Thermal-UAV assisted detection and relocation of roe deer fawns in managed grasslands

Thermal-UAV assisted detection and relocation of roe deer fawns in managed grasslands | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 10 February 2026 V1 Latest version Share on Thermal-UAV assisted detection and relocation of roe deer fawns in managed grasslands Authors : Petter Kjellander 0000-0002-4272-6737 [email protected] , Paula van der Heide , Giorgia Ausilio , Madeleine Christensson , Henrike Hensel , Linda Höglund , Anders Jarnemo , Arvid Norström , and Florent Rumiano Authors Info & Affiliations https://doi.org/10.22541/au.177075198.88223885/v1 285 views 128 downloads Contents Abstract Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Advancements in modern fodder production have unintentionally increased wildlife mortality during grass mowing operations. Roe deer (Capreolus capreolus) fawns are particularly vulnerable because they remain motionless as a predator-avoidance strategy. Unmanned aerial vehicles (UAVs) equipped with thermal sensors offer a promising method for detecting fawns before mowing, but the post-relocation survival of relocated fawns remains uncertain. In this study, 40 roe deer fawns were marked, mainly with VHF collars, several days or weeks prior to mowing. An independent unmanned aerial vehicle (UAV) pilot, unaware of fawn locations, conducted aerial surveys in the early morning before mowing. The pilot detected 22 of 24 radio-marked fawns present in fields scheduled for mowing, corresponding to a >90% detection rate. All detected fawns were relocated less than 300 m away to safe areas and survived for at least 4 days post-relocation, consistent with successful mother - offspring reunion. These short-distance relocations were thus associated with high short-term survival. Overall, UAV-based thermal sensors proved to be an effective approach to reduce fawn mortality in mowed grasslands. Thermal-UAV assisted detection and relocation of roe deer fawns in managed grasslands Introduction Technological advances in agricultural machinery have greatly improved the efficiency of fodder production, but they have also introduced new risks for wildlife inhabiting grasslands. Modern high-capacity mowers, with wider cutting widths and higher operating speeds, can cause substantial mortality among small or concealed animals during harvest operations (Wimmer et al. 2013). Several vulnerable species, include both ground-nesting birds (Frawley & Best, 1991) and young deer (roe deer, Capreolus capreolus and fallow deer, Dama dama ), because of their behavior and reliance on agricultural fields (Jarnemo, 2002; Kjellander et al., 2012). The species most often mentioned as particularly vulnerable are neonate roe deer, which rely on concealment rather than flight to avoid predators (Jarnemo 2002; Cukor et al. 2019). Roe deer are a typical “hider” species (Lent 1974). During their first weeks of life, fawns spend most of the day motionless in dense vegetation while the mother forages nearby, returning several times a day to nurse (Espmark 1969). This behavior effectively protects them from natural predators such as the red fox ( Vulpes vulpes ) (Aanes & Andersen 1996; Jarnemo & Liberg 2005) but renders them extremely vulnerable to mowing. When heavy machinery approaches, the fawns’ instinct is to remain still rather than flee, and even older fawns that attempt to run may fail to escape if they are within only a few meters of the cutting blades (Jarnemo 2002). In some agricultural regions, mowing can account for 25 – 44 % of annual fawn mortality (Jarnemo 2002). Because roe deer are income breeders (sensu Jönsson 1997, Andersen et al. 2000), relying on current food intake rather than stored reserves, they select nutrient-rich agricultural habitats for foraging (McLoughlin et al. 2007; Panzacchi et al. 2010). Consequently, the timing of parturition, May and June in most of Sweden (Jarnemo et al. 2004), coincides directly with the peak mowing season. Beyond the obvious animal-welfare implications, this conflict has ecological and economic dimensions. Roe deer are a key game species in Europe (Cederlund et al. 1998; Burbaitė & Csányi 2009) and an important prey for large carnivores such