Salmonella prevalence and serovar distribution in reptiles: a systematic review and meta-analysis

Peer-reviewed literature was searched in the PubMed, Scopus and Web of Science databases. Relevant studies published between January 1, 1986 and March 7, 2023 were retrieved using the following search algorithm: ((reptile) OR (turtle) OR (tortoise) OR (terrapin) OR (crocodile) OR (alligator) OR (caiman) OR (snake) OR (lizard) OR (iguana) OR (gecko) OR (skink) OR (tuatara) OR (amphisbaenian)) AND (salmonella) AND ((serotype) OR (serovar) OR (subspecies) OR (population) OR (isolate) OR (prevalence)).

After manually excluding duplicated articles, two independent reviewers (P.S-B and C.M) used the Rayyan software [79] to screen titles and abstracts for studies that potentially met the inclusion criteria. The full text of each selected study was then retrieved and independently assessed for eligibility. Disagreements between reviewers were resolved through consultation with a third reviewer (M.C-A). Studies were excluded if we were unable to obtain the full-text article or retrieve missing data despite requests to the authors. For articles referring to the same study, only the most complete version was retained.

We aimed to analyze the available data on the prevalence of Salmonella spp. intestinal tract colonization and the distribution of Salmonella serovars in both wild and captive reptiles worldwide. To maintain consistency, we excluded all studies published prior to 1986, the year the current nomenclature for the genus Salmonella was introduced [37]. We included observational studies conducted on live reptiles, without restrictions on species, age group, or sex, that reported the presence or absence of Salmonella spp. in intestinal tract samples. All sampling methods of intestines (feces, cloacal swab, intestine dissection, etc.) and all detection methods of the genus Salmonella were considered eligible for inclusion. Studies were eligible if they explicitly described the total number of individuals tested and those that tested positive for Salmonella spp. detection.

Articles were excluded based on the following criteria: (1) articles written in languages other than English, Spanish, or French; (2) studies not conducted on a population of reptiles, including those focusing on environmental samples or reptile-derived products (e.g., eggs or meat); (3) studies investigating infectious agents other than Salmonella; (4) non-observational studies, such as reviews, letters not presenting original data, and single-case studies; (5) studies involving experimentally infected animals; (6) studies conducted on carcasses; (7) studies reporting Salmonella isolation in organs or tissues other than the intestinal tract; (8) studies involving animals that received antimicrobial treatment within the 3 months preceding the study; (9) studies using detection methods capable of identifying specific Salmonella serovars only, instead of the entire genus, or lacking a description of the detection method; and (10) studies where raw data was unavailable for calculating the prevalence of Salmonella spp.

A standardized, pre-piloted Microsoft Excel form was used to extract information from the selected articles. The following data were considered and extracted from each study: (i) Study identifier: first author, title, year of publication, collection period, country; (ii) Sample: higher reptile taxon (lizard, snake, turtle, crocodile, tuatara and amphisbaenian), common reptile name, reptile family, reptile species, living conditions (wild or captive), sample size, age, health status, clinical symptoms; (iii) Methodology: sample type, sampling technique, conservation method until testing, test method, test protocol, subspecies and serovar identification method, subspecies and serovar identification protocol; (iv) Results: number of Salmonella-positive reptiles, Salmonella subspecies, Salmonella serovar, Salmonella serovar frequency, administration of antibiotic treatment prior to testing, antimicrobial resistance detected (when tested).

To ensure consistency, reptile families and species were registered under their current names, according to the latest update from the Reptile Database [80].

The methodological quality of the included studies was assessed using the Critical Appraisal Checklist for Studies Reporting Prevalence Data of the Joanna Briggs Institute [81]. This tool includes nine parameters, of which eight are applicable to studies on animal populations: appropriate sampling frame, appropriate sampling method, adequate sample size, description of study subjects and setting, sufficient coverage of the identified sample for data analysis, valid methods for identifying the condition, standard and reliable condition measurement, and appropriate statistical analysis. Each item was scored as 1 point for a “yes” response and 0 for “no” and “unclear” responses, with a maximum score of 8. The quality of each study was rated as low (< 4), moderate (4 to 6), or high (> 6) according to the summary score (Additional file 1). Two reviewers (P.S-B and C.M) independently conducted the quality assessment, and disagreements were resolved by consensus.

