Water samples were taken in the Sellrain Valley in the Stubai Alps along the watercourse of a mountain river (Zirmbach), a connected fishpond, and an adjacent river (Melach). Zirmbach is a stream in Sellrain with a length of 10 km and originates at an altitude of 1830 m through the confluence of Stockacher-Bach and Klammbach. Its drainage area is around 70 km2 – around the same size as the Melach after the Zirmbach disembogues into it. Lakewater samples have been obtained from Lake Gossenkoelle (GKS) and Lake Kaiser (KAI). Both lakes are located at the Kuehtai skiing resort, with GKS being of natural origin (2.413 m a.s.l.) and KAI a man-made reservoir (2,300 m a.s.l.) to produce technical snow and was built in 2016. Snow samples used in this study have been collected next to KAI. These samples have been collected along an altitudinal gradient (Figure 1).

Ice samples were taken from englacial ice from the “Nature Ice Palace” at the Hintertux Glacier (Figure 1). This touristic walkable crevasse is located at 3250 m a.s.l. and has a constant temperature of 0°C. Ice has hereby been taken from different spots all over the cave, including ice from superficial layers with clear signs of anthropogenic influence (e.g., visible handprints).

With the variety of different sampling sites, we took different levels of anthropogenic influence into account with Hintertux Glacier considered the most affected. Even though it is fairly difficult to evaluate the level of human impact on an absolute level, a relative differentiation of the water samples can be done. Therefore, the water sampling sites can be seen as a transect from most remote lakes (GKS and KAI) to intermediate (Zirmbach) and least remote (Melach). This categorization is mainly based on the geographic proximity to urbanized areas, which is the highest at the River Melach and the lowest at the mountain lakes situated at 2300 m and 2,416 m above sea level, respectively.

The “Nature Ice Palace” located at the Hintertux Glacier is of special interest regarding anthropogenic influence by being a tourist attraction. While again no absolute measure can be determined to define the human impact at this sampling site, numerous daily visitors (50.000 visitors/year, pers. comm. Roman Erler), frequent skin contact, and the lack of good ventilation due to the cave being well-isolated are making these superficial samples presumably one of the most affected by humans within this study.

Water was collected into either 250 mL sterile screw-cap bottles or Whirl-Packs®, respectively. Snow was sampled with a cleaned stainless-steel shovel and placed in Whirl-Packs®. For superficial ice, a stainless-steel shovel was used to scratch off ice chips. Cleaning of all sampling devices was done with the respective original sample. Sampling took place between December 2018 and February 2019. After transport to the home institute, snow, and ice were stored at −20°C until further use.

To proceed with the antibiotic screening, pure cultures had to be isolated from all sites. Therefore, 1 mL of each snow and water sample was plated on Reasoner’s 2A agar (R2A, Merck) and incubated at 4°C and 20°C, respectively, until visible growth. For incubation, a temperature of 4°C was chosen due to the original cryospheric habitat, while 20°C was applied in parallel to approach the requirements for standardized tests for antibiotic screening despite knowing this would be a compromise toward the tolerability of cold-loving bacteria. Ice and snow samples were processed the same way, except for an initial melting step at 4°C before plating. R2A agar was chosen due to the suitability of cryospheric organisms to grow best on oligotrophic media (Rüthi et al., 2023). Morphological different colonies from all samples were described, transferred into 7 mL of lysogeny broth (LB-media) with a sterile loop, and incubated for 7 days at the respective temperature. LB suits the criteria for the subsequent EUCAST test. After checking their purity with streaking, all isolates were tested for growth at 4, 20, and 37°C. The latter is the standard temperature for antibiotic screening tests, however, not suitable for cryospheric organisms which needed to be confirmed.

As vancomycin’s effect is limited on gram-positive bacteria, a gram test was performed beforehand to the actual antibiotic susceptibility test. To determine the cell wall structure, we followed the protocol of the Ryu non staining KOH technique (Ryu, 1938). Only gram-positive isolates were tested for vancomycin resistanceLife Science Identifiers.

