Capture and care of moths
Bogong moths (Agrotis infusa) of both sexes were captured in the wild during their autumn migration (2024–2025, austral autumn, age around 9 months) and spring migration (2023–2025, austral spring, age around 5 months), using either a Lepi LED light ( and/or a vertical beam search light (model GT175, Ammon Luminaire Company), positioned in front of a white sheet suspended between two trees. All moths were captured near Thredbo (Dead Horse Gap, south-east NSW, Australia: 36.524° S, 148.260° E, elevation 1,580 m) and near the Mount Selwyn Snowfields (south-east NSW, Australia: 35.914° S, 148.444° E, elevation 1,600 m). Each captured moth was placed in its own small plastic container to isolate it from other moths. On capture, the moths were transported to the testing site at the property Glenhare, near Adaminaby, NSW (36.040° S, 148.864° E, elevation 1,250 m). Moths were fed a 20% honey water solution every other day and housed in a cool, sheltered environment with natural photoperiod conditions.
Attachment of tethering stalks on moths
To attach tethering stalks (thin non-magnetic tungsten rods), the scales on the moth’s dorsal thorax were carefully removed using a brush. The end of the tungsten stalk was bent using needle-nosed pliers to create a small circular footplate for connection to the thorax. The stalk was affixed to the moth’s dorsal thorax using contact cement (Impact Adhesive, EVO-STIK), and the moth was gently restrained by a weighted-down plastic mesh to ensure stability. Following stalk attachment (Extended Data Fig. 1f), each moth was transferred to a ventilated clear UV-transmissive Perspex container. Shortly before sunset (about 1 h before), containers were positioned on elevated rock formations to afford unobstructed views of the western sky, enabling observation of the setting sun and skylight polarization patterns—cues hypothesized to calibrate compass mechanisms51,52.
Controlled behavioural experiments in the laboratory
All of the controlled behavioural experiments were performed in an isolated dark room with extremely low background levels of RF disturbance in our purpose-built, highly ferromagnetic-free laboratory located at the rural property Glenhare, near Adaminaby3 (built on a concrete slab reinforced with fibreglass and constructed entirely from non-magnetic materials). Captured migratory Bogong moths were tethered and flown in our modified Mouritsen–Frost flight simulator3,13,36 in a double-wound 3D Helmholtz coil system to investigate whether moths respond to changes in an NMF (Extended Data Fig. 1a). All electrical grounds in the laboratory were connected in a single-point (star) configuration to minimize electrical noise, with the single point consisting of a metal rod. The experimental apparatus had its own dedicated earth separated from the mains earth (through a 6-mm-thick, 30-mm-wide and 12-m-long copper strap dug into the ground below the concrete slab). The arena consisted of a Perspex cylinder lined with a panoramic black cardboard silhouette of a simulated landscape (including trees, grass and hill peaks) set against a blue cardboard ‘sky’ backdrop (Extended Data Fig. 1d,e). A UV-transmissive Perspex lid (diameter, 50 cm; thickness, 0.5 cm), with a layer of UV-transmissive diffusing paper (251 Quarter White diffuser; Lee Filters), was placed on top of the arena and used for dorsal projection of celestial stimuli. A customized optical encoder53 was mounted at the centre of the lid. This encoder held a carbon fibre rod (length, 15 mm; diameter, 1 mm) that extended downwards inside the arena. The tungsten tether was connected to the end of this rod, thus connecting the moth to the encoder. This encoder allowed tethered moths to rotate freely (0–360°) in the horizontal plane and to continuously record their flight heading directions (five samples per second, angular resolution 0.9°) using customized recording software (Flash Flight Simulator Data Acquisition System, written by H.C.3,53). The recorded data therefore permitted us to reconstruct the moth’s virtual flight path for a specified period of time under different stimulation conditions. In addition, a very dim moving pattern of optic flow was projected from below, using a 45° mirror, onto a screen (251 Quarter White diffusing paper; Lee Filters) placed beneath the arena to simulate a moving view of the ground below the flying moth3,4,53. The direction of the optic flow was instantaneously coupled through feedback to the moth’s flight heading using custom software3,4,53 (H.C.) so that the optic flow always moved from head to tail below the moth (Extended Data Fig. 1g).