A random-effects meta-analysis for proportion calculations was conducted in R [82] using the “meta” package [83] in order to estimate the prevalence of Salmonella spp. and S. enterica in reptiles with their respective 95% confidence intervals (95%CI). Comparisons of the prevalence estimates for Salmonella spp. and S. enterica between higher reptile taxa, wild versus captive reptiles, detection methods, and reptile families were evaluated using the Q-test for comparison of meta-analyses and the Bonferroni post-hoc test for pairwise comparisons. Heterogeneity among the included studies was assessed using the Likelihood Ratio Test and the Wald Test.

The 95%CI for the isolation frequencies of Salmonella enterica subspecies were calculated using the Exact binomial test in R. To statistically compare the isolation frequencies of Salmonella enterica subspecies, pairwise comparisons of proportions were performed. The Bonferroni correction method was employed to adjust p-values for multiple comparisons, ensuring appropriate control of the family-wise error rate. Additionally, a chi-square test with continuity correction was used to analyze the contingency tables for comparing the isolation frequencies.

A total of 1,488 references were retrieved from three electronic databases. After removing duplicates, the titles and abstracts of 735 records were examined for inclusion, leading to further screening of 219 articles in full-text. Ultimately, 179 studies met the eligibility criteria for data extraction and analysis (Fig. 1). The references for these 179 included articles are listed in Additional file 1.

The quality assessment of the 179 included studies revealed that 36.3% were classified as high quality (score 8: n = 18, score 7: n = 47), while the remaining 63.7% were of moderate quality (score 6: n = 74, score 5: n = 30, score 4: n = 10) (Additional file 1). No low-quality studies were identified.

The characteristics of the included studies are summarized in Table 1. A total of 23,411 reptiles were tested across the 179 studies, with 49.9% (n = 11,684) being wild animals and 48.4% (n = 11,338) living in captivity. Among the captive reptiles, 39.1% (n = 4,437) were pet reptiles from pet shops, breeders or households, while 31.8% (n = 3,608) were housed in zoos and wildlife rehabilitation centers. The diversity of the reptile taxon was well represented, as the sampled reptiles belonged to 51 of the 91 currently existing reptile families (Additional file 2) [80]. Most of the reptiles tested were lizards (42.2%, n = 9,884) from 26 of the 37 existing lizard families. Turtles followed, representing 36.1% (n = 8,457) of the sample and including members from 13 of the 14 turtle families. Snakes accounted for 12.9% (n = 3,014) of the tested animals, from 9 of the 30 existing snake families. Tuataras from the Sphenodontidae family constituted 3.2% (n = 753) of the sampled reptiles, and crocodiles made up 2.8% (n = 666) from 2 of the 3 crocodile families. While one study collected samples from amphisbaenians, their number was not specified in the article [84].

The studies covered a total of 56 countries and territories across all continents (Fig. 2). The most common sampling locations were Asia and Europe, accounting for 28.2% (n = 6,592) and 27.9% (n = 6,536) of the samples, respectively, followed by Oceania with 17.8% (n = 4,178), Central and South America with 11.8% (n = 2,762) and North America with 11.6% (n = 2,704). Only 3% (n = 639) of the tested reptiles were from Africa (Additional file 3). Specifically, the U.S.A. was the most frequently studied country with 27 studies, followed by Italy with 15, Spain and Brazil with nine each, as well as Poland, Australia and New Zealand with eight each. Notably, most reptiles tested in Europe were kept in captivity (81%, n = 5,288), whereas most reptiles sampled from the other continents were wild (Fig. 2).

The health condition of the reptiles under study was explicitly described for only 43.9% of the animals sampled, with 40.5% identified as healthy (Table 1). Due to this limited information, the health status will not be considered further in the subsequent meta-analysis, and the estimated pooled prevalence calculated in this study is deemed representative of the overall reptile population.

Intestinal Salmonella spp. colonization was most commonly identified through the collection of cloacal swab (45.8%, n = 10,729), feces (31.8%, n = 7,438) or a combination of both methods (6.3%, n = 147). Another frequently employed method involved direct sampling of the intestines post-necropsy, accounting for 9.1% (n = 2,139) of the samples (Table 1).