DNA of all pure cultures was extracted using Qiagen Blood and Tissue DNA Extraction Kit following the recommended procedure of the manufacturer. For the PCR, the universal primers 27F (5’-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5’-GGTTACCTTGTTACGACTT-3′) were used to amplify the hypervariable regions from V1 to V9 of the 16S rRNA. The PCR reactions were performed in a total volume of 25 μL containing 12.5 μL of 2x PCR MasterMix (Roche), 0.5 μL of each Primer (10 μM), 1 μL of template DNA (50 ng/μL) and 10.5 μL of nuclease-free water. The cycling conditions were as follows: initial denaturation at 95°C for 2 min, followed by 30 cycles of denaturation at 95°C for 30 s, annealing at 60°C for 30 s, elongation at 72°C for 1 min, and final elongation at 72°C for 7 min. Gel electrophoresis was performed to check the correct length of the PCR fragment with a 1 kb + DNA ladder as a reference. Finally, the amplified products were purified using ExoSAP-IT reagent (Thermo Fisher). Sanger sequencing was then performed at the Fondazione Edmund Mach.

Bioinformatics was done in R (4.0.3) and RStudio (1.1.463.0) using the sangeranalyseR package (Chao et al., 2021). The threshold for the quality score was set at 30. Trimmed sequences were blasted against the NCBI database and submitted to the GenBank®.

All statistics were performed in R (4.0.3) and RStudio (1.1.463.0) including the following packages: tidyverse, ggpubr, rstatix, networkD3, dplyr, and dabestr. To evaluate differences in the diameter of the zone of inhibition, the dataset was filtered so that cultures that showed zero inhibition in their growth by the antibiotic at any condition (zone of inhibition = 0 mm) were not part of the analysis. If agar plates showed a zone of inhibition on at least one testing condition, the data was included in the statistical calculations. Estimation plots were performed to estimate the paired mean difference between the conditions MH15, MH17, R2A15, and R2A17. The influence of single parameters on the inhibition zone was tested with a Wilcoxon signed-rank test. Differences between the two cultivation-based antibiotic susceptibility tests were evaluated using descriptive statistics.

In total, we were able to isolate 35 bacterial colonies from all water samples, 21 from ice, and 6 from snow samples. Morphological differences were heeded within but not between sampling sites. Four cultures initially isolated on R2A agar did not grow after transferring them to liquid media, and one culture did not show any sign of growth after the first antibiotic susceptibility test, which is why in total five isolates were excluded from further analysis. Furthermore, no isolate was able to grow at 37°C while all remaining 57 cultures did successfully form colonies on agar plates at 20°C. Seventy percent thereof were able to grow at 4°C. This confirmed the prior assumption that 37°C is not suitable for the growth of cryospheric bacteria. Hence, all further tests were carried out at 20 and 4°C, respectively. All isolates formed colonies on R2A and MH agar. Of the 57 isolates used for further testing, 37 strains were identified as gram-positive bacteria with the KOH test.

The number of resistant isolates detected with the cultivation method varied between 63 and 83% for novobiocin and 67 to 85% for vancomycin. The biggest variations in the identification of resistant cultures were observed within the different incubation temperatures (Table 1) as in general, more resistances were observed on both nutrition media when the incubation temperature was decreased.

According to the cultivation method performed at 20°C, most resistances in the water samples were found in the river Melach (75% novobiocin; 80% vancomycin), followed by the fishpond (60%; 88%), Zirmbach (45%; 59%), KAI (17%; 67%) and GKS (17%; 33%). Samples from the ‘Nature Ice Palace’ located at the Hintertux Glacier contained the highest level of resistant isolates (100%; 81%) within all samples. While snow samples collected next to Lake Kaiser had moderate numbers of antibiotic-resistant bacteria (50%; 50%). On average, these numbers were higher in all habitats when the incubation temperature was decreased to 4°C.

