The golden jackal (Canis aureus): interaction of fur and fecal microbiota, host genetics and animal traits

Abstract

The golden jackal (Canis aureus) is a medium-sized canid widely distributed across Europe and the Mediterranean region. As a synanthropic species, it often inhabits areas near human settlements and is associated with zoonotic diseases such as rabies. This study characterizes the golden jackal’s fur microbiota and examines its interactions with host traits, fecal microbiota, host genetics and environmental factors. Analysis of the 16S rRNA amplicons indicated that the golden jackal fur microbiota is dominated by four bacteria phyla (Proteobacteria, Firmicutes, Actinobacteriota and Bacteroidota). The microbiota composition varies significantly with geographic location, age class, exposure to oral rabies vaccines, canine distemper infection, and environmental conditions.

The fur microbiota patterns found in this study, mirror associations previously observed in golden jackal fecal microbiota [1]. The significant correlation between fur and fecal microbiota composition and genetic distance between individuals indicates a potential link between host genetics and microbiota diversity, reinforcing the hypothesis that host traits and environmental exposure shape microbial communities across different habitats. However, while the fur microbiome demonstrated a significant association with geographic location the genetic diversity, based on three representative genotypes, does not. This can be explained by the strong direct contact of the fur with the environment, maintaining a skin microbiome for a healthy skin, developing and regulating the immune system, thus override the genetic profile. Overall, our findings highlight the golden jackal’s adaptability as a synanthropic species and underscore the intricate relationships between host genetics, microbiota composition, and environmental factors. By elucidating these connections, this study contributes to a broader understanding of microbiome ecology in wildlife and its implications for host health and pathogen transmission.

Supplementary Information

The online version contains supplementary material available at 10.1186/s42523-025-00470-4.

Background

The microbiota describes the microbial communities that inhabit the mucosal body surfaces of animals and humans [2]. Microbiota mainly focused on humans and animal models, but expanded in recent years to veterinary medicine [3, 4] and wild animal conservation through captive and free-roaming wildlife [5–10]. The gastrointestinal tract (GI) microbiota is the most densely populated anatomical part of the mammalian where the microbial cells are thought to outnumber host cells [10]. Wildlife GI microbiota is influenced and altered by diet, different lifestyles, host genetics, habitat, seasonal variation and alterations associated with various diseases [11–15]. For example, environmental factors such as seasonal fluctuations habitat alterations, and fragmentation correlate with the microbiota diversity of black-backed jackals (Canis mesomelas) and wild primates [15, 16]. Other studies of wildlife microbiome, such as the case of wild rodents, identified the correlation with exposure to pathogen(s) [6, 17]. Hence, similar to humans, the study of wild animal microbiota can assist in the surveillance of animal pathogens, even before clinical symptoms, and can help in the health management of the habitat.

The skin microbiota comprises millions of bacteria, though it is assumed to have a lower biomass than gut microbiota [18]. The skin microbiome plays an important role in homeostasis maintaining healthy skin and regulating the immune system. The microbiome of both skin and gut can be altered in various skin diseases [9, 19–22]. The latest human microbiome studies focus on the gut-skin axis exploring the connection between the skin microbiome and the gut and its effects on dermatologic conditions [18, 19]. Animal skin and fur microbiome studies are scarce and focus mostly on dogs and cats in the context of pet diets and skin conditions [4, 21, 23, 24]. Studies of wildlife skin and fur microbiota are rare and conducted mainly on captive animals [25–27]. Variation in fur microbiome was found between species, sex, and geographic location [28]. Infection by pathogens, such as Sarcoptes scabiei mites in the case of three North American wild canids [coyote (Canis latrans), red fox (Vulpes vulpes) and gray fox (Urocyon cinereoargenteus)] have shown altered skin microbial dysbiosis [9] [9]. These findings demonstrated the ecological and host-specific factors that impact microbiota diversity and underscore the need for further research on the microbiome composition of species inhabiting human-modified environments.

The golden jackal (Canis aureus; GJ), a common medium-sized canid, widespread throughout the Mediterranean region, expanding into Europe [29, 30]. The Israeli GJ population went through a strong bottleneck during the early seventies but over the last decade, their population has increased dramatically following invasion to new geographical regions, such as the Negev, mostly in association with human settlements [31, 32]. The revival of the GJ population is due to its excellent reproduction rate (3–8 offspring annually) and its adaptability to a varied diet (vegetables and animals), especially from anthropogenic sources [29, 32–34]. Consequently, the GJ became a synanthropic species, a species that benefits from living in close proximity to humans yet remains beyond their control. The high density of the GJ population in Israel, 25 jackals per 10 km² [35], and its association with humans, makes it a potential exposure for many pathogens, especially zoonotic disease agents [36, 37].

Population genetic study of the GJ population in Israel indicated the presence of gene flow between individuals within and between geographic locations [33]. The movement between populations, increases the likelihood of pathogen transmission, suggesting that the genetic makeup of the immune system in the GJ population may play a crucial role in adapting to and coping with these transmission pressures. The pattern recognition receptor (PRR) families, especially the Toll-like receptors (TLRs) class may play a key role in the GJ respond by stimulating the innate immune system. TLR4 is known for its recognition of bacterial lipopolysaccharide (LPS) and its activation mainly leads to the synthesis of pro-inflammatory cytokines and chemokines [38]. For example, Inflammatory Bowel Disease (IBD) is a chronic debilitating disease that was diagnosed in both humans and dogs. In dogs, polymorphisms in the TLR4 and TLR5 genes were significantly associated with IBD [39]. TLR activity can affect the gut microbiota through pathways of immunity (such as innate immune responses) and metabolism (such as catabolic activity) [40]. In addition, pathogen-mediated selective pressures can also influence the genetic components of host immunity and can cause inter-individual variation in resistance to different pathogens [41]. Despite the significance of this issue, studies on wildlife host genetics and immunogenetic traits remain scarce, primarily due to technical challenges.