Salmonella isolation through cultural methods was the preferred technique for identifying Salmonella spp. in the collected samples, used for 97.0% (n = 22,700) of the tested animals (Table 1). The International Organization for Standardization (ISO) provides a protocol for detecting Salmonella spp. in animal feces and swabs, which involves an initial pre-enrichment step in non-selective broth, followed by enrichment in selective broth, culture plating on selective agars to isolate the bacteria, and finally confirming isolated Salmonella spp. colonies through biochemical tests [85]. This ISO method was the most frequently employed in the studies analyzed, accounting for 54.4% of the samples. The second most commonly used isolation technique (37.1%, n = 8,689) omitted the pre-enrichment step, directly inoculating samples into a selective enrichment broth before plating on selective agars. In 4,5% (n = 1,056) of cases, samples were plated directly onto agars. Detection of Salmonella spp. through PCR, with or without a prior enrichment step in culture broth, was employed for only 1.5% (n = 361) of the samples collected.

The pooled prevalence estimates of Salmonella spp. colonization in the reptile population and its various subgroups are presented Table 2, while the p-values for pairwise comparisons are provided in Additional file 4. The meta-analysis results revealed that the overall pooled prevalence of Salmonella spp. in reptiles was 30.4% (95%CI: 27.4–33.6%). Analysis of heterogeneity showed a high degree of variability among the studies (I² = 63.1%, p < 0.01). Stratified meta-analyses identified several factors contributing to this variability.

The prevalence of Salmonella spp. colonization varied significantly across the different reptile taxa (Table 2). Snakes exhibited the highest prevalence at 63.1% (95%CI: 57.4–68.4%), followed by lizards at 33.6% (95%CI: 28.6–39.0%, p < 0.001). Turtles and crocodiles had similar prevalence rates of 11.2% (95%CI: 8.8–14.2%) and 10.5% (95%CI: 5.7–18.6%), respectively (p = 0.8). Notably, all four studies on Salmonella spp. isolation in tuataras failed to detect the bacteria in a total of 753 animals.

Furthermore, within the turtle taxon, sea turtles and tortoises showed a significantly higher prevalence of Salmonella spp. (18.8%, 95%CI: 9.6–33.6% and 16.3%, 95%CI: 11.3–23.0%, respectively) compared to freshwater turtles (7.6%, 95%CI: 5.1–11.2%, p < 0.05) (Table 3 and Additional file 4).

The prevalence of Salmonella spp. colonization varied based on the housing environment, with captive reptiles showing significantly higher carriage rates (37.8%, 95%CI: 34.3–41.4%) compared to wild ones (14.8%, 95%CI: 11.0-19.6%, p < 0.001) (Table 2). This difference was observed across all higher reptile taxa, though it was statistically significant only for lizards. Captive lizards had a prevalence rate more than three times higher than wild lizards (46.7%, 95%CI: 41.0-52.4% vs. 13.4%, 95%CI: 8.8–19.9%, p < 0.001). In contrast, turtles displayed the smallest difference in Salmonella spp. carriage rates between captive and wild animals (11.9%, 95%CI: 9.0-15.6% and 10.6%, 95%CI: 6.3–17.3%, respectively, p = 0.7) (Tables 2 and 3).

When comparing the rates of Salmonella spp. isolation among the different detection methods, the ISO method, which involves pre-enrichment followed by selective enrichment and plate isolation, was found to have a significantly higher isolation rate (36.7%, 95%CI: 32.9–40.7%) compared to the direct placement in selective enrichment broth followed by plate isolation method (21.8%, 95%CI: 17.1–27.4%, p < 0.001) (Table 2). This demonstrates that pre-enrichment in a culture broth significantly enhances the sensitivity of detection.

Notably, although PCR-based detection methods were used in a small number of studies, significantly higher rates of Salmonella spp. prevalence were reported in reptiles tested using a protocol that included pre-enrichment and selective enrichment steps followed by PCR assay (49.4%, 95%CI: 41.7–57.1%) compared to the ISO method (p < 0.01), suggesting it may be a valid alternative option.

Pooled prevalence estimates of Salmonella spp. colonization were calculated for each reptile family with over 100 tested animals (Fig. 3, Additional files 4 and 5).

No heterogeneity was found among the studies within each of the five selected families of snakes, indicating that family is the main factor influencing variability in Salmonella spp. carriage in snakes (Additional file 5). The Viperidae family had the highest prevalence rate at 84.2% (95%CI: 67.8–93.1%), whereas the Boidae and Pythonidae had the lowest rates at 44.5% (95%CI: 32.8–56.9%) and 53.1% (95%CI: 42.5–63.4%), respectively.