On average, 61% of the isolates were resistant to novobiocin and 64% of the gram-positive bacteria were resistant against vancomycin when the agar disk diffusion method was performed at 20°C. These numbers increased to 75% for novobiocin and 78% for vancomycin when the incubation temperature was reduced to 4°C. However, results differed between different incubation conditions such as temperature, nutrition media, and agar concentration. When performed at 20°C, the agar disk diffusion method detected the majority of resistant bacteria within samples from the touristic ice cave at the Hintertux Glacier (98% novobiocin; 81% vancomycin). Among the other sampling sites, Melach contained the highest number of resistant isolates (50%; 50%) followed by the fishpond (50%; 50%), Zirmbach (32%; 64%), and Gossenkoellesee (33%; 33%). The least amount of resistance was observed within the isolates from Kaiser-See (25, 42%) while isolates from snow samples collected next to the Kaiser-See were resistant in 50 and 60% of the cases. The relative number of resistances increased with a decrease in the incubation temperature from 20°C to 4°C in all sampling sites except for the river Zirmbach.

The two cultivation-based antibiotic susceptibility tests used in this study delivered different results regarding the classification of the bacterial isolates (resistant or sensitive) in approximately 23.8% of the cases. In total, 147 antibiotic susceptibility tests were performed across different nutrition media and testing strategies, including 57 isolates for novobiocin susceptibility tests at 20°C and 40 at 4°C, as well as 31 and 19 gram-positive bacteria being tested for vancomycin resistance at each temperature, respectively. Among these tests, 36.1% of the test results differed when the antibiotic susceptibility tests were performed on R2A agar, and 32% of the test results differed when testing isolates for antibiotic resistance with both testing procedures on MH agar (see Figure 2).

Compared with the officially recommended MH agar (The European Committee on Antimicrobial Susceptibility Testing, 2021a) with an agar concentration of 17 g/L, all other nutrition media compositions used in this study resulted in increased inhibition zone sizes (Figure 3). The biggest mean difference could be observed between MH17 and R2A15 with 9.87 mm, while MH17 and R2A17 only resulted in a mean difference of 5 mm. On average, the zone of inhibition increased by 6.66 mm on MH15 compared to MH17. The difference in the inhibition zone size caused by the varying agar concentration was significant (p < 0.001). Furthermore, different nutrition media (p < 0.001) and agar concentrations show significance when using R2A (p < 0.001) and MH (p < 0.001).

A reduction of the agar concentration from 1.7 to 1.5% resulted in a significant increase of the inhibition zone by 5.17 mm for novobiocin (p < 0.001) and 4.36 mm for vancomycin (p < 0.01) on R2A agar. Similarly, on MH agar, the inhibition zone increased by 8.74 mm for novobiocin (p < 0.001) and 3.18 mm for vancomycin (p < 0.05). The largest absolute mean difference was observed between MH17 and R2A15 for novobiocin (+11.8 mm) and vancomycin (+6.6 mm). A significant effect of the incubation temperature could only be seen for novobiocin (p < 0.001) but not for vancomycin (p = 0.3).

Sequences trimmed and controlled for quality had an average length of ~675 bp. No chimeric sequences were found. A total of 15 cultures did not yield satisfactory BLAST results either due to low percentage identity (< 97%) or insufficiently long reads. The Gram status from the KOH test was confirmed in all but six cases. The Bacillaceae family was found to be the most prevalent (14) among the isolated colonies with Pseudomonadacea following close behind (8) (see Supplementary Table S4).

Out of the initial isolate yield (62), 57 bacterial colonies were successfully grown in liquid media (Luria-Broth) and tested for antibiotic susceptibility.

All cultures used in this study did grow at 20°C whereof 43 cultures had the ability to grow at 4°C, which highlights the psychrophilic character of the isolates usually growing best between 0 and 20°C (Morita, 1975). However, due to limitations in the cultivation of microorganisms from environmental and cold habitats, isolates in this may not be strictly limited to low growing temperatures.

Even though 56% of all bacterial colonies isolated during this study originated from water samples, none were able to grow at 37°C (the recommended temperature for common antibiotic susceptibility tests). This emphasizes the necessity of adaptations for the cultivation-based antibiotic susceptibility tests for bacteria from environmental and especially cryospheric habitats.