as lynx ( Lynx lynx ) and wolves ( Canis lupus ) (Andrén & Liberg 2015). Moreover, undetected carcasses incorporated into silage can cause outbreaks of Clostridium botulinum in livestock (Galey et al. 2000; Moeller & Puschner 2007), and contamination by a single carcass may result in the disposal of entire silage units, corresponding to losses of several tonnes of feed. Traditional mitigation approaches, such as delaying mowing, driving patterns that start from the field center, or placing deterrents like plastic bags or sticks, are often impractical or have achieved limited success (Jarnemo 2002; Green 1998). In contrast, the recent introduction of unmanned aerial vehicles (UAVs) equipped with radiometric thermal sensors operating in the long-wave infrared radiation (LWIR) spectrum (≈8–14 µm) has transformed wildlife rescue operations in Europe. These systems record LWIR naturally emitted by objects and convert it into temperature-calibrated (i.e., radiometric) pixel values. This capability enables the detection of subtle thermal contrasts such as those between the warm body of a concealed fawn and the cooler surrounding vegetation, facilitating rapid and reliable identification, particularly at dawn when thermal differentials are strongest (Steen et al. 2012; Cukor et al. 2019; Obermoller et al. 2021). Studies have reported detection rates from 67 % to 100 %, depending on vegetation density, ambient conditions, and pilot experience. However, detection is only one part of the conservation challenge. Once fawns are located, they are usually relocated to nearby refuge sites before mowing begins. Yet, the effectiveness of these relocations - defined both as the fawn’s survival and as the likelihood of successful reunification with their mother - remains poorly documented. Maternal separation can have fatal consequences, as fawns nurse up to seven times per day during their first month of life (Espmark 1969). The key challenge is to balance operational efficiency and animal welfare: relocations must be sufficiently distant to ensure safety from machinery, yet close enough to allow mother–offspring reunion. In the present study, conducted at the Stora Bjurum estate in south-western Sweden, we address this challenge by combining transmitter-marked roe deer fawns with independent UAV-based surveys during active mowing operations. Specifically, we aimed to (1) quantify the detection rate of UAV-based thermal imaging for marked fawns under real farming conditions, and (2) evaluate the short-term post-relocation survival and reunification success of fawns moved to safety. Materials and Methods Study area Intense fieldwork was conducted from 13 May to 24 June 2024 within the 2,100 ha Stora Bjurum estate, located in Västra Götaland County in south-western Sweden. The landscape is a mosaic of agricultural land and forest. The estate comprises approximately 483 ha of grazing land, 240 ha of grass fields, 140 ha of cereals, and 1,200 ha of hemiboreal forest. Mowing operations form a major component of land use at Stora Bjurum, with nineteen fields scheduled for harvest during the study period. Fields were typically cut using a 9-m cutting width operating at 15 – 20 km h⁻¹, corresponding to a working capacity of roughly 8 – 10 ha h⁻¹. Prior to the introduction of UAV-scanning in 2023, the landowner estimated that approximately 25 roe deer fawns were killed annually during mowing. Figure 1. The Stora Bjurum estate research area in Västra Götaland County, Sweden. All red and orange fields were eventually mowed while fields indicated with orange boundaries also contained radio-collared fawns, i.e., Exeprimental fields”. Numbers indicate field size (ha). Background imagery: © Google Earth, image © [2025] Maxar Technologies. Study setup Fawns were captured, marked, and equipped with transmitters ≥ 4 days to several weeks before the anticipated mowing dates to minimize any marking effects prior to disturbance. As mowing approached, farmers selected fields less than 24 hours in advance based on weather. The research team arrived at each field several hours before the machinery to conduct systematic UAV searches and VHF-based checks, allowing us to locate both marked and unmarked fawns. An independent