For lizards, Salmonella spp. prevalence was assessed across eleven families, revealing notable differences. The four lizard families with the highest prevalence rates were Teiidae (62.6%, 95%CI: 41.5–79.8%), Varanidae (54.8%, 95%CI: 37.3–71.2%), Iguanidae (53.8%, 95%CI: 37.9–69.0%) and Agamidae (48.7%, 95%CI: 37.0-60.5%). These rates were significantly higher than those of the four families with the lowest prevalence rates, namely Diplodactylidae (8.7%, 95%CI: 1.7–33.8%), Scincidae (12.7%, 95%CI: 7.6–20.6%), Anolidae (15.6%, 95%CI: 5.7–36.3%) and Gekkonidae (23.6%, 95%CI: 14.9–35.4).

As previously mentioned, freshwater turtles were found to harbor significantly lower Salmonella spp. prevalence rates than sea turtles and tortoises. However, when comparing the prevalence rates among different families of freshwater turtles, substantial disparities were observed. The Trionychidae (33.2%, 95%CI: 19.5–50.5%) and Kinosternidae (28.8%, 95%CI: 13.2–51.9%) families exhibited prevalence rates that were significantly higher than the overall prevalence in turtles, as well as compared to the three lowest prevalence rates found in the families of freshwater turtles Chelidae (5.1%, 95%CI: 2.7–9.6%), Geoemydidae (6.2%, 95%CI: 2.4–15.3%) and Emydidae (8.4%, 95%CI: 4.9–14.0%).

Finally, there was no significant difference in Salmonella spp. prevalence between the two crocodile families Alligatoridae and Crocodylidae (8.7%, 95%CI: 3.6–19.2% and 12.8, 95%CI: 5.6–26.8%, p = 0.5). Notably, no heterogeneity was found among the studies within each family of crocodiles.

Salmonella species and subspecies were identified for 4,756 (74.1%) of the 6,416 reptiles that tested positive for Salmonella spp. (Additional file 6). Among these, only two reptiles (a free-ranging lizard and a captive turtle) tested positive for S. bongori [86, 87], while the rest (99.96%) carried the Salmonella enterica species. The most common cause of Salmonella colonization in reptiles was found to be from the subspecies S. enterica (68.5%), followed by S. diarizonae (13.2%), S. houtenae (6.7%), S. salamae (6.5%), S. arizonae (5.0%) and S. indica (0.08%) (Figure 4A and Additional file 6).

Statistically significant differences were observed when comparing the isolation frequencies of Salmonella enterica subspecies between wild and captive reptiles (Fig. 4A and Additional file 6). Colonization by S. enterica and S. houtenae was significantly more common in wild reptiles compared to captive ones (71.0% vs. 67.6%, p = 0.03, and 8.2% vs. 5.8%, p < 0.01, respectively), while S. diarizonae and S. salamae were more frequently detected in captive reptiles than their wild counterparts (14.3% vs. 11.7%, p = 0.02, and 7.8% vs. 3.9%, p < 0.01, respectively). No significant differences were observed in the isolation of S. arizonae and S. indica with respect to the living conditions of the reptiles (p = 0.34 and p = 0.66, respectively).

Interestingly, the comparison of the isolation frequencies of the different subspecies of Salmonella enterica within each reptile group revealed significant differences (Fig. 4B and Additional file 6). S. enterica was the most frequently reported subspecies in every type of reptile except for crocodiles, which predominantly carried S. arizonae (68.3%). Notably, the isolation frequency of S. enterica was higher in turtles (87.3%) than in lizards (70.1%), snakes (46.7%) and crocodiles (19.5%), with all these differences being statistically significant (p < 0.01). Following S. enterica, lizards were significantly most frequently colonized by S. houtenae (14.2%) and snakes by S. diarizonae (38.2%). These findings suggest a potential specialization of certain Salmonella subspecies towards particular groups of host reptiles.

A total of 542 distinct Salmonella serovars were identified across the included studies (Additional file 7). The greatest variety of serovars was found in snakes (248 different serovars), followed by lizards (232 serovars) and turtles (151 serovars). Only three studies focused on identifying Salmonella serovars in crocodiles, resulting in a total of 11 distinct serovars identified. Interestingly, a much wider diversity of serovars was observed in captive reptiles, with more than twice as many distinct serovars compared to wild reptiles (443 vs. 194 different serovars, respectively). Figure 5 shows the most frequently detected serovars among all positive reptiles and within each specific reptile group. S. enterica serovars Oranienburg, Newport, Weltevreden, Pomona and Muenchen were the most frequently linked to reptile colonization. Specifically, serovar Weltevreden was most frequently detected in lizards, Newport in snakes, Abony in turtles and Adelaide in crocodiles. Notably, Enteritidis and Typhimurium, which are the Salmonella serovars most commonly associated with human salmonellosis, ranked among the top ten serovars colonizing both reptiles overall as well as snakes and turtles specifically.