The three-axis Helmholtz coil system, custom built in aluminium and copper (University of Oldenburg workshop; outer coil diameters: x, 1,245 mm; y, 1,300 mm; z, 700 mm), manipulated the magnetic field around the experimental moth. The coil system was powered by three constant-current power supplies (Kepco, BOP 50-2M) and the current running through the coil system was controlled through high-speed USB carriers (USB-9162, National Instruments) and custom-written codes in MATLAB (v.R2019a; MathWorks). Before each experimental session, Meda FVM-400 magnetometer measurements ensured that the magnetic field was identical to the software settings (Extended Data Fig. 6).
Stimulation with starry skies
In the laboratory, a natural moonless starry austral night sky was projected onto the arena lid using a downward-pointing projector (LED projector ASUS S1 Mobile). The projector was mounted sufficiently high above the arena to provide clear and correctly sized dorsal images of the night sky. To avoid unwanted stray light, each projector was housed in a custom-built 3D-printed plastic box featuring an opening in front of the lens and ventilation slits above the projector. The free software Stellarium (v.24.1)54 was used to generate a celestial image simulating the moonless starry night sky over Canberra (around 80 km from Adaminaby in a straight line) at 21:30 on the night of 22 October 2022 (southern hemisphere spring 2022) (Extended Data Fig. 1b). In addition, to prevent Bogong moths from using their stellar compass3 for orientation during geomagnetic orientation experiments, a featureless starry sky of randomized stars3 (Extended Data Fig. 1c) was projected. These images were projected onto the circular screen created by the lid of the arena, with the projected sky’s dimensions matched to the screen size. Because the projectors did not emit UV light, we installed a custom-made LED-ring (built by T. McIntyre, University of South Australia; diameter, 120 mm; inner diameter, 50 mm) featuring eight UV LEDs (LED370E Ultra Bright Deep Violet LED; Thorlabs) centred around the front lens of the projector. The brightness of the LED-ring was reduced to quasi-natural levels by neutral-density filters and custom-written software in MATLAB (MathWorks).
To manipulate the intensity and spectral composition of light entering the flight arena (Extended Data Fig. 7), we used filter sheets (Lee Filters) placed in front of the projector lens. Neutral-density filters (210 ND0.6; neutral-density 0.6 log units, transmitting λ > 310 nm) were used to attenuate light intensity uniformly across all wavelengths, preserving spectral integrity. To lower illumination intensity to natural starlight levels, six layers of 210 0.6ND filter were placed in front of the projector lens. To block the UV–violet–blue part of this illumination spectrum, and to maintain approximately the same starlight illumination intensity, one layer of 768 Egg Yolk Yellow long-pass filter (λ > 500 nm; Lee Filters) and five layers of 210 0.6ND filter were placed in front of the projector lens. Upward illumination intensity and spectrum measurements of the projected starry night sky using these filter combinations were made from within the arena using an Ocean Optics QE65PRO Spectrometer (Ocean Optics) located underneath the projection screen at the position of the moth3 (Extended Data Fig. 7).