Monitoring a synanthropic species and controlling both animal and human health requires a holistic approach. We recently described the Israeli GJ fecal microbiota and its associations with animal characteristics, geographic distribution, and burden of pathogens [1]. These findings stress the importance of further study looking at the GJ. Hence, the goal of the current study is to characterize the fur microbiota of the GJ in Israel in association with the gut microbiota and the genetic characteristics. We hypothesize that these factors are interlinked with the animal’s traits, including host genetics, gut microbiota, and habitat, as well as its geographical location. This approach can improve our understanding of the GJ as a successful synanthropic species with implications for future conservation programs.

Methods

Sampling and data collection

GJ sampling was performed in four different geographical regions in Israel (Fig. 1) as previously described [1]: Beit-Shean Valley (1), Ha-Sharon (2), Menashe Heights (3) and the Upper Galilee (4) following the predator control activity carried out by Israel Nature and Parks Authority (INPA). Culling of the GJ is conducted mainly for rabies control, surveillance and agricultural damage control. Beit-Shean Valley and the Upper Galilee are considered hot spots for rabies, while the Menashe Heights (region 3) is considered a moderate hot spot for rabies and Ha-Sharon (region 4), is considered to be free of rabies [42]. The four studied regions were characterized for proximity to domestic animals and anthropogenic food sources and annual climate measurement [1].

Fig. 1.

Sampling regions of GJ in the center-north of Israel: Beit-Shean valley (Region 1; ~347 km²), Ha-Sharon (Region 2; ~258 km²), Menashe Heights (Region 3; ~115 km²) and the Upper Galilee (Region 4; ~100 km²)

In order to calculate the sample size for this study, we used the established relative abundance at phylum level from our fecal microbiome study [1]. We estimated the sample size needed to detect a decline of at least 3-fold between typical fecal phyla such as Bacteroidota and Fusobacteriota, between fecal microbiota and fur microbiota. Given a respective relative abundance of 39% and 24%, the sample size required for detecting a 4-fold decline between proportions [43] with 95% confidence interval and 80% power, was 58 fecal and fur samples for Fusobacteriota and 31 for Bacteroidota.

Sampling and data collection for each specimen was conducted shortly after culling of the jackal (usually minutes and up to one hour) and included: (1) General Information: gender (male/female), estimated age (puppy to old), body mass (kg), body score (1–5; emaciated-obese) and body length (nose to the tip of tail in cm); (2) Presence of ectoparasites and skin disease was also noted by inspection; (3) Biological samples: fur samples (plucks from axilla and dorsal flank), rectal swab (ESwab™; Copan Italia S.p.A, Brescia, Italy) and blood (Serum tube- VACUETTE^®, Greiner Bio-One, Kremsmünster, Austria; EDTA tube- BD Vacutainer^®, BD, Plymouth, UK). All samples were stored in a cooler immediately after sampling and transferred between 30 and 360 min to a deep freezer (-80 °C) for storage until sample processing. The carcasses of GJ were transferred to the Kimron Veterinary Institute for necropsy and further diagnostic tests.

Burden of pathogens

The blood and tissue samples obtained were tested at the Kimron Veterinary Institute for eight pathogens: Rabies (FAT - brain stem, RFFIT - Rabies antibody titer, tetracycline test from bone- exposure to Rabies oral vaccination test), Canine Distemper virus (PCR- brain), Brucella spp. (serological test), Leptospira interrogans (serological test), Coxiella burnetii (Q fever; serological test), Nespora caninum and Toxoplasma gondii (serological test) and internal parasite (feces and diaphragm- flotation method). Description of the diagnostic methods employed was previously described [1].

Host genetic characterization

DNA was extracted from whole blood samples (preserved with EDTA) using Maxwell^® RSC Blood DNA Kit (Promega™, Wisconsin, USA) following the manufacturer’s protocol. The extracted DNA was used for genetic characterization of the GJ individuals based on neutral nuclear markers (Short Tandem Repeats (STR; microsatellites)) and Toll Like Receptors (TLR) gene regions. Negative controls were included through the entire process to determine the authenticity of the results.

Nuclear markers: All samples were genotyped using the Canine Genotypes™ Panel 1.1 (Thermo Scientific™, Vilnius, Lithuania). The panel included 18 autosomal STR loci and one sex chromosome locus, Amelogenin. The positive amplifications were genotyped on an ABI3130xl Genetic Analyser (Applied Biosystems, Foster City, California, United States) using the Orange DNA Size Standard (MCLAB). Identification of peaks and detection of allele length was performed using GeneMapper^® software (version 3.7; Applied Biosystems, Foster City, California, United States).

The genotype panel result was analyzed using GenAlEx 6.5 [44], and Structure [45] software’s in order to determine the individual identification, the subpopulation structure, the genetic diversity within and between subpopulation and the rate of gene flow. Relatedness among the sampled GJ was determined using a relatedness estimator W [46] in kingroup-v2 [47].