UAV-pilot, inexperienced in UAV-based radiometric thermal sensing to detect wildlife and unaware of fawn locations, conducted the operational detection flights 1–2 hours before mowing began. Regardless of whether the independent pilot detected a given fawn, all individuals located by the research team were relocated to safe refuge sites prior to mowing. Unmarked fawns found during morning UAV flights were ear-tagged and, where collars were available, instrumented before relocation. UAV All UAV operations were conducted using DJI Mavic 3T platforms (DJI 2022), each equipped with a multisensor payload comprising a dual-RGB camera system and an LWIR radiometric thermal sensor. The RGB system includes a wide-angle camera with a 24 mm equivalent focal length and a 48 MP 1/2” CMOS sensor, providing high-resolution contextual imagery and supporting orthophoto generation, as well as a telephoto camera with a 162 mm equivalent focal length and a 12 MP sensor. The telephoto module offers up to 56× hybrid zoom, enabling precise long-distance visual confirmation of potential detections without altering flight altitude or disturbing wildlife. The thermal sensor operates in the 8–14 µm spectral spectrum and delivers radiometric temperature measurements at a resolution of 640 × 512 pixels, with an equivalent focal length of 40 mm. The thermal sensor acquires frames at 30 Hz, ensuring stable thermal anomaly detection during flights. All temperature estimates rely on a default emissivity value of 0.95, appropriate for natural vegetation and mammalian fur. Although radiometric accuracy can be affected by humidity, solar loading, and atmospheric transmissivity, conducting operations at dawn minimizes these influences and maximizes thermal contrast between endothermic animals and surrounding vegetation. To limit disturbance, flights were conducted at 50–55 m above ground level, yielding a ground sampling distance of approximately 5.8 cm per pixel. This spatial resolution is consistent with practical detection thresholds for small warm-bodied animals such as roe deer fawns, which typically produce thermal signatures spanning at least 2–3 pixels under favorable conditions. Flights were conducted in a semi-systematic manner where the pilot first flew over the field systematically to ensure that the whole field was covered. When thermal signals of interest were spotted, the pilot then deviated from the flight path to investigate signals manually. Thermal signals that were suspected to be fawns were marked with waypoints before the drone then returned and continued along the systematic flight path. Batteries were swapped at approx. 15% capacity. Operations were constrained by fog, rain, and strong wind. Capturing and monitoring roe deer fawns From 13 May, one to three observers surveyed farmland at dawn and dusk for lactating females likely to have fawns. Fields with suspected parturition were prioritized for UAV flights (04:00 – 07:00 hrs), when thermal contrast was greatest. On detection, waypoints were stored and a ground team approached. Each fawn was fitted with either a or a 151 MHz VHF collar (Followit AB, Lindesberg, Sweden) with a mortality sensor, plus ear tags. VHF collars (~60 g) were designed to gradually expand and eventually detach after 2 – 6 months of UV exposure. In total, 20 VHF, 5 GPS, and 4 older VHF collars were deployed, with older VHF units used only when other collars were unavailable. Tracking began immediately after collaring and continued for ≥ 4 days post-relocation and an additional last survival check approx. one month later. UAV surveys to find unmarked fawns were also used to verify locations of marked fawns, combined with radiotelemetry using handheld receivers and 4-element Yagi antennas. Triangulation (bearings from ≥2 locations) was used to estimate positions. Survival was checked daily; locations were determined at least every three days. Deceased fawns were recovered for field necropsy or submitted to the Swedish Veterinary Institute (Uppsala) for autopsy. Relocation protocol On mowing days, the independent, uninformed pilot flew 04:00 – 07:00 or slightly later (until reduced thermal contrast). Harvesting typically started 07:00 – 09:00 and spanned six days (June 11 - 16). Prior