Antibiotic-resistant microorganisms were found in every sampling site of this study. However, the percentage of resistant bacteria increased along the altitudinal transect from the high alpine lakes (GKS and KAI) to the river Melach, located in a highly populated area. Even though it is difficult to actually measure human impact, an increase in the anthropogenic influence downstream of the water-sample transect can be assumed. On a relative scale, the river Melach, which is showing higher proximity to urbanized areas than any other water sample in this study, is likely to face the highest level of anthropogenic influence. On the other hand, GKS and KAI are only minimally impacted directly by humans, whereas the relative abundance of antibiotic-resistant isolates in these sampling sites is high compared to other little human-impacted areas (Scott et al., 2020). The rural Zirmbach is flowing through some sparsely populated areas and, the observed increase in antibiotic-resistant microbes downstream of the sampled water transect (Figure 4) goes hand in hand with an increased anthropogenic influence due to proximity to humans. These results agree with the general assumption that antibiotic resistance is to a large extent caused by anthropogenic pressure and thus especially present in environments shaped by humans (Kim and Aga, 2007; Berendonk et al., 2015). Furthermore, the same increase in antimicrobial resistance with increased proximity to areas with bigger anthropogenic influence has already been reported before in river networks (McConnell et al., 2018). The effect has been revealed by the selection and the character of the transect but was not the prior scope of this study. The fishpond along the course of the Zirmbach embodies a special case, as it is used for commercial aquaculture. Generally, antibiotics are still widely applied in the fish farming industry across the world and include antibiotics from different classes like penicillins or quinolones (Tuševljak et al., 2012). The use of antimicrobial substances would explain the high numbers of resistance found within the fishpond (Figure 4), even though it is situated outside urbanized areas. However, vancomycin and novobiocin are not approved for antibacterial therapies in aquacultures within the European Union (VO 37/2010 EU), which indicates that these resistances are likely not caused by the use of antibiotic therapies in treating potential fish pathogens.

The river Melach revealed the highest resistances along this transect (Figure 4) which might also be in context with populated areas. In order to cover a high variety of cryospheric habitats for our comparative study, we included an englacial cave open for touristic usage: “Nature Ice Palace” located at the Hintertux Glacier and which is not connected to the transect. The cave is exceptional due to heavy pressure by ca. 50.000 yearly visitors and the lack of ventilation. Large numbers of antibiotic-resistant bacteria were found originating from englacial ice. In comparison to other glacial environments around the world, the results indicate that antibiotic resistance does occur in very high numbers within the ice cave (Segawa et al., 2013) (see Figure 4).

The comparison of the two different antibiotic susceptibility tests clearly showed that they potentially provide different results regarding the resistance status of bacterial isolates. While it cannot be stated for sure which method is more accurate, the agar disk diffusion method has some clear methodical advantages. It has been tested for years and was already described as an applicable method to evaluate antibiotic resistance in the mid-20th century (Bauer et al., 1959). Apart from the disk-diffusion test being a standardized procedure with an official protocol to follow, it also considers the antibiotic concentration to match its minimum inhibition concentration for each bacterial species – an aspect the cultivation method lacks completely. This absence may be the main reason for the differences between the two methods. The more precise reading guide connected to breakpoints of the agar disk diffusion method – which identifies an isolate as either resistant, susceptible, or (in some cases) intermediate – poses another major cause for differences in the resistance profile of an isolate based on the testing procedure. In total, the two methods delivered different results in ~24% of the treatments (all incubation temperatures and nutrition media). This again highlights the importance of a standardized testing procedure for antibiotic susceptibility testing of environmental and especially cryospheric samples.

In general, an effect on the size of the zone of inhibition has been observed by nutrition media, agar concentration, and incubation temperature.

The effect of the nutrition media on the size of the zone of inhibition is mainly caused by the differences in the agar concentration. A decrease of the agar concentration from 1.7% (original recipe of MH) to 1.5% (original recipe of R2A) ultimately leads to an increased zone of inhibition size at both nutrition media tested. This is likely caused by the enhanced diffusion rate of the antibiotics at a higher diffusion coefficient. According to Fick’s (1855) law the diffusion rate is inversely proportional to the viscosity of the media the molecule has to pass through. The reduction of the agar concentration does decrease the viscosity of the nutrition media, which causes an increased diffusion coefficient and thus an increased diffusion rate for the antibiotics. When comparing R2A15 with R2A17 and MH15 with MH17, respectively, highlights the strong significant effect the agar concentration has on the size of the zone of inhibition (Figure 3).