Toll Like Receptors: Three TLR gene regions, TLR2, TLR4 and TLR7, were studied to understand the immune response of GJ to pathogens and specific microbiome profile. The three TLR gene regions [TLR2 (338 bp), TLR4 (312 bp) and TLR7 (918 bp)] were Sanger sequenced and analyzed in order to determine the genetic variation within the GJ cohort and the association of the immune response system of GJ specimens with their pathogens and microbiome. In the lack of published GJ whole genomes the obtained sequences were compared to each other and to the domestic dog genome (NIH database, CanFam3.1; [48], Table S1). For each individual at each gene region both strands (sense and antisense) were Sanger sequenced, visually checked and consensus sequences were obtained. The TLR4 gene region targeting the mutations associated with inflammation was amplified using the published primer set [39] (Table S1). Successful amplifications were Sanger Sequenced using ‘EPPiC Fast’ (A&A Biotechnology, Gdynia, Poland) to purify the PCR products and cycle-sequenced using ‘BigDye’ (Applied Biosystems, Foster City, California, United States). Sequences were obtained on an ABI PRISM 3730xl DNA Analyzer (Thermo Scientific™, Vilnius, Lithuania). The two DNA strands of the same gene region for the same individual were assembled together for authentication using Sequencher software (Version 5.4; Gene Codes Corporations). A consensus sequence representing the gene region of the individual was obtained and used for further analysis. Bioinformatic analysis focused on identification of mutations, especially those that are associated with clinical diseases such as TLR4 with GI/IBS [39]. Genetic diversity and number of genotypes was conducted using DNAsp software (Version 6) [49]. Phylogenetic relationships between individuals representing four geographical regions were conducted for concatenated sequences representing each individual’s three TLR gene regions with nine additional representative species as an outgroup. Sequences for the outgroups were obtained from the “DNA zoo” web page [50] and the NCBI dataset (https://www.ncbi.nlm.nih.gov/datasets/, accessed in July 2023), for the following species: Vulpes vulpes, Vulpes lagopus, Nyctereutes procyonoides, Urocyon littoralis, Otocyon megalotis, Chrysocyon brachyurus, Lycaon pictus, Canis lupus dingo and Canis lupus. Reciprocal best-hit BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi, accessed in July 2023) was used to gain the fragments of interest for each TLR sequence. The three TLR’s representative sequences (TLR2, TLR4, and TLR7) were concatenated using the SEDA tool (Version 1.5.0) [51]. The concatenated sequence was viewed in UGENE (Version 49.1) [52] and aligned with the MAFFT (Version 7) [53] using default parameters. The concatenated alignment for 74 jackals substituted with nine outgroup sequences for the related TLR’s fragments resulted in 1548 positions aligned with gaps. The alignment farther was analyzed using IQ-TREE 2 [54] tool with the following parameters: MFP 1000 and alrt 5000. After auto-testing the models, TN + F + I was chosen according to BIC for the tree model. The consensus tree was viewed using iTOL online webpage (Version 5) [55], rooted using the Vulpes sp. branch, and colored using a default script according to the geographical region information in this study.

Fur and fecal microbiome analysis

The methodology used for fecal microbiome analysis was described in our recent study [1]. The following methodology pertains to the processing of fur samples and analysis of both fur and fecal microbiota.

DNA extraction and amplicon sequencing

DNA extraction from fur samples was conducted using DNeasy PowerSoil (QIAGEN^®, Hilden, Germany) kit according to manufacturer’s instructions. Negative control samples were added to rule out contamination during the extraction process. DNA extracts, including the extraction and amplification negative controls, were subjected to two rounds of amplification to prepare the libraries for sequencing. The first amplification of 20 cycles was performed on the V3-V4 region of 16S rDNA (amplicon size ~ 300 bp) using 16S rRNA primers from the Earth microbiome project (515F: 5’GTGCCAGCMGCCGCGGT3’, 807R: 5’GGACTACHVGGGTWTCT3’), with universal adapters CS1 and CS2. The second amplification of 10 cycles was performed using the Access Array Barcode Library for Illumina Sequencers from Fluidigm. The final library concentration for each sample was determined using the Qubit (Invitrogen) and the Denovix dsDNA High Sensitivity Kit according to the Denovix kit instructions. The size of each library was determined by TapeStation analysis using the D1000 Screentape according to the manufacturer’s instructions. Sequencing of all amplicons was carried out on an Illumina MiSeq sequencing platform using a Miseq V2-500 cycle kit to generate 2 × 250 paired-end reads) (Illumina, San Diego, CA, USA).

Illumina paired-end sequence data (as FASTQ files) were processed using the QIIME2 software package (ver. 2021.8) and its plugins [56]. Specifically, the ‘demux’ plugin was used to import the demultiplexed paired-end sequencing reads and to create the ‘artifact’ file (i.e. QIIME2 data format required for subsequent analyses). Read merging, adapter and quality trimming, identification of chimeric sequences, and clustering of sequences with 97% similarity threshold to amplicon sequence variants (ASVs) was conducted with ‘dada2’ plugin [57]. Our quality requirement for sample inclusion was raw read count per sample > 2,000, quality scores of reads > 20, and percentage of chimera sequences < 30%. Taxonomic annotations of ASVs with 97% similarity was assigned using the SILVA reference database [58] (https://www.arb-silva.de/, version: silva-138-99-nb-classifier, date of access: 28.12.2021). The final ASV table was rarefied at 3,200 sequences per sample, followed by downstream analyses including alpha and beta diversity, and differential abundance using the appropriate QIIME2 and ‘R’ plugins.

Fur microbiome data analysis

Alpha diversity was assessed using observed ASVs, Shannon index, evenness and Faith’s PD tests. Statistical significance between tested groups was assessed using the Kruskal-Wallis test. Beta diversity was assessed using the Bray-Curtis dissimilarity index, weighted-UniFrac and unweighted-UniFrac metrics, and tested for statistical significance using the PERMANOVA test. Mantel test was used to assess beta-correlation between quantitative variables with Bray-Curtis and unweighted-UniFrac dissimilarity indices. Relative abundance of specific taxa in significant beta correlations was assessed using LEfSE (Linear discriminant analysis effect size) analysis (using the R package microeco). Significant results were considered as LDA (Linear discriminant analysis) > 3. False Discovery Rate (FDR) adjustments (Benjamini-Hochberg (BH)) were applied to p-values for all statistical tests.

Comparison of the golden jackal fur microbiota to gut microbiota

We sub-sampled our published fecal microbiota dataset [1] to include only samples for which matched fur microbiota samples were available, for a total of 66 paired fur and fecal microbiota samples. The comparison was made through phylogenetic levels using Venn diagrams, alpha diversity (Shannon index) and beta diversity (Bray-Curtis dissimilarity index) (using the R package microeco). We tested correlations between fur and fecal microbiota using a Mantel test that was used to assess beta-correlation between quantitative variables with Bray-Curtis, unweighted-UniFrac and weighted-UniFrac dissimilarity indices.

