment. 16 naive sunbirds were captured and divided into two groups (i) Control group that were fed with sunbird artificial nectar free of alkaloids (ii) Treatment group that were fed with artificial nectar with the addition of nicotine (0.5 ppm) and anabasin (5 ppm). Excreta samples were collected from both groups at 0, 2, 4 and 7 weeks. All samples were used for generating a 16S rRNA libraries for bacterial community analyses using Illumina sequencing. Excreta that were collected at day 0 and at 4 weeks from the treatment group were used for culturing bacterial isolates that can degrade nicotine or anabasine. W, sampling week; C, control; T, treatment Non-metric Multidimensional scaling analysis (nMDS) with Bray-Curtis dissimilarity matrix.
The figure demonstrates the dynamic changes in the gut microbiota composition of sunbirds from the different feeding groups. Control group (C) with no addition of secondary metabolites and treatment group with the addition of nicotine and anabasine (T). Each treated group contained 8 birds that were held in separate cages. Excreta samples were collected at day 0 and after 2, 4 and 7 weeks. Each sampling time is colored differently; day 0 -blue; week 2 -red; week 4 -pink; week 7 -green. All controls symbols are empty shapes and all treatment symbols are full shapes. Significant differences were found in the microbiota composition between the control and the treatments at each sampling time and also between the same treatment in the different time intervals (See Table 1 for the statistical analyses). W, sampling week; C, control; T, treatment The relative abundances of 10 genera that contributed most saliently to differences among the control and the treatment groups at various time intervals -(a) 2, (b) 4 and (c) 7 weeks.
Bars indicate means ± STD.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.
Supplementary Files.zip