The research concerns the genus Proteus, a group of neotenic salamanders that retain larval features, including external gills, throughout their lives. These obligate groundwater vertebrates inhabit limestone and dolomitic formations in the Dinaric Karst and are highly sensitive to contamination, hydrological alteration and disruption of aquifer connectivity. The International Union for Conservation of Nature classifies the genus as Vulnerable, and the European Union Habitats Directive places the species complex under special and strict protection. Reliable individual identification is important for studies of survival, population size, recruitment, movement and reproduction, all of which may require researchers to recognize the same animals during repeated surveys. Conventional marking can be especially problematic for a small, long-lived and protected population adapted to permanent darkness.

Toe-clipping is particularly unsuitable as a long-term identifier for olms because their high capacity for digit regeneration can alter the original mark and lead to mistaken identifications. Passive integrated transponder tags, which are read electronically, have uncertain retention in olms and can perform differently among species. Visible Implant Elastomer, or VIE, tagging is currently the most commonly used artificial marking technique in olm research. It involves injecting a silicone-based material beneath the skin so that it cures in place as a readable code, but this requires capture, handling and injection. A comparison cited in the study included 72 olms from another project: 61 were successfully VIE-tagged, producing an identification efficiency of 84.7%, while four were too small to tag and seven were not tagged. Those figures were not generated as a direct experimental comparison with the UV-A method.

The new approach instead exploits fluorescence naturally present in olm skin. Under 365-nanometre illumination, the animals displayed distinctive epidermal patterns that were especially clear on the head and dorsal body. Illumination at 395 nanometres exposed additional dermal features, including less abundant greenish spots. The researchers assessed the presence, position and configuration of individual markings rather than depending on one isolated spot. Across the sampled animals, the non-uniform distribution of fluorescent compounds produced body signatures that appeared to be unique to each specimen. Infrared illumination at 850 nanometres was the least effective of the tested wavelengths and did not improve the detection of body patterns or identification success. Visible light remained useful for recording amputated appendages, hematomas and other morphological abnormalities, while the UV-A images supplied the discriminatory patterns used for individual recognition.

To reduce the chance of observer bias, each new animal was compared with the complete collection of photographs from previously identified olms, rather than with a small subset. At least two observers conducted each comparison independently and blindly. One had participated in fieldwork, while the other had not and therefore was unaware of the animals’ tail-clipping information. A capture was classified as a recapture only when both observers considered the photographic characteristics consistent with those of a previously identified animal. Field samples carried progressive numbers rather than assigned identities, and individual IDs were added only after analysis. Tail clips taken for genetic work were used retrospectively as a secondary check, reducing the risk that an animal would be rejected because an observer had overlooked part of its fluorescent pattern. They were not the primary basis for photo-identification, and the monitoring program subsequently used visible and UV-A photography for recognition.

Repeated encounters indicated that the natural markers remained stable during the study. The longest interval between a first capture and a recapture was approximately 12 months. An olm photographed on 17 August 2025 and encountered again on 25 February 2026 retained recognizable clusters of spots on its head and trunk. A juvenile that weighed 3.20 grams and had a total length of 12.00 centimetres at its first capture weighed 3.60 grams and measured 12.12 centimetres more than a year later, while its identifying pattern remained recognizable under UV-A light. Across the first year of monitoring, no recaptured olm, whether juvenile or adult, showed a detectable change, reduction or rearrangement in its fluorescent markings under 365-nanometre illumination. The authors caution that this is a limited observation period relative to the potential lifespan of the animals, and studies of longer-term persistence are continuing.

Photo-identification also worked at sites containing many similar-looking animals. At Pozzo presso Jamiano, 64% of documented individuals were recaptured, while 53% were recaptured at Pozzo della Ferrovia. Some animals at both caves were encountered on three or four separate occasions. The campaign was described as the largest sampling of olm populations conducted within their natural distributional range to date, exceeding earlier studies that sampled 13 olms, monitored 19 adults or collected 76 animals. Approximately 200 individuals from a single cave were distinguished through manual comparison of their photographs. Because many documented Proteus populations contain only a few dozen observed animals per site, the findings suggest that manual “photo passport” matching is feasible for routine monitoring, although the study did not test how well the procedure would scale to thousands of animals.

The equipment requirements are modest. Researchers needed a mobile phone or camera, a headlamp, UV-A torches operating at 365 and 395 nanometres and an infrared torch. A 12-megapixel Samsung Galaxy S21 was among the simplest photographic systems used, while the most professional setup was an OM-5 Mark II camera with a 20.4-megapixel M.Zuiko 12–45 mm f/4 Pro lens. The investigators observed no differences in identification performance between smartphone and professional-camera images or between RAW and JPG files. At least three and as many as 10 images were obtained under each light condition, and exposure to UV-A was kept below 30 seconds per animal. Identification by comparison with the photographic database took 15–45 minutes, depending on the number of specimens collected from a cave.

The estimated costs were also lower than those of VIE tagging. An initial UV flashlight was valued at €15–150, or US$17–169, and a camera or smartphone may already be available, whereas a VIE injection kit was estimated at €245–1,138, or US$280–1,300. UV photography uses no consumable marking material and has negligible per-animal costs after the initial equipment purchase. It could also be applied to the smallest olm identified in the study, which weighed 2.6 grams and measured 11 centimetres in total length. Although routine collection still involved placing animals in a box for documentation, successful identification was possible while an olm remained submerged, reducing the need to remove it from water. The method is therefore described as virtually non-invasive when photographs are taken underwater and minimally invasive when temporary containment is necessary.

Animal welfare and the biological basis of the markings remain important limitations. No behavioral flight response was observed during the study, and each olm received less than 30 seconds of UV-A exposure at a distance of at least 35–40 centimetres. However, no published study cited by the researchers had specifically measured the physiological or welfare effects of short-term UV-A exposure in Proteus, so illumination should remain limited to the minimum needed for documentation. The fluorescence may be associated with riboflavin-like compounds naturally present in the skin, but the precise molecules and the mechanisms producing their spatial distribution remain unknown because skin sampling was outside the study’s scope. The evidence therefore supports reliable individual recognition over approximately one year, not permanent stability or the absence of possible effects. The fluorescent patterns nevertheless provide a high-discrimination, low-cost and readily deployable tool for long-term monitoring of vulnerable Proteus populations, and may also be applicable to other rare or cave-dwelling animals with individually distinctive ultraviolet patterns or pigmentation.

Subject of Research: Individual recognition of wild olms using natural UV-A fluorescent skin patterns.

Article Title: Non-invasive UV-A photo-identification for individual recognition in olms (genus Proteus )

Article References: Restaino, M., & Manfrin, C. (2026). Non-invasive UV-A photo-identification for individual recognition in olms (genus Proteus). Herpetozoa, 39, 429-440. https://doi.org/10.3897/herpetozoa.39.e208806

Image Credits: AI Generated

DOI: 10.3897/herpetozoa.39.e208806

Keywords: olms, Proteus, UV-A fluorescence, photo-identification, cave salamanders, Classical Karst, wildlife monitoring, animal welfare, capture-recapture, Dinaric Karst, Non-invasive, UV-A