Correlation analysis between the golden jackal fur and gut microbiota to golden jackal genetic characteristics (STR and TLR’s genotyping)

We tested correlations between the GJ microbiota (whole fecal and fur dataset) and host genetics using pairwise comparison of individuals’ microbiome dissimilarity using Bray-Curtis and unweighted UniFrac metrics to genetic similarities of STRs, based on pairwise relatedness, following the Lynch and Ritland estimator (LRM) [59] and single nucleotide polymorphisms (SNPs) of TLR2 and TLR7 sequences. We used ANOVA tests on individual distance matrices described to test correlations between fur and fecal microbiota and host genetics (STRs and TLRs).

Results

Golden jackal cohort metadata - geographic features, host traits and pathogen burden

The sampling was conducted between 2019 and 2020, during which 111 GJ specimens were collected and the metadata collected with them including geographic features (Table S2, Fig. 1); specimens information (Table S3); and pathogen prevalence (Table S4) is available in the supplementary materials.

Golden jackal microsatellites (STR)

All 111 specimens of GJ were amplified and genotyped using the Thermo Scientific Canine Genotypes Panel 1.1 panel. Out of the 19 STR’s loci the amelogenin and loci FH2054 were excluded from the analysis, due to the fact that FH2054 was found to be a tetranucleotide which could not be analyzed together with the rest of the STRs The amelogenin loci are included in the panel determine the sex in cases where it was unknown, but it was not used in the genetic profile of neutral markers. In this study, sex was determined by morphological observation of the carcass that was taken to necropsy and confirmed by the amelogenin STR. Hence, the profile of each individual was determined by 17 polymorphic loci (10–16 alleles per locus) and was found to be unique. Comparison between individuals based on the location of sampling (region 1–4) showed no association, low Fst values (0.005, P < 0.05) indicated very low variation between subpopulations and high variation within individuals of the same location. PCoA analysis indicated that there was no structure associated with the geographical location of sampling (Fig. 2A) and supported by the Assignment test indicating that only 53% assigned according to their location group. In addition, the STRUCTURE analysis designated the presence of three genotypes (K = 3) not related to the geographical location (Fig. 2B). The majority of individuals clustered together into one group representing GJ individuals from all four geographical regions, suggesting gene flow among them regardless of their habitat. Furthermore, these findings shed light on the presence of three main genotype profiles among the GJ population in the North of Israel (Fig. 2C).

Fig. 2.

(A) PCoaA analysis demonstrates no variance between four region groups per low Fst values (P < 0.05). (B) Harvester Plot indicating the number of optimal groups represented in the data. X = K = Number of genotypes and Y = mean of estimated Ln. probability of data. (C) STRUCTURE bar plot demonstrates bayesian clustering analyses for GJ as three main genotype profiles (K = 3). X = GJ individuals and Y = the estimated membership coefficients (Q) for each individual. Plot emphasizing the lack of structure related to the geographical location of sampling (X = individuals noted by region 1–4)

Analysis of golden jackal TLRs

The number of consensus sequences, representing high quality sequences, used in the analysis varied between the three gene regions: TLR2 was represented by 104 sequences, TLR4 by 96 sequences and TLR7 by 82 sequences.

TLR2

Alignment of all 104 consensus sequences indicated the presence of 22 different genotypes. Ten individuals had unique genotypes while the majority (n = 94) of samples were represented by 12 genotypes. Comparison between sequences representing individuals from four different regions found that the samples representing geographical regions 2, 3 and 4 vary from each other (Gst = 1) while samples representing geographical region 1 do not vary from the other samples (Gst < 0.007).

TLR4

All 96 consensus sequences obtained in the study mainly represented one genotype. In order to determine the presence of mutations associated with IBD the GJ sequences were compared to the dog genome. Comparison to dog genome of mutations associated with IBD found in GJ highlighted four mutations:

M1 mutation (position 11:71,365,652; non-synonymous mutation A > T associated with IBD directly): 92 individuals carry the wild type (WT) and four individuals (R08, R11, R15, R29) are heterozygous to the mutation (A/T, W).

M2 mutation (position 11:71,365,810; non-synonymous mutation A > G) all sequences were identical carrying the WT (A).

M3 mutation (position 11:71,365,875; synonymous mutation A > G) all sequences were identical carrying the mutation (G).

M4 mutation (position 11:71,365,888; non-synonymous mutation G > A) 92 individuals carry the wild type (WT) and the same four individuals that were heterozygous to M1 mutation (R08, R11, R15, R29) were heterozygous to the mutation (A/G, R).

The four individuals that carry the heterozygote profile in two mutations (M1 and M4) are from region 1 (R08, R11 and R15) and region 4 (R29) regions where the prevalence of Rabies is high.

TLR7

Alignment of all 82 consensus sequences indicated the presence of 18 different genotypes. Eleven individuals had unique genotypes while the majority (n = 71) of samples were represented by seven genotypes. Comparison between the genotypes and origin of the sample (geographic regions) indicated no polymorphism between the regions.

In order to understand the relationships between the GJ individuals, a concatenated consensus sequence was established for 74 specimens that had all three TLR gene fragments, for a total of 1568 bp. Phylogenetic analysis of the 74 GJ specimens together with an outgroup of nine Canidae species (Canis lupus and Vulpes vulpes) showed that all the GJ shared the same clade without any structure (bootstrap values less than 75%), especially there was no association to geographical origin as individuals from different locations cluster together (Fig. 3). These findings support the STR results indicating gene flow among GJ individuals.

Fig. 3.

Phylogenetic relationship among 74 GJ specimens. Cladogram of the combined analysis of TLR2, TLR4 and TLR7 of 74 studied GJ individuals. Geographic regions are marked in color: region 1 = green, region 2 = red, region 3 = yellow and region 4 = blue. Outgroup consisted of nine Canidae species (from Canis lupus to Vulpes vulpes) colored in light blue

Golden jackal fur microbiome analysis

We conducted 16S rRNA amplicon sequencing on fur samples from the entire cohort (n = 111). However, due to low yield in 30 samples, the final analysis was performed on 81 samples. The fur sequencing output consisted of a total of 2,349,122 sequences after demultiplexing. Read abundances per individual ranged from 4765 to 293,218 with an average of 29,001.51 ± 31,621.61. After completing the DADA2 pipeline, read abundance per individual ranged from 431 to 191,057 with an average of 13,233 ± 21,788. For downstream analyses the cutoff was a minimal 3,200 sequences per sample after rarefaction; only 67 samples met this threshold.