Generation and measurement of RF fields
RF fields were produced by a customized antenna and a noise generator (DG1022; Rigol) operating in noise mode. An output intensity of 2.5 Vpp (volts peak-to-peak) was selected for our experiments (Extended Data Fig. 5). The produced RF field intensities lay in the range of RF field intensities and broadband frequencies previously shown to disrupt magnetic compass orientation in night-migratory songbirds11,12 (Extended Data Table 3). A custom-constructed ring-shaped antenna with an inner diameter of approximately 51 cm was made from a coaxial cable (RG58C/U MIL-C-17 BNC connector), precisely designed to fit over the exterior wall of our flight simulator arena. In brief, a 180-cm-long coaxial cable with a BNC connector at one end was used. The braided shield at the opposite (cut) end was stripped to expose the central conductor. After determining the circumference to fit the arena (approximately 160 cm), the plastic sheath was removed over a 5-cm length of cable at a point 20 cm from the BNC connector end, exposing the braided shield (Extended Data Fig. 5a). The central conductor was then connected to this exposed shield to form a closed loop. The 2.5 Vpp noise-mode signal was selected and then transmitted through a 5-m-long coaxial cable (BNC connector) from the control room to the antenna (now fitted over the arena) to generate broadband noise in the range of roughly 150 kHz to 60 MHz (Extended Data Fig. 5c and Extended Data Table 3). The noise generator was used in ‘RF OFF’ mode and only switched to ‘RF ON’ mode for the experiments or during measurements. To calibrate the intensity of the RF noise, measurements of the generated RF fields were made at the University of Oldenburg (Germany) with a Signal Analyser (Rhode and Schwarz; FSV3004 Signal and Spectrum Analyser (10 Hz to 4 GHz)) in an RF-shielded experimental chamber29,30,55 using a calibrated active loop antenna (Schwarzbeck Mess-Elektronik; HFS 1546, 150 kHz to 400 MHz). This antenna was connected to the FSV3004 signal analyser through the wall of the chamber by means of a coaxial cable with N-connectors located on the outside of the chamber. RF fields were measured for the duration of one experimental test round (5 min). For comparison with other studies that explored the effects of broadband RF fields on the orientation behaviour of animals11,29,30,55, ‘average’ and ‘maxhold’ measurements were recorded. As described previously55 for maxhold measurements the detector was set to ‘Maxpeak’, whereas for average measurements the detector was set to ‘RMS’ (root-mean-square). The values of maxhold and average broadband RF density (\({\overline{b}}^{\max }\), \(\overline{b}\)) and intensity (\({B}_{{\rm{r}}{\rm{m}}{\rm{s}}}^{max}\), \({B}_{{\rm{r}}{\rm{m}}{\rm{s}}}\)) were calculated using equations 2.1 and 2.3 of ref. 55 for consistency and comparability (Extended Data Table 3). Notably, the ‘RF OFF’ mode (stand-by mode) provided similar measurements of density and intensity to measurements obtained with the generator set to the lowest possible output (0.004 Vpp) or turned off (Extended Data Fig. 5c and Extended Data Table 3).
Behavioural experiments under the sky in a natural landscape
Two non-magnetic Mouritsen–Frost flight simulators (the same type used in the laboratory) were placed on a hilltop at Glenhare and used to record the heading directions of tethered migratory Bogong moths experiencing the full local surrounding landscape in different light-filtered arenas (Extended Data Fig. 3c). Four flight simulator arenas were constructed from transparent UV-transmissive Perspex cylinders and lids (diameter, 50 cm; height, 35 cm), each fully enclosed (sides, lid and bottom) by distinct filter combinations: (1) a clear arena (one layer of Lee filter 130, λ > 310 nm); (2) a UV-blocking arena (two layers of Lee filter 226, λ > 380 nm); (3) a UV–blue-blocking arena (two layers of Lee filter 226 plus one layer of Lee filter 768, λ > 500 nm); and (4) a clear–neutral-density arena (one layer of Lee filter 130 plus one layer of Lee filter 211 0.9ND, λ > 310 nm). A tethered moth was connected to an optical encoder mounted at the centre of the lid (as described above). To prevent Bogong moths—known to use a stellar compass3—from accessing sky cues, a black cardboard disk 30 cm in diameter was positioned 3 cm above the tethered test moth, completely blocking its visual access to celestial cues. We ran two flight simulators in parallel on each night. The simulators were placed on each of two aluminium tables with clear UV-transmissive Perspex tops, and were placed 8 m apart. The two flight simulators were controlled from computers that were operated from inside a black light-tight cubical tent that was 10 m from the two arenas (Extended Data Fig. 3a). The tent was therefore a landscape feature (among others) that the moths could potentially see from different directions during their tethered flights. Thus, in the experiments described below, each cohort of moths was run in both arenas to control for any possible effects of differences in visual landmarks seen from each arena.
