This study provides insights into the temperature-induced GI microbiome dynamics of semi-aquatic turtles, a group of largely understudied Testudines. All turtles in the experimental group experienced a significant change in ambient-induced body temperature at each time point. Three animals from the control group and two from the experimental group were removed from the data set because they experienced prolonged higher or lower temperatures than expected due to observed behavioral thermoregulation. Despite differentially enriching some individual lineages (Fig. 5), the magnitude of temperature change and/or the total exposure time failed to elicit significant diversity or community composition changes in the experimental relative to control groups (Fig. 3, 4, additional file 1: S1). Our results contrast with other studies investigating effects of warming climates on the vertebrate and invertebrate gastrointestinal microbiome where significant decreases in both diversity and community structure have been reported [26, 27] .
T.c.triunguis are commonly found in the midwestern and southern United States (e.g., Alabama, Arkansas, Georgia, Illinois, Kansas, Louisiana, Mississippi, Missouri, Oklahoma, Texas) [7] which experience a wide range of temperature differences from periods of extreme cold to extreme heat. Free ranging box-turtles have been observed to be active at temperature ranges from 11.0°C to 36.0°C [28]. Unlike most reptiles, ectotherms native to these areas must be able to rapidly adapt to temperature swings. These adaptations may include rapid changes to their microbiome or alternatively, as reported here, a microbial community that exhibits minimal change despite environmental temperature shifts. The lack of significant community-level change between the two groups in our experiment could be due to the fact that the microbiome of T.c.triunguis has a level of innate robustness against rapid temperature swings
Recent work with active-season ground squirrels showed that the microbiome did not change at different timepoints in the seasons, which the authors identified as a potential insensitivity to change of the obligate hibernator microbiome [29]. Similarly, metatranscriptomic work on arctic ground squirrels shows that active season cecal microbiome composition is insensitive to seasonal time of collection and high versus low fat diet, however, both variables indeed affect microbiome transcriptional patterns [30]. These hibernating mammal studies, despite not being directly comparable to our ectotherm study model, suggests that animals living in habitats with extreme seasonality may be adapted to resist change in their gastrointestinal microbiome during the active season. The compositional robustness of the microbiome may allow for time-optimal energy harvesting needs. Further, the unchanging diversity of the microbiome throughout the active season also suggests a highly diverse GI microbial functional repertoire that through transcriptomic responses, as reported elsewhere [30], may optimally address environmental/dietary seasonality. Future work involving the effect of temperature on the metatranscriptomic dynamics of the T.c.triunguis GI microbiome is warranted. Although there were no significant changes induced by temperature in the T.c.triunguis microbiome at the community level, we detected significant experiment group-driven enrichment of a few dozen individual lineages or ASVs (Fig. 5). Despite these few taxa being individually significantly enriched as a function of experiment group, their net contribution to community-wide alpha diversity and ordination clustering is negligible.
At week 2 (24°C ambient temperature), there was a total of 16 and 21 differentially enriched lineages in the experimental and control group, respectively (Fig. 5A). No obvious enrichment trends were observed in terms of the number or taxonomic affiliation of these lineages. Given that there was no difference in temperature between the control and experimental group at this time point, we consider this observation to reflect minor methodological or biological variance in our data.
At week 4 (28.5°C ambient temperature) the experimental group had experienced an increase of 3.1° C mean body temperature for two weeks relative to controls. Here a clear enrichment trend emerged: 75 lineages were significantly enriched in the controls relative to only 5 enriched in the experimental group (Fig. 5B). This trend indicates that most differentially enriched taxa were disproportionally depleted by an increase in temperature and therefore enriched in the control group. This observation, despite being inconsequential to community-level diversity metrics, supports the notion of a net decrease in abundance of 75 microbiome members in response to a sustained average temperature increase of only 3.1°C.
At week 6 (33°C ambient temperature) the experimental group experienced an additional step increase in temperature of 3.7°C for two weeks (i.e.: a net increase of 6.8°C relative to controls over the previous two weeks). Here, a similar pattern as observed in week 4 was evident: more lineages (78 total) were depleted from the experimental group (enriched in the controls) while only 19 lineages were enriched by the sustained temperature increase (Fig. 5C).
Within the thermal and temporal range of our experiment (two week-spaced stepwise increases in temperature by 3.4°C), most susceptible lineages were disproportionally depleted in abundance by rising temperatures. Perhaps longer experimental time windows (> 4 weeks) or more drastic temperature increases (> 6.8°C) could have resulted in more dramatic differential enrichment trends with a potential impact on net community metrics.
Interestingly, one of the 19 temperature-enriched lineages observed at week 6 is a member of the Erysipelothrix spp. (Fig. 6). Members of this genus are ubiquitous Gram-positive bacteria that can infect a wide variety of hosts such as mammals, reptiles, fish, birds, and even insects [31]. The genus is comprised of several species, the most notable is Erysipelothrix rhusiopathiae, causing significant clinical disease in livestock and has zoonotic potential. Classically E. rhusiopathiae is the causative agent of swine erysipelas resulting in significant losses in outbreaks due to acute septicemia, endocarditis, cutaneous lesions, and chronic arthritis [32]. It is also a pathogen of zoonotic concern causing erythematous cutaneous lesions (erysipeloid) and possible fatal endocarditis if left untreated [33, 34]. Zoonotic infection with E. rhusiopathiae is most commonly as a result of handling infected animals [35]. This pathogen notably has also caused clinical disease and zoonotic transmission in American alligators (Alligator mississippiensis) and American crocodiles (Crocodylus acutus), [36] and has been isolated from a common snapping turtle (Chelydra serpentina) [37]. Phylogenetic analysis (Fig. 6) of the sequences recovered here placed the week 6 experimentally enriched Erysipelothrix spp. sequence within a clade that is most similar to previously sequenced E. rhusiopathiae. Although considered environmentally ubiquitous, the ASV sequenced in this experiment was preferentially enriched in the experimental group at the highest temperature. Enrichment of this ASV at this temperature may be expected since Erysipelothrix spp. is optimally incubated at 37°C [34, 38]. This supports the likelihood that this enriched ASV would appear in the experimental group at the highest temperature step.