Mobbing is a widespread, vocally coordinated behaviour where species approach and harass a threat. The noisy miner (Manorina melanocephala) is a notorious native Australian honeyeater, well-known for its hyperaggressive mobbing. Numerous studies have identified negative impacts of their mobbing behaviour, highlighting the exclusion of competitors from colony areas and the resulting loss of woodland-bird biodiversity. Despite this, few studies have investigated mobbing itself, and our understanding of the factors which influence its expression remains limited. Here, we use a field-based playback experiment to investigate whether mobbing responses vary in relation to colony borders and caller familiarity. Noisy miners were more likely to respond, reacted more quickly, and responded more strongly to mobbing calls broadcast inside as opposed to outside the colony. These behavioural differences likely arise from variation in the relative costs and benefits of responding. When noisy miners did mob outside the colony, more individuals joined in response to unfamiliar as opposed to familiar callers. Our results reveal that noisy miner mobbing may not be as indiscriminate as often assumed, with caller familiarity and location influencing this behaviour. We suggest there are benefits to greater consideration of the factors impacting noisy miner mobbing behaviour.

Study site and species

Noisy miners are medium-sized (~ 70g) native Australian honeyeaters that live in persistent, year-round colonies across eastern and southern Australia (Higgins 2001). Consisting of up to several hundred birds, colonies occupy discrete areas of open eucalypt and urban habitat, preferring areas with a high edge-to-interior ratio (Taylor et al. 2008). Colonies are made up of highly-social cooperatively breeding subgroups, known as ‘coteries’, containing up to 20 individuals in over-lapping ‘activity spaces’ (Dow 1978; Dow & Whitmore 1990). Coterie members regularly interact with one another within these activity spaces, including foraging, roosting, and regular mobbing behaviour. Cross-coterie groups also form during larger mobbing events as birds leave these centres of activity to chase other noisy miners or mob heterospecifics (Arnold 2000). Noisy miners are highly aggressive to a range of species, actively expelling both avian predators and other small to medium-size birds (i.e. potential food competitors) from their colonies (Dow & Whitmore 1990, Clarke & Oldland 2007). Like much of noisy miner behaviour, mobbing is coordinated vocally; miners use two functionally referential alarm calls (Cunningham and Magrath 2017; Farrow et al. 2017; Holt et al. 2017), a high-pitched ‘aerial’ alarm call given to raptors in flight and a broad-frequency mobbing (‘chur’) call given to ground and perched predators, as well as competing heterospecifics (Kennedy et al. 2009; Magrath and Bennett 2012). Past work has established that chur calls provide information about individual identity (Kennedy et al. 2009), but whilst noisy miners can differentiate between callers in a captive setting (McDonald 2012), field-based playback experiments suggest that it may not always be beneficial to do so (Kennedy et al. 2009). Individuals can adjust their response to vocalisations in relation to the relative costs and benefits of doing so, but the potentially fatal cost of not responding to alarm calls may outweigh the costs of discriminating between callers, even when information about caller identity is vocally encoded (Schibler and Manser 2007). 

This study was conducted on a wild population of noisy miners at Newholme Field Research Station (30° 25’ S, 151° 38’ E), a rural property approximately 7 km north-west of Armidale managed by the University of New England, chosen as it supports a high density of noisy miners. This colony of ~100 adults inhabits a patch of remnant eucalypt woodland (~ 75 ha) in a heavily-cleared agricultural landscape, and has been studied since 2011 (Barati et al. 2016, 2018; Barati and McDonald 2017, Farrow et al. 2020). We elected to run playback experiments on a single colony, as whether there is geographic variation in the mobbing behaviour of the species is currently unknown. While only one colony was tested, the sampling decision as to mob or not is taken at the individual level in this species and the experimental protocol was such that multiple coteries were exposed to playbacks (akin to testing multiple neighbouring social groups within a non-colonial species). To increase the likelihood that different playback sets (see ‘Playback experiments’ below) were targeting different colony members, the six playback pairs within the colony took place at least 145 m apart (mean±SE: 304±34 m, range 145–595 m), a distance greater than the average activity space of individual noisy miners (mean diameter 129 m; Dow 1979). Moreover, different individuals from within the large colony respond to different mobbing events, so by moving the stimulus to different locations within and external to the colony, different birds were challenged in each playback. All research was authorised and conducted under NSW National Parks and Wildlife Service Scientific License (SL100314) and UNE Animal Ethics Committee (AEC20-099).

Call collection and preparation of playback stimuli

Playback stimuli were constructed with chur-call recordings (Fig 1) from noisy miner mobbing events collected during February and March 2023 at the Newholme colony and at a colony of noisy miners at Baker’s Creek, 25 km from Newholme. Recordings were made using a Marantz PMD661 professional solid-state recorder (sampling rate 44.1 kHz, 16-bit resolution; Marantz America, Mahwah, NJ) with a Sennheiser ME67 directional microphone (Sennheiser UK, High Wycombe, Buckinghamshire, UK) and Rycote softie windshield (Rycote Microphone Windshields, Stroud, Gloucestershire, UK). Mobbing events were provoked in calm conditions using a taxidermy model of a boobook (Ninox boobook) – a local owl species known to be aggressively mobbed by noisy miners – secured on a tree branch, with recording equipment mounted on a tripod 5–10 m away.