Behavioural procedures
Indoor behavioural procedures
For tethering in the panorama-walled arena, procedures were conducted under dim red light to maintain the moth’s dark-adapted visual state. The arena lid holding the optical encoder was lifted and the tungsten stalk of a vigorously flying moth (held with medical forceps) was attached to the bottom end of the encoder rod by means of a 1.5-cm length of thin rubber intravenous medical tubing that connected the stalk to the rod. The orientation of the moth was manually aligned to magnetic north and set as 0° in the software (0° = magnetic north). The optical encoder, controlled using custom software, recorded the moth’s flight heading under the projected night sky. Data were sampled at 5 Hz with a 0.9° horizontal angular resolution. Each moth was tested for exactly 5 min per stimulus condition and the resulting virtual flight trajectory was constructed from the sampled data.
Moths chosen for analysis had to meet three ante hoc criteria, two before the experiment and one during the experiment: (1) the tethering stalk had to be perfectly vertical; (2) wing flapping had to be vigorous with a large and equal amplitude for both wings, indicating that the contact cement had not interfered with the wings; and (3) the moth had to fly continuously for the full 5 min. If a moth stopped flying, the arena was gently tapped to stimulate continued flight. A moth that stopped flying four times was rejected, and the recording aborted.
Outdoor behavioural procedures
The same behavioural methodologies and inclusion criteria used in indoor experiments were applied outdoors. These field experiments aimed to characterize migratory behaviour in Bogong moths experiencing the real geomagnetic field and local landscape cues—specifically open grasslands, trees and rolling hills—from within light-filtered arenas. Experiments were performed under clear starry skies or partly overcast conditions during autumn 2025 (over ten nights during early and late April), as well as at two times of night, to test whether the migratory orientations of moths were affected by the spectrum of light.
Each night, we conducted experiments with roughly 40 Bogong moths, testing all moths in each of two arenas across two experimental phases. We used a within-subject counterbalanced design in which each moth underwent orientation measurements in both phases (in each of the two arenas) to control for individual variation (Extended Data Fig. 3c). The study comprised two main experiments (experiments 1 and 2) and an intensity control (experiment 3). In experiment 1, in which we blocked UV light, we simultaneously compared two cohorts of roughly 20 moths. In phase 1 of the experiment, one cohort was flown in the clear arena (one layer of Lee filter 130, λ > 310 nm), and one in the UV-blocked arena (two layers of Lee filter 226, λ > 380 nm). In phase 2 of the experiment, the cohorts were then swapped to the other arena and the experiment was repeated (Extended Data Fig. 3c). In experiment 2, in which we blocked UV, violet and blue light, we again simultaneously compared two cohorts of roughly 20 moths in phase 1 and swapped them to the other arena for phase 2 (Extended Data Fig. 3c). This time the two arenas were the clear arena and a UV–blue-blocked arena (two layers of filter 226 plus one layer of filter 768, λ > 500 nm; Lee Filters). Finally, in experiment 3, we controlled for the possibility that any observed disorientation in experiments 1 and 2 was not simply caused by a reduction in illumination intensity arising from the UV and blue blocking filters, rather than by the reduced spectrum itself. To do this, all 40 moths were flown in a clear–neutral-density arena (one layer of filter 130 plus one layer of filter 211 0.9ND, λ > 310 nm; Lee Filters), whose internal intensity was reduced to a level similar to the reduction caused by the UV and UV–blue spectral filters. In all experiments, the instantaneous orientations of moths were recorded in the same way as for indoor experiments. Between earlier and later experiments, tethered moths were kept isolated and in the dark in a suitcase that was warmed with hot water bottles.