Due to their ability to infect both warm-blooded and cold-blooded animals, bacteria from the S. enterica subspecies are responsible for most reptile-associated salmonellosis cases. Thus, to further assess the risk of human salmonellosis posed by reptiles in general and by each higher reptile taxon in particular, a new database was established that exclusively included only studies conducting Salmonella subspecies identification following Salmonella detection. Out of the initial 179 studies, 111 studies were included in this subset, comprising a total of 12,333 reptiles. The pooled prevalence estimates of S. enterica colonization in the overall reptile population and in different subgroups are presented in Table 4, and the p-values for pairwise comparisons are provided in Additional file 8. The meta-analysis results from this new database revealed an overall pooled prevalence of S. enterica in reptiles of 35.1% (95%CI: 31.4–39.1%).

In concordance with the distribution of Salmonella enterica subspecies isolation frequencies in each higher reptile taxon described in the previous section (Fig. 4), the prevalence of S. enterica in snakes was significantly lower than the previously reported prevalence of Salmonella spp. (45.3%, 95%CI: 37.9–53.0%, and 63.1%, 95%CI: 57.4–68.4%, respectively) (Table 4). In addition, lizards showed a slightly but not significantly lower prevalence of S. enterica than snakes (38.5%, 95%CI: 32.1–45.4%, p = 0.19). Conversely, turtles exhibited a significantly lower S. enterica prevalence (26.0%, 95%CI: 21.1–31.7%, p < 0.01), followed by crocodiles (4.0%, 95%CI: 1.8–8.6%, p < 0.001).

In contrast to the general prevalence of Salmonella spp. in reptiles, there was no significant difference in the occurrence of S. enterica colonization between free-ranging and captive reptiles (33.8%, 95%CI: 36.1–42.4%, and 36.2%, 95%CI: 31.9–40.1%, respectively, p = 0.6). However, distinctive patterns were observed when comparing these rates within each higher reptile taxon. For lizards and turtles, the prevalence of S. enterica among captive animals was higher than their free-ranging counterparts, though this difference was statistically significant only within the lizard taxon (p = 0.03). In contrast, free-ranging snakes exhibited a nearly double rate of S. enterica colonization compared to captive snakes (71.9%, 95%CI: 50.7–86.5%, and 37.3%, 95%CI: 30.6–44.6%, respectively, p < 0.01).

Considerable disparities in Salmonella detection rates have been observed across studies investigating the prevalence of Salmonella spp. in reptile populations. Our systematic review and meta-analysis revealed that the detection method employed is a critical factor underlying this variability. The established protocols recommended by the World Organization for Animal Health (WOAH) and the World Health Organization (WHO) for isolating of Salmonella spp. from animal feces are based on ISO standards and involve a three-step cultural-based method: pre-enrichment in a non-selective broth, followed by enrichment in a selective broth and plating on selective agars, before confirming the presence of Salmonella spp. through biochemical and serological tests [89, 90]. Despite these guidelines being available since 1981 [91], the majority of studies included in our review (55.3%) did not adhere to these protocols. In particular, 41.3% of the studies used a two-step method that omitted the pre-enrichment step. Our meta-analysis revealed that studies utilizing the three-step cultural-based method conforming to ISO standards yielded significantly higher Salmonella spp. isolation rates compared to those foregoing the pre-enrichment step. This suggests that the omission of this pre-enrichment phase, which is believed to enable stressed or injured Salmonella bacteria to recover before exposure to selective media [90], may reduce the sensitivity of the detection method. Consequently, we strongly encourage researchers studying the prevalence of Salmonella spp. intestinal colonization in reptiles to strictly follow the WOAH and WHO guidelines when employing culture-based detection methods. A small number of studies (10 out of 179) used a PCR-based approach to detect Salmonella spp. in reptile intestinal samples. Interestingly, higher Salmonella prevalence rates were reported when the PCR assay was preceded by non-selective pre-enrichment and selective enrichment steps. This finding strongly supports the crucial role of these two initial culture steps in significantly enhancing the sensitivity of any method used for detecting Salmonella spp. in reptile intestinal samples. Furthermore, although this PCR-based protocol with pre-enrichment and selective enrichment was only employed in 3 studies, involving a total of 160 animals (150 wild and 10 captive turtles), the pooled prevalence of Salmonella spp. in reptiles tested using this method was significantly higher compared to other detection methods. These findings are particularly notable given that they were obtained in turtles, which the overall meta-analysis found to have the lowest Salmonella spp. prevalence alongside crocodiles. Therefore, these observations suggest that an appropriate PCR assay could be considered a valid complementary approach to the plating onto selective agars and biochemical confirmation steps recommended in the ISO method. This supplementary technique may help capture additional potentially false-negative samples. Given that this PCR-based alternative method would substantially reduce processing times, further studies comparing the efficacy of these two protocols for Salmonella spp. detection in animal feces or other sample types would be of great interest.