While other factors like starch may also influence the diffusion rate of antimicrobial substances, these seem to have a much lower impact. However, the differences in the nutrition media compositions might explain the discrepancy between MH15 and R2A15 or MH17 and R2A17 (Figure 3). The significance of the differences indicates that the agar concentration itself is not the only factor that influences the diffusion rate of the antibiotics. Especially, the increased starch concentration of MH agar potentially impacts how easily molecules can diffuse through the agar. Nonetheless, the size of the inhibition zone differed the least when MH17 was compared with R2A17 for both novobiocin (+6.64 mm) and vancomycin (+2.24 mm). Overall, these results suggest that the use of different nutrition media and especially of a different agar concentration can lead to a change in inhibition zone size and, therefore, to a potential misinterpretation of the antibiotic susceptibility of an isolate. Using R2A agar (with 1.5% agar concentration) instead of the officially recommended MH agar (with 1.7% agar concentration) resulted in an average increase of the inhibition zone by 11.8 mm for novobiocin and 6.6 mm for vancomycin, respectively. These differences caused by different nutrition media and agar concentrations can have a drastic impact on the final result of the antibiotic susceptibility test. This is especially relevant because these increases equal roughly 74% (novobiocin) and 44% (vancomycin) of the thresholds given (according to the EUCAST breakpoint table, The European Committee on Antimicrobial Susceptibility Testing, 2021b) to identify a bacterium as either resistant or susceptible to the antibiotic and could therefore lead to a misinterpretation of the actual susceptibility.

The different increase in the zone of inhibition size between the two antibiotics can potentially be explained by the differences in their molecular weight (novobiocin 612.6 g/mol, vancomycin 1449.3 g/mol) as it is again inversely proportional to the diffusion coefficient. According to Fick’s law, diffusion decreases with an increasing radius of the diffusing molecule. Therefore, the diffusion rate is expectedly higher for novobiocin than for vancomycin, which is also represented in the results (Figure 3).

Previous studies described an effect of a decreased incubation temperature and the consequent prolonged incubation period on the results of the agar disk diffusion (e.g., Smith and Kronvall, 2015; Smith et al., 2018) which has also been observed in this study. However, these experiments are usually performed on mesophilic bacteria, while, to the best of our knowledge, nothing is yet reported for psychrophilic and psychrotolerant microorganisms. Our results show that a decrease in the incubation temperature from 20°C to 4°C can cause an increase in the zone of inhibition diameter size when the agar disk diffusion method is applied to psychrophilic and psychrotolerant bacteria. This result is contrary to the expected decreased diffusion rate of molecules at lower temperatures. However, this effect seems to be irrelevant due to the prolonged incubation period of up to 2 weeks for cultures to visibly appear. Interestingly, the significant difference in the size of the zone of inhibition at the two temperatures of interest could only be observed when the antibiotic susceptibility test was performed with novobiocin – a molecule that needs to enter the bacterial cell to inhibit its growth by inhibiting the DNA and/or RNA synthesis (Smith and Davis, 1967). In general, the transport of molecules into bacteria can be mediated by either passive diffusion, facilitated diffusion, or active transport through energy-dependent systems (Chopra, 1989). However, passive diffusion of hydrophilic antibiotics through the cytoplasmic membrane is limited to a size of a maximum of 100 g/mol (Franklin and Leive, 1973). Nevertheless, it is known that even though passive diffusion rates across membranes are lower at cold temperatures, bacterial cells up-regulate membrane transport proteins (de Maayer et al., 2014). This could impact the effective novobiocin uptake at lower temperature levels and explain the average increase in the inhibition zone size. However, even though the zone of inhibition increased significantly when isolates were tested for novobiocin susceptibility at 4°C, the relative number of bacteria determined as resistant did not change. No significant change in the size of the zone of inhibition was observed when the agar disk diffusion method was performed with vancomycin at 20°C and 4°C.