Irradiances and filter transmission characteristics were again measured (Extended Data Fig. 3b) using a spectrophotometer (QE65PRO; Ocean Optics). The spectral transmission of each arena was measured at the normal location of the tethered moth under natural stellar illumination.
Data analysis
The data used for the analyses described below are available online56. Because there were no statistical differences in results obtained from male and female moths, the results were pooled.
The instantaneous heading directions (and the virtual trajectory) of each tethered flying moth were saved as a text file. Circular plot vectors, representing each moth’s mean orientation (α) and flight path directedness (r), were calculated via the classic Rayleigh test. Normality of r values was assessed using the Shapiro–Wilk test, with parametric t-tests or non-parametric Wilcoxon or Kruskal–Wallis tests used for group comparisons based on normality results. Flight paths were visualized in R (v.4.4.3, and mean orientation vectors were computed for each trajectory. To leverage both mean direction (α) and directedness (r) derived from flight trajectories, we applied a one-sided MMRT with Bonferroni correction for multiple comparisons using custom R scripts (developed by H.C., adapted from refs. 53,56,57,58). The MMRT-derived R* value quantifies population-level directedness and orientation, indicating whether the flight directions of a population of tested moths deviated significantly from random. The exact MMRT-derived P value was calculated using Moore’s formula4,58 (equation 3.3 in ref. 57). For comparing flight direction distributions across conditions (for example, 0° or 180° rotated magnetic fields, and different light-filtered conditions), we used the Watson–Wheeler test59 with Bonferroni correction in R to assess significant differences in circular distributions.
Statistics and reproducibility
Behavioural data obtained under open natural skies were obtained in independent experiments over successive nights (with new moths used each night). Behavioural data obtained in the laboratory under projected starry skies generated using Stellarium were obtained in independent experiments run over several nights with each night’s cohort of moths used only once. Thus, all these experiments deal with biological replicates. Sample sizes were based on the availability of moths, with at least 50 moths (and as many as 130) being used for each experiment. Experiments were conducted without blinding and without a specific randomization protocol (apart from the random selection of moths used as replicates). Because the experimental procedure and analysis were controlled entirely by computer (with no possibility for human intervention or bias), the conditions were considered effectively blinded. However, to rule out the possibility of any human bias, one of us not involved in the experiments (B.L.) wrote a new analysis code and re-analysed a blinded data set of 975 flight trajectories, each with 1,500 flight headings (recorded five times per second over 5 min), from 16 experimental conditions (Extended Data Fig. 2 and Supplementary Methods). Before the re-analysis, H.C. (who performed the experiments and the original analysis), assigned all 975 trajectory files with randomized identifiers (for example, Trajectory_1, Trajectory_2, …, Trajectory_975), and the identities of the 16 experimental conditions were replaced with arbitrary condition codes (A, B, C, …, P). The key linking the arbitrary condition codes (A to P) to the 16 original experimental conditions was retained by H.C. and was not disclosed to B.L. Moreover, B.L. conducted the analysis without access to the original condition identities, results or analysis code. He independently wrote a new R script to calculate the mean direction of each trajectory, generate the circular plots and calculate the group-level mean directions (α), R* values, CI95 and P values. All plots and statistical outputs were finalized before the experimental-condition codes were revealed. After completion of the blinded re-analysis, the condition codes were unblinded and matched to the corresponding panels in Extended Data Fig. 2. The plots generated from this re-analysis, once unblinded, matched those from Extended Data Fig. 2, in terms of individual flight vectors, the population mean vectors and the statistics, proving that the procedure adopted was effectively blinded (Supplementary Methods).
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.










Leave a Reply