Our systematic review and meta-analysis identified several key factors influencing the prevalence of Salmonella spp. among reptile populations. Specifically, Salmonella prevalence was found to be affected by the reptile’s habitat, taxonomic group and family. Numerous prior studies have compared Salmonella colonization rates between captive and wild reptiles, with contradictory findings regarding the impact of captivity on prevalence. However, our meta-analysis conclusively demonstrated that captive reptiles harbor a significantly higher prevalence of Salmonella spp. intestinal colonization compared to their wild counterparts. These results are particularly concerning, as captive reptiles are the source of human reptile-associated salmonellosis. One hypothesis proposed to explain this phenomenon is that captive reptiles acquire Salmonella through direct or indirect contact with humans, domestic animals or contaminated food sources [47, 49, 92‐95]. For instance, outbreaks of reptile-associated salmonellosis have been linked to feeder mammals [96‐99]. According to this hypothesis, the distribution of Salmonella enterica subspecies colonizing captive reptiles would be expected to be enriched in S. enterica transmitted by humans or mammals. However, our findings revealed that the proportion of S. enterica was actually lower in captive Salmonella-colonized reptiles compared to their wild counterparts, resulting in a similar prevalence of S. enterica between captive and wild reptiles (see Fig. 4A; Table 4). These observations suggest that the higher Salmonella spp. prevalence reported in captive reptiles would more likely be the consequence of enhanced colonization and/or active shedding of Salmonella enterica subspecies other than S. enterica, rather than attributable to exposure to humans or food. Captivity-related chronic stress is known to induce immunodeficiency in reptiles, potentially increasing their susceptibility to Salmonella spp. colonization and shedding [47, 92, 100]. Additionally, the high density and/or diversity of reptiles in confined captive environments may also contribute to the spread of these contaminations [50, 87]. Specifically, our analysis indicated that captive reptiles exhibited statistically higher frequencies of S. diarizonae and S. salamae colonization compared to their wild counterparts. A plausible explanation is that the stress-induced immunodeficiency frequently observed in captive animals may predispose healthy reptiles to colonization by and shedding of these two Salmonella enterica subspecies, which could therefore be considered opportunistic pathogens in this context. Although S. diarizonae and S. salamae were found to be the second and third most frequently isolated Salmonella enterica subspecies in captive reptiles, respectively, after S. enterica, they are rarely responsible for reptile-associated salmonellosis cases in humans, with S. enterica, S. arizonae and S. houtenae being the three subspecies most frequently associated with this condition over the past two decades [101]. Furthermore, a greater diversity of Salmonella serovars was identified in captive reptiles compared to their wild counterparts. This observation can also be explained by the increased direct or indirect contacts between reptile specimens from the same or different species, particularly in the case of animals from pet shops and zoos, as well as by stress-induced immunodeficiency, both factors facilitating the circulation of numerous Salmonella serovars among captive reptiles. Collectively, these findings suggest that captivity-related stress and the increased contact between individuals appear as key factors in enhancing Salmonella spp. colonization in captive reptiles.