For use in the ‘familiar’ treatments, six mobbing events were recorded at different locations within the Newholme colony, while for use in the ‘unfamiliar’ treatments, six mobbing events were recorded at different locations within the Baker’s Creek colony. Baker’s Creek is far enough for the colonies to be considered discrete and callers to be unfamiliar; recent analysis found little evidence of marked gene flow between colonies over 8 km apart (Barati unpublished data 2023; Dow 1979).

Twenty-four (i.e. 12 familiar and 12 unfamiliar) 10-min playback tracks were constructed by extracting chur calls from original recordings and inserting them into ambient sound (recorded at the start/end of mobbing events) using Raven Pro (version 1.5, The Cornell Laboratory of Ornithology, NY, USA). To avoid pseudoreplication, two different tracks were constructed per mobbing event. Mobbing recordings contained chur calls from 8–12 individuals, with the number of individuals calling matched within each set of playback tracks (see below). To ensure consistent between playbacks in the amount of time a signal was present, all tracks were constructed as 30 s of chur calls followed by 30 s of silence for the 10 min duration. As Baker’s Creek is adjacent to a major road, a low-pass filter at 500 Hz was applied to all recordings from both sites to remove low-frequency disturbance. Tracks did not include any other vocalizations from noisy miners or other species.

Playback experiments

To assess whether the response to chur calls is influenced by call location in relation to colony borders and/or caller familiarity, a playback experiment was performed. Six playback sets were conducted, where each playback set comprised four treatments: familiar chur calls broadcast inside the colony (‘familiar inside’), unfamiliar chur calls broadcast inside the colony (‘unfamiliar inside’), familiar chur calls broadcast outside the colony (‘familiar outside’), and unfamiliar chur calls broadcast outside the colony (‘unfamiliar outside’). Colony borders were based on long-term observations of noisy miner occurrence and the distribution of recorded nests (Barati et al. 2016; Barati 2017). The order of the four treatments was counterbalanced between the six sets of trials. Unique tracks were used for each of the six playback sets, with the two familiar tracks in a given set constructed from recordings of the same Newholme mobbing event and the two unfamiliar tracks from the same Baker’s Creek mobbing event. To avoid habituation, models were placed in a different tree for each pair of inside/outside treatments within a given playback set. Paired trees were <100 m of each other (mean±SE: 52±11 m, range 20–96 m) – this distance is less than the mean activity space of individual noisy miners (Dow 1979), thus in each pair, the nearest miners were likely members of the same coterie (see Study species above).

Playbacks experiments were conducted on calm, dry days from 07:00 – 12:00 h in February and March 2023. Experiments took place outside of the breeding season to avoid any changes in mobbing behaviour that might be associated with nest protection and brood provisioning (Arnold et al. 2005). When weather allowed, a maximum of two playbacks were conducted per day, at least 3.5 hours apart. Playback locations were alternated such that if a day’s first playback took place inside the colony the second took place outside and vice versa. To ensure that consecutive playbacks were not targeting the same colony members (and so minimising risk of habituation to the stimuli), consecutive playbacks took place at least 455 m apart (mean±SE: 545±28 m, range 455–646 m), over 3 times the average activity space of individual noisy miners (Dow 1979).

Before each playback, one of four 3D-printed boobook models was placed in a tree within arm’s reach when no noisy miners were visible in the immediate vicinity and covered by a pillowcase (as in Mesken 2021). All four models were used in a given set and their order of presentation counterbalanced. A portable Bluetooth speaker (JBL Xtreme 2, Harman International, Connecticut, US) was placed in the tree below the model. A Digitech QM-1589 sound meter was used to determine playback amplitude prior to experiments and all calls were broadcast at natural levels (~80 dBC at 1 m; Holt et al. 2017). After setting up the equipment, the observer removed the pillowcase and then entered a hide with it positioned >25 m away, before starting the relevant playback. While it was impossible to be blind to location (inside/outside the colony), the observer was blind to caller type (familiar/unfamiliar). If a natural disturbance occurred (e.g. aerial alarm call or natural mobbing event), the trial was abandoned and repeated on another day (N=2).

During trials, the following data were recorded to a Dictaphone (Sony Europe Limited, Surrey, UK): response latency (time between start of playback and the first noisy miner to approach <10 m of the model); latency to give the first chur call (time between start of playback and the first noisy miner <10 m of the model to produce a chur call); maximum number of birds that responded; duration of the response (total time where noisy miner/s were <10 m radius of the model and responding to the model). Responding to the model was categorised as either looking toward the speaker, movement toward the speaker, and/or chur calling (Cunningham and Magrath 2017). As playbacks only commenced when no birds were within the immediate vicinity, we were able to count individual birds as they flew within the 10 m radius, aided by the patchy nature of the habitat. During noisy miner mobbing events, the number of birds generally peaks early on, with individuals flying in singly or in small clusters to the nearest trees where they perch until gradually leaving the mob as it subsides. Data collection ended at the end of the 10-min track. Thus, where no birds had responded or called by the end of the 10 min trial, latency was assigned a cut-off value of 600 s; if birds were still responding response duration was recorded as 10 minutes.