to the independent flights, the research team conducted preliminary checks (02:30–03:45) using both UAV and VHF-triangulation to confirm which marked fawns were in target fields as well as if new unmarked fawns were present in the field. The independent pilot generated waypoints for putative fawns and, after full coverage, returned to each waypoint and descended for the ground team to recover the fawn. Fawns were placed in a bag and carried to the nearest site providing cover (adjacent grass fields or forest edge), guided by recent doe movements. Older fawns that could not be captured were gently herded to safe habitat. Marked fawns were checked by VHF immediately before mowing to ensure they had not returned to the field. Any newly detected fawns were ear-tagged, collared when possible, and relocated. Post-relocation monitoring Post-relocation survival was monitored by daily VHF checks and opportunistic visual confirmations for ≥ 4 – 9 days. An extended follow-up was performed late July 2024 to document longer-term status (alive, collar drop, transmitter failure, or mortality). Statistical analysis To evaluate detection efficiency of thermal-UAV surveys, we modelled the probability that a radio-marked fawn known to be present in a field was detected by the independent operator/pilot (i.e., sensitivity). At each of 25 mowing events, we recorded the number of radio-marked fawns present in each field (n_marked) and the number detected by the independent operator (hits). We fitted generalized linear mixed models (GLMMs) with a logit link, including random intercepts for mowing event and field to account for repeated sampling and spatial structure. Because detection was near-perfect, we additionally fitted a beta-binomial model to account for potential overdispersion. Empirical sensitivity was calculated using Wilson 95% confidence intervals. Statistical analyses were conducted in R 4.5.1 using the packages glmmTMB, emmeans, and DescTools. Only fawns confirmed to be present via VHF telemetry prior to the independent UAV survey were included in detection analyses; fawns not located by VHF telemetry were excluded. Results Marking and fawn survival During the 2024 field season, 40 fawns were captured and marked within 12 of the 19 fields at Stora Bjurum estate. Thirty-six were equipped with transmitters (32 VHF, 4 GPS), while four were ear-tagged only. Of these, 30 faws were marked ≥4 days before mowing began (between 25 May and 7 June), and an additional six fawns were captured and marked on the day of mowing. Two GPS prototypes failed after a few days. Consequently, survival could be reliably monitored for 34 fawns with fully operational transmitters. Within 1 – 8 days after marking, ten fawns died (approx. 30%): seven due to fox predation (21%), two from undetermined causes (6%), and one during mowing (3%) in a non-experimental field. Fawn mortality during mowing One fawn was killed during mowing in 2024 within the broader study area (estate), compared to approx. 25 per year prior to 2023. The killed fawn had been captured and marked together with a sibling earlier the same morning in a field adjacent to the experimental field scheduled for mowing. The experimental field was surveyed in the morning and no fawns were detected. After marking, the fawn subsequently moved into the field scheduled for mowing, where it was killed. The sibling remained in the original, unmowed field and survived. Detection rate Drone flights spanned approx. 20 hours over six days, with approx. 48 ha mowed per day. Across 25 mowing events on 19 fields, 24 radio-marked fawns were known to be present at 10 events across nine fields. Of these, 22 (91.7%) were detected by the independent UAV operator directly on the remote controller. Detection was perfect (100%) at 24 of 25 events; at one event, two marked fawns were missed (Table S1). All detections were verified on the ground, and no false positives occurred. Empirical sensitivity was 0.92 (95% CI: 0.74 – 0.98; Table S1 and S2). Relocation distance In total, 24 fawns were relocated prior to mowing. For one fawn, coordinates were incomplete and excluded from distance analysis. For the remaining 23 fawns, we recorded 26 individual relocation events. Euclidean distances