Golden jackal fur microbiome taxonomic profile analysis

Bacterial taxa varied largely in their proportions between individuals. Overall, 40 phyla were found between all individuals but only 16 phyla exhibited over 0.1% abundance, and these accounted for more than 99.68% of relative abundance (Table 1; Fig. 4). The most abundant bacterial phyla were the Proteobacteria (45.63%, range 2.81–83.88%), Firmicutes (29.22%, range 0.84–93.41%) and Actinobacteriota (12.84%, range 0.03–46.44%). The most abundant bacterial families were the Moraxellaceae (17.62%, range 0-76.44%), Staphylococcaceae (7.8%, range 0-89.02%), Planococcaceae (4.38%, range 0-29.61%) and Bacillaceae (4.2%, range 0-35.93%). In general, in almost all individuals, 380 bacterial families were found but only 86 were present above 0.1% abundance accounting for 95.56% of total abundance (Table 2; Fig. 5).

Table 1.

Relative abundance of main phyla (> 1%)

Phylum	mean	min	max
Proteobacteria	45.63	2.81	83.88
Firmicutes	29.22	0.84	93.41
Actinobacteriota	12.84	0.03	46.44
Bacteroidota	6.71	0	47.61
Fusobacteriota	1.24	0	22.28
Deinococcota	1.07	0	13.08
Chloroflexi	0.30	0	3.76
Myxococcota	0.21	0	2.71
Planctomycetota	0.19	0	6.22
Acidobacteriota	0.18	0	1.50
unidentified	1.48	0	16.26
other	0.94	0	21.1
Total	100.00

Fig. 4.

Relative abundance of top eight most abundant bacterial phyla found in the fur of studied GJ specimens, clustered by geographical regions (1–4)

Table 2.

Relative abundance of leading families

Family	mean	min	max
Moraxellaceae	17.62	0	76.47
Staphylococcaceae	7.80	0	89.02
Planococcaceae	4.38	0	29.61
Bacillaceae	4.20	0	35.93
Enterobacteriaceae	3.70	0	62.14
Oxalobacteraceae	3.34	0	31.12
Micrococcaceae	2.99	0	24.69
Clostridiaceae	2.88	0	66.81
Sphingomonadaceae	2.34	0	11.47
Pasteurellaceae	2.33	0	55.64
Pseudomonadaceae	2.02	0	13.78
Erwiniaceae	1.73	0	46.59
unidentified	3.23	0	18.53
Other	41.44	0	796.12
Total	100

Fig. 5.

Relative abundance of top eight most abundant bacterial families found in the fur of studied GJ specimens, clustered by geographical regions (1–4)

Bacterial genera analysis revealed 876 genera, from which 115 were above 0.1% and accounted for 89.47% from total abundance. The most abundant genera in the GJ were Acinetobacter (13.26%, range 0-74.84%), Staphylococcus (5.36%, range 0-89.01%), Bacillus (3.8%, range 0-35.931%), Massilia (3.02%, range 0-29.08%) and Psychrobacter (2.23%, range 0-22.37%).

Alpha diversity analysis

Alpha diversity, using Faith-PD, between geographical region groups is demonstrated in Fig. 6A. Significant differences were found between geographical region groups (Faith-PD; H = 17.94, P = 3.9 × 10^− 3) and body condition, where thinner specimens had lower diversity (Shannon index; H = 4.4, P = 0.03; Evenness; H = 6.888, P = 8 × 10^− 3) (Figure S2A). Age-class was removed from analysis due to scarcity of samples. Sex was not found as a significant contributor to alpha-diversity in (Faith-PD; H = 0.35, P = 0.69).

Fig. 6.

Alpha diversity (A) per Faith’s PD between regional groups (1–4); (B) per Shannon index between negative to positive bone tetracycline test specimens

For pathogen burden, negative specimens in the bone tetracycline test had significant lower diversity compared to positive specimens using the Shannon index (H = 6.82, P = 9 × 10^− 3) and Evenness (H = 7.69, P = 5.3 × 10^− 3) (Fig. 6B). Canine distemper (Faith-PD; H = 4.37, P = 0.04) was also found as significant factor contributes to alpha-diversity with higher diversity in the positive specimens (Figure S2B). Skin disease (Faith-PD; H = 0.41, P = 0.52) and proximity to domestic animals and anthropogenic food sources were not found to contribute to alpha-diversity in all comparisons.

Spearman’s correlation coefficient was used to test correlation between the quantitative measurements (GJ body and climate measurements) and alpha diversity (using the Shannon index) (Table S5). We found a significant positive correlation in body weight (Spearman value = 0.12 and P = 0.04) and warmest month temperature (Spearman value = 0.31 and P = 0.01). Other quantitative measurements were not found significant (Spearman value = 0.2 and P = 0.1 for length; Spearman value= -0.14 and P = 0.24 for annual Precipitation; Spearman value = 0.12 and P = 0.33 for annual mean temperature; Spearman value= -0.06 and P = 0.6 for coldest month temperature).

Beta diversity analysis

Beta diversity group significance was measured using Bray-Curtis dissimilarity index and unweighted and weighted Unifrac. A significant difference was observed between geographical region groups (Bray-Curtis; F = 2.15, P = 9 × 10^− 3) (Fig. 7A) while no significant difference was found between sex (Bray-Curtis; F = 0.73, P = 0.96). Age-class was removed from analysis due to scarcity of samples. Most of the pathogen burden was not found as a significant contributor based on the various statistical methods used for the analyses skin disease (Bray-Curtis; F = 1.13, P = 0.22), external parasites (Bray-Curtis; F = 0.88, P = P = 0.76), Distemper (Bray-Curtis; F = 1.11, P = 0.22) and Toxoplasma (Bray-Curtis: F = 1.06, P = 0.36). Bone tetracycline (Fig. 7B) was the only significant contributor to beta diversity in Bray-Curtis (F = 1.39, P = 0.04).