Our study revealed significant differences in the prevalence of Salmonella spp. across higher reptile taxa. Snakes exhibited a notably high prevalence of 63.1%, suggesting that Salmonella spp. is a common component of their normal gut microbiota. Lizards were the second most frequently colonized reptile group, with a prevalence of 33.6%, followed by turtles and crocodiles (11.2% and 10.5%, respectively). Several hypotheses have been proposed to account for the elevated Salmonella spp. prevalence in snakes. One potential explanation is their carnivorous diet, as snakes may acquire the bacteria through consumption of contaminated prey [61, 102, 103]. However, this dietary factor does not seem to apply to crocodiles, which are also carnivorous yet display the lowest rates of Salmonella spp. carriage. Another hypothesis posits that the predominantly ground-dwelling and terrestrial nature of snakes, particularly when confined in terrariums, increases their likelihood of exposure to contaminated substrates and fecal matter, thereby elevating their risk of Salmonella acquisition [103, 104]. Regarding the prevalence of Salmonella spp. in turtles, previous studies have suggested that seasonal variations, specifically the reduced feeding of turtles during their hibernation preparation, may contribute to the lower rates of Salmonella isolation observed in chelonians compared to other reptile groups [65, 69]. However, our review of the 104 studies on Salmonella spp. prevalence in turtles found no evidence of a higher frequency of sampling during the coldest months of the year (data not shown). Consequently, hibernation alone does not appear to contribute to the lower prevalence of Salmonella spp. in turtles compared to snakes and lizards.

The analysis revealed that tortoises had a significantly higher prevalence of Salmonella spp. compared to freshwater turtles. This may suggest that tortoises are more susceptible to intestinal Salmonella colonization or that fecal-oral Salmonella transmission is less effective in freshwater turtles. However, Salmonella spp. demonstrates high survival in aquatic environments and is frequently isolated from water sources, which can effectively mediate its transmission in aquatic animals [24, 105, 106]. Given that most studies in the meta-analysis identified Salmonella spp. through cloacal swabs, the difference in Salmonella detection between tortoises and freshwater turtles may be attributed to the shorter duration Salmonella persists in the cloaca in aquatic animals before being flushed out, whereas in terrestrial habitats Salmonella remains for longer periods and is directly transmitted between individuals [46, 51]. Supportive of this hypothesis is the observation that, along with freshwater turtles, crocodiles, which are aquatic or semi-aquatic, have the lowest prevalence of Salmonella spp. In contrast, the majority of snakes and lizards included in the meta-analysis are terrestrial [107] and exhibited the highest Salmonella spp. prevalence rates. However, this environmental pattern does not appear to extend to sea turtles, which display Salmonella spp. isolation rates comparable to tortoises, although the results for sea turtles exhibited high variability, potentially due to the limited number of studies and tested animals. The notable variations in prevalence estimates across turtle and lizard taxa suggest the involvement of additional factors, such as immune system or diet.

Variations in Salmonella spp. prevalence were observed across different reptile families within each higher reptile taxon, with some families displaying significantly higher colonization rates compared to others. Interestingly, while significant heterogeneity was observed between studies within each higher reptile taxon, no significant between-studies variability was detected when the meta-analysis was stratified by reptile family for 19 out of the 27 families examined (70%). These findings suggest that reptile family is a crucial factor in determining susceptibility to Salmonella spp. Further investigation into the underlying factors, whether genetic characteristics related to the immune system, environmental conditions associated with the living environment of different reptile families, or dietary factors, would provide valuable insights into the complex host-pathogen interactions between Salmonella spp. and its reptile hosts.

S. enterica accounts for approximately 99% of human Salmonella infections [37] and nearly 90% of reptile-associated salmonellosis cases [101]. Our analysis found S. enterica to be the predominant Salmonella subspecies colonizing reptiles. The meta-analysis of 111 studies that identified Salmonella species and subspecies following detection estimated a 35.1% prevalence of S. enterica in reptiles, highlighting the substantial zoonotic risk posed by reptiles as a source of human salmonellosis. Of particular concern are the three reptile taxa commonly kept as pets - snakes, lizards, and turtles. Snakes and lizards exhibited the highest rates of S. enterica colonization at 45.3% and 38.5%, respectively, while turtles had a 26.0% prevalence. Accordingly, the 15 Salmonella serovars most frequently identified in reptiles belonged to the S. enterica subspecies, several of which are significant public health concerns. The most common Salmonella serovar colonizing reptiles was Oranienburg, a host-generalist serovar that has been linked to numerous outbreaks of human salmonellosis worldwide [108‐113]. The second most frequent serovar isolated from reptiles was Newport, which has been reported as the third most common Salmonella serovar associated with human salmonellosis in the United States over the past two decades [5, 114] and the fifth most common on average in Europe since 2018 [115]. Following these, the most common Salmonella serovars found in reptiles were Weltevreden, Pomona, and Muenchen, all of which are host-generalist serovars responsible for multiple salmonellosis outbreaks in otherwise healthy individuals [108, 116‐123]. Importantly, the two Salmonella serovars most frequently associated with human salmonellosis globally, Enteritidis and Typhimurium [5, 7, 115], were also identified among the top ten serovars colonizing reptiles. Collectively, these findings indicate that reptiles constitute a considerable reservoir of human-pathogenic Salmonella serovars with high zoonotic potential.