between points were calculated using the RT90 projected coordinate reference system (CRS) and ranged from 74 m to 293 m (mean 167 m). Most relocations were within 150 meters of the original site, typically to adjacent fields or forest edges. Four fawns were relocated twice. Three radio-collared fawns remained outside fields and did not require relocation. Post-relocation survival Of the 24 relocated fawns, 22 could be monitored for at least four days (range: 4 - 9 days) after the final handling and relocation. All monitored individuals remained alive during this period. Subsequent follow-up, 24 July (> 1 month after last handling) documented 12 fawns alive, seven collar drops, two transmitter failures, and one mortality of unknown cause. Multiple observations suggested successful reunion with does in several cases (proximity, coordinated movements, or nursing behavior). Discussion Our findings demonstrate that UAV-based radiometric thermal sensing can dramatically reduce mowing-related mortality in roe deer fawns under real agricultural conditions. The independent drone pilot detected more than 90 % of marked fawns known to be present, and only a single fatality occurred during mowing in the entire 2024 season, compared with roughly 25 per year in previous years at the same site. These results highlight that UAV-based radiometric thermal sensing, when properly timed and executed, was associated with a marked reduction in observed mowing-related mortality during fodder harvest. The exceptionally high detection rate observed at Stora Bjurum is consistent with recent European studies reporting success rates of 85 – 100% (Wimmer et al. 2013; Cukor et al. 2019). Several factors contributed to this performance. Flights were conducted between 04:00 and 07:00 h, when thermal contrast between vegetation and fawns was greatest, and flight altitudes of 50 – 55 m provided an effective balance between field coverage, ground sampling distance, and minimal disturbance to adult does. In the two cases where fawns were missed, vegetation was unusually tall or dense, indicating that local microhabitat structure can strongly influence detectability. Operator experience also played an important role: the same fawns were detected by a trained pilot at the same locations prior to a flight conducted by an uninformed and inexperienced pilot, supporting previous findings that human interpretation of radiometric thermal signatures from UAV-mounted LWIR sensors remains a limiting factor (Delisle et al. 2023). From a practical perspective, the usability of thermal sensors was strongly constrained after sunrise under the prevailing summer conditions. Once solar heating progressed, warm ground and vegetation features increasingly generated false thermal targets, making reliable detection difficult. Under cloudy conditions, the effective flight window could be extended by approximately one hour, highlighting the importance of both timing and weather conditions for successful UAV-based fawn detection. While the technology is reliable, practical constraints must be acknowledged. Battery capacity constrains UAV flight duration to approximately 25 minutes, and environmental conditions such as wind, rain, or fog can further reduce operational efficiency and sensor performance. Wind increases power demand and can affect flight stability, while moisture in the air may interfere with sensor optics and reduce thermal contrast, limiting detection reliability. Nevertheless, under favorable conditions a single UAV covered over 20 ha per flight, sufficient for real-time coordination with mowing schedules. The high-resolution RGB and radiometric thermal sensors enabled accurate detection of target animals without disturbing wildlife. These results demonstrate that UAV deployment is not only feasible but also scalable, offering a practical approach that integrates seamlessly with modern agricultural workflows while maintaining minimal ecological impact. Equally encouraging was the high short-term survival (100%) of fawns following relocation. All monitored individuals survived for at least 4 days (range: 4 – 9 days) after being moved, suggesting rapid reunion with their mothers and no evident increase in predation risk during