Fig. 7.

PCoA plots based on Bray-Curtis diversity dissimilarity metric demonstrated the differences (A) between GJ regional groups (1–4), (B) negative and positive for tetracycline in bone

Mantel correlation was used to test correlation between the quantitative measurements (GJ body and climate measurements) with beta diversity (using the Bray-Curtis and unweighted UniFrac metrics) (Fig. 8, Table S6). Significant correlation was not observed in the length (Bray-Curtis; Spearman value = 0.08, P = 0.1) but was weakly observed in weight (Bray-Curtis; Spearman value = 0.12, P = 0.02) of the specimens. Environmental quantitative measurements were found significant contributable to beta-diversity: annual Precipitation (Bray-Curtis; Spearman value = 0.19 and P = 0.01), annual mean temperature (Bray-Curtis; Spearman value = 0.14, P = 0.01), warmest month temperature (Bray-Curtis; Spearman value = 0.19 and P = 0.01) and coldest month temperature (Bray-Curtis; Spearman value = 0.1, P = 0.03).

Fig. 8.

Distance-based redundancy analysis (Db-RDA) based on Bray-Curtis diversity dissimilarity metric of the fur microbiota compositions between GJ climate measurements among regional groups

Marker-gene based LEfSE analysis

The LefSe analysis was used to identify microbiota compositions associated with variables that showed significant correlations between GJ features and beta-diversity. Nine significant taxa differentiated the geographical region: region 3 is enriched with Acinetobacter genus, region 4 with Acetobacterales order, Acetobacteraceae family, Polaromonas, Rubellimicrobium, Methylobacterium-Methylorubrum, Noviherbaspirillum, Buchnera genera and Methylobacterium goesingense species (Fig. 9).

Fig. 9.

LefSe analysis regional group 1–4. (A) Score of the linear discriminant analysis (LDA, significant threshold > 3). (B) Cladogram of LEfSE results

LefSe analysis between age-class of GJ specimens revealed 37 significant taxa between age groups (Fig. 10). Only the adult group was enriched with taxa with LDA score larger than 3: The genera Wenzhouxiangella, Ulvibacter, Sedimentitalea and the species Nitrosomonas halophila.

Fig. 10.

LefSe analysis age-class GJ specimens. (A) Score of the linear discriminant analysis (LDA, significant threshold > 3). (B) Cladogram of LEfSE results

The LefSe analysis between bone-tetracycline positive and negative specimens revealed 45 significant taxa between response groups (Fig. 11). The positive group was enriched with Chloroflexi phylum, Chloroflexia class, Propionibacteriales and Corynebacteriales orders, Corynebacteriaceae, Propionibacteriaceae, Nocardioidaceae families, and Corynebacterium genus while the negative group was enriched only with Pantoea genus.

Fig. 11.

LefSe analysis bone tetracycline positive and negative GJ specimens. (A) Score of the linear discriminant analysis (LDA, significant threshold > 3). (B) Cladogram of LEfSE results

Comparison between fecal and fur microbiota

We compared taxonomic abundance, alpha diversity, and beta diversity across the 66 paired microbiota samples. Phylum level comparison between GJ fecal and fur microbiome (Fig. 12A) demonstrated the dominance of Bacteroidota, Fusobacteriota and Firmicutes in the fecal microbiome and for the fur microbiome the dominance of Proteobacteria, Firmicutes, Actinobacteriota and Bacteroidota. We found that 70.5% of the taxonomic groups were present in both fur and fecal samples. (Fig. 12B), although the fur taxa were varied greatly from fecal. Regional group variation was minimal for fecal microbiome (Fig. 12C), as 85.3% taxa were shared, and the respective figures was lower (53.7%) for fur microbiome (Fig. 12D). The comparison between age-class groups demonstrated the variation was minimal for fecal microbiome (Fig. 12E) as 90.5% taxa were shared but varied for fur microbiome (Fig. 12F) as only 12% taxa were shared, mostly between adult to subadult (66.6%).

Fig. 12.

(A) Relative abundance of top ten most abundant bacterial phyla found in the fur and fecal of studied GJ specimens. Venn diagrams demonstrate relationships of taxonomic groups between fur and fecal (B), between regional groups ((C) for fecal and (D) for fur) and between age-class ((E) for fecal and (F) for fur)

Alpha diversity using Faith’s PD demonstrated higher diversity (P < 1 × 10^− 4) for fur microbiome than fecal microbiome (Fig. 13A). Regional group comparisons also demonstrate higher diversities for fur microbiome with exception of region 3 (Fig. 13B). Higher diversities were only demonstrated in the adult group of fur microbiome (Fig. 13C). Variation between fur samples was noted by significant differences (P ~ 0.04) that were found between fur regional and age-class groups and only one of regional fecal groups.

Fig. 13.

Alpha diversity per Faith’s PD between fur and fecal samples (A), regional groups (B) and age-class (C). Group distance plots based on Bray-Curtis diversity dissimilarity metric between fur and fecal samples (D), Regional groups (E) and age-class (F)

Beta diversity between fur and fecal samples (Fig. 13D) demonstrated notable distance from each other (P = 0.001) and also higher variation for fecal than fur, as fur samples demonstrates higher distance from each other. Those findings are similar between regional groups (Fig. 13E) and age-class (Fig. 13F) groups. The distance differences found significant (P = 0.001) for both regional and age-class groups except in fecal age-class groups. Mantel test was used in order to determine a correlation between fur and fecal microbiome (Table S7). Regardless of the geographical origin and the specimen age no correlation was found in all performed tests with beta diversity (using the Bray-Curtis, weighted UniFrac and unweighted UniFrac metrics) (P > 0.091).

