Ecological and environmental traits can influence avian escape behaviour but most data underpinning our current understanding relates to continental and temperate areas and species. We conducted a phylogenetically controlled comparative analysis of flight-initation distance (FID) against a variety of environmental, behavioural and life-history attributes for Sri Lankan birds (202 species; n = 2540). As with other studies, body mass was positively associated with FID, and longer FIDs occurred in areas where human population density was lower. We also found that the effect of human population density was more pronounced in larger birds. Birds that were in groups when approached tended to have longer FIDs. Unlike the findings of other comparative analyses, based mostly on continental, temperate populations, most other ecological variables did not feature in the best models predicting FID (time of year, breeding system, clutch size, habitat, migratory behaviour, development [altricial/precocial], elevation and diet). Thus, some associations (body mass and exposure to humans) may be universal, while others may not manifest themselves among tropical avifaunas. Further tropical datasets are required to confirm truly universal associations of environmental and ecological attributes and escape distances among birds.

We measured flight-initiation distance (FID), the distance at which birds commence escape from an approaching person, a widely used method of indexing escape behaviour in animals (methods follow Blumstein 2003). Observations were made of 2540 individuals from 202 species of birds at 124 locations across Sri Lanka. Most distances were collected using a laser range-finder, except where obscuring vegetation or reflection prevented use, in which case distances were paced. FIDs were collected in fine weather only. Observers wore dull pants, shirts, and hats. We recorded the distance at which approaches commenced (StDist; starting distance), which is usually positively associated with FID. All apporaches involved direct line of sight. We sampled singlets or single-species groups. Repeat sampling of individuals was avoided by collecting data at many sites, not resampling the same location, and not sampling the same species < 50 m from a point at which it had already been sampled. FIDs were recorded in a range of habitats by opportunistically collecting FIDs across as many habitats as could be visited. FIDs for perched birds were corrected for observer height (Livezey and Blumstein 2016), and straight line distances were calculated. 

Every FID was geolocated, enabling each to be allocated a corresponding elevation (Data aggregated to 30 second grid resolution - https://csdms.colorado.edu/wiki/Data:CGIAR-CSI_SRTM), human population density, and location (geographically proximate areas, that exhibit the same prevailing regimes of human occurrence, land use and activity). Date and time of day were collected along with StDist, for each FID. Human population density (for 2015) mapped in 30 seconds resolution and extracted from (https://sedac.ciesin.columbia.edu/data/collection/gpw-v4/sets/browse). Habitat at locations was classified as “Open” (e.g., wetland, grassland), “closed” (i.e., forest, woodland) or “both”. Finer habitat classifications are reduced to open, closed or both following Blumstein (2003). Because group size was highly skewed towards single individuals and estimation of exact group size when with others was often ambiguous, this was categorized as either solitary or with other birds at the time of encounter. Likewise, elevation was clearly bimodally distributed and so was categorically divided into highland sites > 1500 m and lowland sites < 1000 m above sea level.

Blumstein DT (2003) Flight-initiation distance in birds is dependent on intruder starting distance. J Wildlife Manage 67:852-857. doi:10.2307/3802692

Livezey KB, Blumstein DT (2016) “Shortest-distance” method is more accurate than conventional method in estimating flight-initiation distances for close perched birds. J Ornithol 157:923-925. doi: 10.1007/s10336-106-1337-3

For full details of methods with justifications and analyses please see Ekanayake et al. (2022) Behavioral Ecology and Sociobiology. Please also cite this reference if you use these data.