this period. More than one month after mowing, 12 of the 13 fawns with functioning transmitters were still alive. However, it cannot be excluded that the single mortality (< 8%) observed in July was related to a separation from the mother following relocation disturbance, since the death cause could not be determined. The relatively short relocation distances, averaging 167 m, appeared sufficient to remove fawns from immediate risk posed by machinery while keeping mother–offspring pairs within their familiar spatial range. This is consistent with behavioural observations indicating that roe deer does typically remain within 150 – 200 m of hidden fawns (Espmark 1969). While longer-distance translocations may increase the risk of separation, short-distance relocations within the same field complex appear to be a low-risk mitigation measure. A potential concern is that relocated fawns may move back into the field scheduled for mowing, or into adjacent fields when fields are spatially aggregated, if mowing is delayed. In our study, continuous radio-tracking prior to mowing mitigated this risk in all but one case. On two additional occasions, fawns moved into neighbouring fields, but both individuals were detected visually and successfully herded out of the fields, either by the research team or following notification from the tractor driver. In both cases, the fawns appeared to be somewhat older and more mobile, suggesting that age and mobility may influence post-relocation movement and should be considered when selecting relocation sites. In practical implementations, temporary containment methods may further reduce the risk of re-entry. For example, Rehkitzrettung Schweiz (https://www.rehkitzrettung.ch/) employs shaded containment boxes that safely hold fawns until mowing is completed. Such protocols, when combined with real-time UAV guidance and close coordination with machinery operators, could contribute to a near-complete mitigation system for mowing-related fawn mortality. Beyond the immediate welfare benefits, widespread adoption of UAV-based fawn detection could have broader ecological and economic implications. Preventing mowing-related fawn mortality improves juvenile recruitment, supports sustainable hunting yields, and reduces the risk of silage contamination. The method also offers valuable research opportunities, as thermal UAVs enable rapid and minimally invasive detection of neonates for ecological monitoring and tagging. In addition, the approach is cost-efficient: the type of UAV required represents a relatively modest investment, the learning curve for basic piloting is short, and operational proficiency can be reached quickly. UAVs can be deployed on short notice and efficiently survey large areas within a limited time frame, making the method both practical and scalable. Future work should focus on increasing the automation and standardization of UAV-based wildlife detection. Machine-learning approaches, such as convolutional neural networks (CNNs) and object detection models (e.g. YOLO, Faster R-CNN), trained to recognize the characteristic shape, size and radiometric thermal signature of fawns, could reduce reliance on pilot expertise and improve detection consistency across operators and environmental conditions. The integration of multisensor data, combining radiometric thermal imagery with high-resolution RGB data, may further enhance detection performance by exploiting both thermal contrast and visual features. In addition, linking detection outputs to GIS-based planning tools and georeferenced flight paths could improve search efficiency by enabling real-time mapping of detected animals and adaptive adjustment of mowing operations to reduce wildlife disturbance. Finally, longer-term monitoring of relocated fawns, ideally using GPS transmitters providing fine-scale movement data, would allow assessment of post-relocation movement patterns, habitat use and subsequent survival. This study confirms that UAV-assisted thermal imaging represents a powerful, practical, and welfare-friendly tool for mitigating wildlife mortality in agricultural landscapes. By demonstrating both high detection reliability and successful short-term survival of relocated fawns, the Stora