Fecal and fur microbiota association with host genetic markers

The correlations between the GJ microbiota and host genetic profile indicated that genetic relatedness as determined by means of STRs was correlated with beta diversities of fecal and fur microbiota (per Bray-Curtis dissimilarity) (Fig. 14A, R= -0.053, P = 1.3 × 10^− 3; Fig. 14D, R= -0.054, P = 0.037). The correlation of the genotypes representing the TLR regions indicated that the fur microbiota correlated to both TLR2 and TLR7 (Fig. 14E, R = 0.089, P = 1.8 × 10^− 4; Fig. 14F, R = 0.21, P = 3.9E-10 correspondingly). The fecal microbiota on the other hand correlated to TLR2 genotypes (Fig. 14B, R = 0.047, P = 9.1 × 10^− 4) but not to TLR7 (Fig. 14C, R= -0.025, P = 0.19).

Fig. 14.

Pairwise dissimilarity between fecal (A-C) and fur (D-F) microbiome beta diversity and genetic relatedness (A and D) using LRM method; SNPs number of TLR2 (B and E); SNPs number of TLR7 (C and F)

Discussion

A novel characterization of free-ranging canid – the GJs fur microbiota was determined. The synthesis of this study findings with our previous study on the GJ fecal microbiome [1] and new insights of the host genetics demonstrate a multi-faceted approach to understanding the microbial ecology of this synanthropic species.

Fur microbiome composition

We found that the bacterial taxa in GJ fur, sampled from the axilla and dorsal flank, were primarily dominated by Proteobacteria, followed by Firmicutes, Actinobacteriota, and Bacteroidota. When comparing this microbiome profile to published studies on the dog skin microbiome, we observed a high similarity in bacterial taxa composition, though differences were noted in their relative abundance. These variations were associated with the anatomical site of sampling [4, 21, 60]. Compared to the fecal microbiome, fur microbiota exhibited greater diversity and taxonomic variability. While fecal microbial biomass is generally higher than that of skin microbiota [18], this trend may differ in GJ due to environmental influences such as climate and habitat exposure. Further exploration of this axis could lead to novel insights into microbiome-mediated disease mechanisms.

Environmental influences on fur microbiome

Gut microbiome is directly influenced by the host diet but indirectly variation can occur due to environment and host genetic variability [2, 5, 61]. On the other hand, fur microbiome is more subjected to external factors such as geographical location and environmental factors in the host habitat. The diversity of bacterial species in a microbial ecosystem and the variation of microbial communities between samples were obtained via alpha and beta diversity analysis of the GJ fur microbiome. The finding of significant differences correlated to the regional location, echoed similar trends observed in our prior fecal microbiome analysis [1]. This phenomenon of fur and skin microbiome variation (mainly by beta diversity) between different geographic populations was found in scarce studies of wild animals such as the humpback whales (Megaptera novaeangliae) [27] and Myotis bats - between caves and net catching [62]. Our findings can reflect geographic distance, environmental effect or different diet in each geographical area between the populations. To further study these regional differences, we used LEfSe to identify taxonomic distinctions among geographical regions [63]. Notably, Polaromonas, associated with cold environments, was relatively abundant in the fur of individuals from region 4, at the Upper Galilee, north of Israel with the coldest month temperature. Similar findings, the psychrophilic bacteria associated with cold, was observed on beluga whales (Delphinapterus leucas) skin [26] Additionally, Rubellimicrobium, linked to urban environments in dogs [64], was more prevalent in region 4, suggesting an interaction between environmental conditions and microbial populations. These findings highlight the significant role of environmental surroundings in shaping fur microbial communities.

Climate conditions also played a role in structuring the GJ fur microbiome, as annual precipitation and temperature (annual, maximal, and minimal) correlated with microbial diversity, particularly in beta diversity, although only weak positive correlation was found. The clustering of geographical regions in the RDA analysis (Fig. 8) further supports climate as a primary driver of regional effects. Similar correlations have been observed in human skin microbiomes, where climatic conditions influence microbial composition [65]. However, while regional differences were apparent in both the fur and fecal microbiomes, climate conditions did not significantly affect fecal microbiota composition in our previous study [1]. The impact of climate on gut microbiota in wildlife remains poorly understood [66, 67], and further research is needed to clarify these relationships.

Contrary to expectations, proximity to anthropogenic food sources, such as domestic waste and agricultural byproducts, did not significantly influence the fur microbiome. This may be due to the generalist dietary habits of GJ, which enable them to exploit multiple food sources, thereby mitigating direct microbial shifts linked to specific dietary exposures.

Host traits and fur microbiome

Host traits also contributed to variations in the fur microbiome. Body condition was a significant factor, as individuals with lower body condition scores exhibited reduced microbiome diversity. This aligns with prior observations in other canids, where poor body condition correlates with altered skin microbial communities [68]. While mange infection significantly affected the fecal microbiome in our previous study [1] and among coyotes [9] it did not induce notable changes in the fur microbiome. The lack of similar microbiome patterns in GJ may be explained by the gut-to-skin axis, where gut microbiota alterations influence systemic immune responses rather than direct shifts in the fur microbiome [18, 19]. Additionally, personal observations suggest that mange-infected GJ may recover under favourable conditions, such as abundant food availability (personal communication; R.Lapid). Unlike the fecal microbiome, the fur microbiome is closely correlated with the physical condition of the GJ, such as weight, health and overall body condition, which in turn are influenced by climate and environmental factors, including the availability of food.

Pathogen burden and fur microbiome

Pathogen burden also correlated with fur microbiome composition. While rabies and Brucella were not detected in our cohort, tetracycline biomarker analysis, used to assess oral rabies vaccine (ORV) uptake, revealed significant microbial differences between tetracycline-positive and -negative individuals. Tetracycline-negative specimens exhibited lower alpha diversity, while LEfSe analysis showed enrichment of resident skin bacteria in tetracycline-positive individuals, such as Propionibacteriales [69] and Corynebacteriales [70]. Correlation with the fecal microbiome found that specimens positive to tetracycline had significantly lower alpha diversity with clear dissimilarity in beta diversity as compared to negative specimens [1]. The Pantoea genus, which includes species with potential pathogenicity in humans and plants [71, 72], was more abundant in tetracycline-negative specimens. These results suggest that ORV uptake may influence fur microbiome composition, potentially through gut-to-skin microbial interactions, though further studies are required to elucidate this mechanism.