The results of our systematic review and meta-analysis demonstrated that snakes exhibit exceptionally high rates of Salmonella spp. and S. enterica carriage, and serve as hosts to a remarkably diverse range of Salmonella serovars. Of particular concern, the serovars S. Newport, S. Typhimurium and S. Enteritidis were identified as the first, second, and sixth most commonly detected among Salmonella-colonized snakes, respectively. These findings underscore the significant hazard that snakes pose in terms of reptile-associated salmonellosis. As such, snake pet owners and the pet industry should exercise caution and appropriate handling practices when dealing with snakes. Furthermore, given the elevated risk of Salmonella transmission through direct or indirect exposure, snakes should not be considered suitable pets in households with children, elderly individuals and immunocompromised people. Following snakes, lizards were found to be the second most frequently colonized by both Salmonella spp. and S. enterica among all reptile taxa. Lizards are popular pet reptiles and our research revealed that they constitute the group most susceptible to increased Salmonella spp. colonization and excretion associated with captivity, with captive lizards exhibiting a prevalence rate over three times higher than their wild counterparts. Furthermore, notable differences in Salmonella spp. prevalence were observed across distinct lizard families. Of particular concern, the Agamidae and Iguanidae families, which encompass two of the most popular pet reptiles globally, the bearded dragon (Pogona vitticeps) and the green iguana (Iguana iguana), respectively [124], were found to have some of the highest Salmonella spp. prevalence rates among lizards. This is troubling, as pet lizards generally have more direct contact with their owners compared to snakes, especially with young children, who face elevated risks of developing invasive salmonellosis [125]. Indeed, bearded dragons and iguanas were the most frequently identified squamate sources of reptile-associated salmonellosis cases from 1997 to 2017, primarily involving children under five years of age [126]. Despite turtles exhibiting significantly lower Salmonella spp. prevalence rates compared to snakes and lizards, they are the primary reptile associated with human salmonellosis [101]. This can be attributed to the finding that 87.3% of Salmonella colonization cases in turtles are caused by S. enterica, with serovars such as S. Enteritidis, S. Typhimurium, and S. Newport ranking among the top ten serovars colonizing turtles. Furthermore, the public often incorrectly perceives turtles as safe pets, especially for children, which exacerbates the risk of human salmonellosis linked to these reptiles [127]. Finally, although crocodiles showed the lowest prevalence of both Salmonella spp. and S. enterica among the reptile taxa examined, the commercial farming of crocodiles for leather and meat production poses an additional risk of foodborne Salmonella transmission to humans. Indeed, the majority (89.0%) of the 345 captive crocodiles included in this systematic review were commercially raised for these purposes.

In conclusion, our study enables the characterization of the substantial risk posed by reptiles as a source of human salmonellosis and identifies several factors that influence the prevalence of Salmonella spp. in reptiles, including captivity status, higher reptile taxon, and family. Our results confirm that reptiles represent a serious zoonotic hazard for Salmonella transmission. Despite the well-established role of pet reptiles as reservoirs for Salmonella spp., and the recommendation from the U.S. Centers for Disease Control and Prevention that children under five should avoid contact with these animals [128], several studies have reported that reptile owners commonly lack awareness that Salmonella can be acquired from their pets [32, 126, 129]. This knowledge deficit is associated with poor hygiene practices in reptile husbandry, which consequently increase the risk of reptile-associated salmonellosis [19]. As reptiles are becoming increasingly popular as companion animals, there is a continued need to educate both reptile owners and the pet industry about the risks of reptile-associated salmonellosis, particularly for young children, the elderly, and immunocompromised individuals. Public health authorities should maintain their collaborative efforts with the pet industry, human and animal health agencies, and healthcare providers to ensure that adequate information and guidelines are widely accessible and disseminated.