Bjurum project provides a model for integrating precision technology into conservation-minded farming. With modest training and logistical coordination, similar methods could be implemented widely across Europe to safeguard millions of neonate animals each spring. Conclusion UAV-based radiometric thermal sensing is a highly effective tool for detecting and protecting roe deer fawns during mowing. Detection exceeded 90%, and short-distance relocations (mean 167 m) resulted in no observed short-term mortality. These results support drone-assisted rescue as a scalable, humane method for wildlife protection in modern agriculture. Acknowledgements The landowner at Stora Bjurum (Anders Bergqvist), agricultural staff and a determined game keeper (Andreas Lindgren) and the UAV pilot at Stora Bjurum, assisting with housing, fieldwork and logistics. The Marie-Claire Cronstedt foundation for funding. References Aanes, R., Andersen, R. The effects of sex, time of birth, and habitat on the vulnerability of roe deer fawns to red fox predation. Canadian Journal of Zoology. 1996; 74: 1857-1865. Andersen, R., Gaillard, J.-M., Linnell, J. D. C., Duncan, P. Factors affecting maternal care in an income breeder, the European roe deer. Journal of Animal Ecology. 2000; 69: 672-682. Andrén, H., Liberg, O. Large impact of Eurasian lynx predation on roe deer population dynamics. PLoS ONE. 2015; 10(3): e0120570. Burbaitė, L., Csányi, S. Roe deer population and harvest changes in Europe. Estonian Journal of Ecology. 2009; 58(3): 169-180. Cederlund, G., Bergquist, J., Kjellander, P., Gill, R., Gaillard, J. M., Boisaubert, B., Ballon, P., Duncan, P. Managing roe deer and their impact on the environment: maximising the net benefits to society. In: Andersen, R., Duncan, P., Linnell, J. D. C. (Eds.), The European roe deer: the biology of success. 1998; Scandinavian University Press, Oslo, 337–372 p. Cukor, J., Bartoška, J., Rohla, J., Sova, J., Machálek, A. Use of aerial thermography to reduce mortality of roe deer fawns before harvest. PeerJ. 2019; 7: e6923. Delisle, Z. J., McGovern, P. G., Dillman, B. G., Swihart, R. K. Imperfect detection and wildlife density estimation using aerial surveys with infrared and visible sensors. Remote Sensing in Ecology and Conservation. 2023; 9(2): 222-234. DJI Enterprise. 2024; Mavic 3 Enterprise Specifications. Retrieved from https://enterprise.dji.com/mavic-3-enterprise/specs Espmark, Y. Mother–young relationships and development of behaviour in roe deer (Capreolus capreolus). Viltrevy, Stockholm. 1969; 6: 461–540 p. Frawley, B. J., Best, L. B. Effects of mowing on breeding bird abundance and species composition in alfalfa fields. Wildlife Society Bulletin. 1991; 19(2): 135–142. Galey, F., Terra, R., Walker, R., Adaska, J., Etchebarne, M., Puschner, B., Fisher, E., Whitlock, R., Rocke, T., Willoughby, D., et al. Type C botulism in dairy cattle from feed contaminated with a dead cat. Journal of Veterinary Diagnostic Investigation. 2000; 12: 204–209. Green, C. Reducing mortality of grassland wildlife during haying and wheat-harvesting operations. Oklahoma State University, Stillwater, USA. 1998; NREM-5006, 1–4 pp. Jarnemo, A. Roe deer Capreolus capreolus fawns and mowing – mortality rates and countermeasures. Wildlife Biology. 2002; 8: 211-218. Jarnemo, A., Liberg, O., Lockowandt, S., Olsson, A., Wahlström, K. Predation by red fox on European roe deer fawns in relation to age, sex, and birth rate. Canadian Journal of Zoology. 2004; 82: 416-422. Jarnemo, A., Liberg, O. Red fox removal and roe deer fawn survival – A 14-year study. The Journal of Wildlife Management. 2005; 69(3): 1090-1098. Jönsson, K. I. Capital and income breeding as alternative tactics of resource use in reproduction. Oikos. 1997; 78: 57-66. Kjellander, P., Svartholm, I., Bergvall U.A., Jarnemo, A. Habitat use, bed-site selection and mortality rate in neonate fallow deer ( Dama dama ). Wildlife Biology. 2012; 18:280-291. Lent, P. C. Mother–infant relationships in ungulates. In: Geist, V., Walther, F. (Eds.), The behavior of ungulates and its relation to management. IUCN, Morges, Switzerland. 