Canine distemper virus (CDV) infection was also associated with increased microbial diversity in the fur microbiome, mirroring findings from our fecal microbiome study [1]. Similar effects have been reported in other species, such as giant pandas (Ailuropoda melanoleuca), where CDV infection correlates with gut microbiome alterations [73]. These changes may be linked to the progressive immunosuppressive effects of CDV, which disrupt microbial homeostasis via the gut-to-skin axis [74]. Overall, similar to our fecal microbiome study [1], this suggests that pathogen burden can significantly influence microbial composition, which may affect host immunity and disease susceptibility.

Host genetics and microbiome interactions

The knowledge that the skin microbiome is developing and regulating the immune system, among humans and model animals was the initiative to genetic characterize toll-like receptor (TLR) genes associated with canid pathogens such as rabies. While most of the cohort carried the TLR4 wild-type genotype, four individuals (three from region 1, one from region 4) were heterozygous for inflammatory bowel disease-associated mutations [39]. TLR2 and TLR7 genotypes correlated with fur microbiota composition, whereas only TLR2 showed associations with fecal microbiota. Genetic variation in TLR2 was regionally structured, with specimens from region 1 differing from those in regions 2, 3, and 4. Two potential explanations for this pattern include selective adaptation to ORV exposure in region 1 or restricted gene flow due to geographical barriers. Although animal movement was not directly studied, previous telemetry research suggests that GJ home ranges typically span 1–10 km², with minimal long-distance dispersal near human settlements [35]. If a geographic barrier, such as the Jordan Valley, limits gene flow, this could explain the observed genetic differentiation. However, similar studies, such as movement tracking and genetic analysis of African golden wolves, (Canis anthus) are needed to substantiate this hypothesis [75].

Genetic analysis using neutral markers (STRs) and functional TLR genes revealed no significant genetic structuring by geographical region, indicating high gene flow within the cohort. However, significant correlations were found between microbiome composition and host genetic distance, consistent with findings in black-backed jackals [15] and humans [76, 77]. These results suggest that host genetics shape both gut and fur microbiota, reinforcing the importance of genetic factors in microbial community assembly.

Study limitations

There are several possible limitations to this study. The sample size of 67 fur microbiome samples, while informative, may not fully capture the microbial diversity across the entire GJ population. Fur samples are low biomass samples which led to exclusion of a substantial proportion of the samples due to low DNA concentration. A potential limitation is the use of 16S rRNA amplicon sequencing rather than whole genome metagenomics, which could enable strain-level metagenomics and the reconstruction of metabolic pathways, thereby expanding our knowledge and comprehension of the GJ fur microbiome and possibly also mycobiome. Since skin and especially fur studies are challenging, it is possible that some of our findings of the fur microbiome population originated from contamination or a transient microbiota, which represents animal exposure to the environment, including insects and not necessarily the permanent host microbiota. A limitation of the analysis studying the association between fecal and fur microbiota to host genetic markers is the potential impact of pseudo-replication, as each sample was compared to all the others without adequately accounting for this redundancy. This could obscure the true relationships between microbial composition and host genetic traits. To address this, future analyses could treat each pairwise comparison as a random variable, select representative comparisons for each sample, or apply resampling techniques to improve statistical rigor. Additionally, while multiple pathogen detection methods were employed, their sensitivity and specificity may have limitations compared to advanced molecular techniques. The GJ is a mobile nocturnal predator, making sample collection challenging and thus we were unable to test the dynamics of the microbiota over time because we lacked multiple sampling points.

Another limitation was the inability to track GJ movements, preventing confirmation of regional isolation. While previous radio-collared studies did not detect long-distance dispersal [35], further movement ecology research is needed to rule out potential interregional migration. Similar analyses in related species, such as African golden wolves [75], could provide valuable comparative insights.

Conclusion

Using a novel approach of studying the GJ fur and fecal microbiota and its relation to host traits, pathogen burden and host genetics, we found that host genetics together with environmental factors were associated with changes in the GJ microbiota. The identified associations between fur microbiota to many of the variables studied together with correlations related to the host traits point to the gut-to-skin axis as an explainable factor and deserve further study.

The fur and fecal microbiota correlation with ORV bioindicator suggest specific microbiome profiles might be associated with vaccine uptake and that further study of the microbiome can be used to improve rabies control through better targeting of the oral vaccine. Additionally, a better understanding of the GJ microbiota and the factors shaping it is crucial, as these factors may interact with various One Health aspects, including zoonotic disease dynamics, environmental health, and wildlife management. Overall, the GJ’s adaptability as a synanthropic species, combined with insights from its fur and fecal microbiome [1], enhances our understanding of the human-wildlife interface from a ‘One-Health’ perspective. These findings inform strategies for managing human-wildlife interactions, addressing ecological and health implications, and promoting coexistence in urbanized environments.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Abbreviations

GJ

Golden jackal

INPA

Israeli national and parks authority

STR

Short Tandem Repeats

TLR

Toll-like receptors

Author contributions

RL – field sampling, wet lab work, data analysis, drafting of manuscript. YM – data analysis. KJ– data analysis. RK – study design, field sampling. JMG – study design, drafting of manuscript, overall project management. GKBG – study design, critical review of manuscript.

Funding

No funding was received for this research. Dr. Roi Lapid received a PhD scholarship for delivering this project from the Kimron veterinary institute.

Data availability

16S rRNA gene reads deposited ate sequence read archive (SRA) under the bioproject number PRJEB71615.

Declarations

Ethics approval and consent to participate

All specimens were achieved during routine predator control of the INPA and disease surveillance of both INPA and the Kimron veterinary institute.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.