1974 ; Vol.1, Publication Series 24, 14–55 p. McLoughlin, P. D., Gaillard, J. M., Boyce, M. S., Bonenfant, C., Messier, F., Duncan, P., Delorme, D., Van Moorter, B., Saïd, S., Klein, F. Lifetime reproductive success and composition of the home range in a large herbivore. Ecology. 2007; 88(12): 3192-3201. Moeller, R. B., Jr., Puschner, B. Botulism in cattle – A review. The Bovine Practitioner. 2007; 41(1): 54-59. Obermoller, T. R., Norton, A. S., Michel, E. S., Haroldson, B. S. Use of drones with thermal infrared to locate white-tailed deer neonates for capture. Wildlife Society Bulletin. 2021; 45(4): 682-689. Panzacchi, M., Herfindal, I., Linnell, J. D. C., Odden, J., Andersen, R. Trade-offs between maternal foraging and fawn predation risk in an income breeder. Behavioral Ecology and Sociobiology. 2010; 64(8): 1267-1278. Rehkitzrettung Schweiz. Infos über BFH-HAFL Methode. Rehkitzrettung Schweiz. Retrieved [Nov 12, 2025], from https://www.rehkitzrettung.ch/infos/infos-ueber-bfh-hafl-methode Steen, K. A., Villa-Henriksen, A., Therkildsen, O. R., Green, O. Automatic detection of animals in mowing operations using thermal cameras. Sensors. 2012; 12: 7587-7597. Wimmer, T., Israel, M., Haschberger, P., Weimann, A. Der Fliegende Wildretter in Aktion: DLR und BJV nutzen ferngesteuerte Flugplattform zur Rehkitzrettung. Hege und Bejagung des Rehwildes, Schriftenreihe des Landesjagdverbandes Bayern. 2013; 71-77 p. Supplements Table S1. Summary of detection results from 25 mowing events at Stora Bjurum, June 2024. Number of mowing events Total number of independent mowing occasions surveyed 25 Total marked fawns present Radio-marked fawns known to be present on fields 24 Total detected fawns Fawns detected by UAV 22 (verified; no false positives) Empirical sensitivity Proportion of marked fawns detected 0.92 (0.74–0.98) Wilson 95% CI Missed fawns Fawns present but not detected 2 (both missed at a single event) Mean modelled detection probability Estimated average sensitivity from beta-binomial GLMM ≈ 1.00 (no overdispersion) Between-event variance (SD) Random intercept variance (event_id) 2.16 (SD = 1.47) Moderate event-level variation Between-field variance (SD) Random intercept variance (field_id) 3970.64 (SD = 63.0) Unstable (few observations) Table S2. Event-level detection results from 25 mowing events with ten events that included ≥1 radio-marked fawn at Stora Bjurum, June 2024. 3 2 0 2 0.0 % [0.0, 65.8] % 4 2 2 0 100.0 % [34.2, 100.0] % 5 2 2 0 100.0 % [34.2, 100.0] % 10 2 2 0 100.0 % [34.2, 100.0] % 11 3 3 0 100.0 % [43.9, 100.0] % 12 4 4 0 100.0 % [51.0, 100.0] % 14 4 4 0 100.0 % [51.0, 100.0] % 21 1 1 0 100.0 % [20.7, 100.0] % 24 2 2 0 100.0 % [34.2, 100.0] % 25 2 2 0 100.0 % [34.2, 100.0] % Total 24 22 2 91.7 % [74.2, 97.7] % Information & Authors Information Version history V1 Version 1 10 February 2026 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords detection rate fawn mowing mortality uav Authors Affiliations Petter Kjellander 0000-0002-4272-6737 [email protected] Swedish University of Agricultural Sciences View all articles by this author Paula van der Heide Swedish University of Agricultural Sciences View all articles by this author Giorgia Ausilio Swedish University of Agricultural Sciences View all articles by this author Madeleine Christensson Swedish University of Agricultural Sciences View all articles by this author Henrike Hensel Swedish university of agricultural sciences View all articles by this author Linda Höglund Swedish university of agricultural sciences View all articles by this author Anders Jarnemo Halmstad University View all articles by this author Arvid Norström Swedish university of agricultural sciences View all articles by this author Florent Rumiano Swedish University of Agricultural Sciences View all articles by this author Metrics & Citations Metrics Article Usage 285 views 128 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Petter Kjellander, Paula van der Heide, Giorgia Ausilio, et al. Thermal-UAV assisted detection and relocation of roe deer fawns in managed grasslands. Authorea . 10 February 2026. DOI: https://doi.org/10.22541/au.177075198.88223885/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.177075198.88223885/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9fe119f1ca1e3fe2',t:'MTc3OTE3MjczMw=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();