Fine-scale changes in speed and altitude suggest protean movements in homing pigeon flights

Abstract

The power curve provides a basis for predicting adjustments that animals make in flight speed, for example in relation to wind, distance, habitat foraging quality and objective. However, relatively few studies have examined how animals respond to the landscape below them, which could affect speed and power allocation through modifications in climb rate and perceived predation risk. We equipped homing pigeons ( Columba livia ) with high-frequency loggers to examine how flight speed, and hence effort, varies in relation to topography and land cover. Pigeons showed mixed evidence for an energy-saving strategy, as they minimized climb rates by starting their ascent ahead of hills, but selected rapid speeds in their ascents. Birds did not modify their speed substantially in relation to land cover, but used higher speeds during descending flight, highlighting the importance of considering the rate of change in altitude before estimating power use from speed. Finally, we document an unexpected variability in speed and altitude over fine scales; a source of substantial energetic inefficiency. We suggest this may be a form of protean behaviour adopted to reduce predation risk when flocking is not an option, and that such a strategy could be widespread.

Emergence of splits and collective turns in pigeon flocks under predation

“…According to our model, early collective escape of a flock may be the result of protean movement of its members. Thus, given the variability of the emerging collective turns and splits, the escape path of the group may become less predictable [ 59 , 60 , 76 ], counteracting anticipation-based hunting strategies of the predator [ 58 , 67 , 96 ].…”

“…Specifically, the discrete escape manoeuvre in our model involves a turn away from the heading of the predator [ 46 ] during multiple time steps. These turns are not fixed [ 59 , 60 , 76 ]; for each manoeuvre the turning angle and duration of turning are drawn independently from two gamma distributions (electronic supplementary material, table S3). We parametrized these distributions to our measures of individual turns in the empirical data ( figure 2 b and electronic supplementary material, figure S2).…”

“…U-turns in fish) [ 49 , 56 – 58 ]. Empirical data further suggest that the manoeuvres may not be fixed but varying and unpredictable in terms of turning angle and speed (referred to as ‘protean’ movement) to increase the complexity of the escape path of a prey [ 58 – 60 ].…”

The role of wingbeat frequency and amplitude in flight power

“…An examination of accelerometer data revealed some slight variation in sampling rate between logger types (up to 3 Hz), which was accounted for in the calculation of wingbeat frequency. Tags were attached to the back feathers using Tesa tape (Wilson et al, 1997) in all species apart from pigeons, where tags were attached via Velcro strips glued to the back feathers (Garde et al, 2021, Biro et al, 2002). The total mass of the tag, including housing and attachments, was under 5% of bird body mass and 3% in most cases.…”

“…Yet at fine-scales, disentangling the impact of the biological and physical environment on flight costs can be challenging, given that a range of factors often vary simultaneously. These include the topography birds are flying over, individual position within a flock (Garde et al, 2021, Portugal et al, 2014) and social context (Sankey & Portugal, 2019; Usherwood et al, 2011), as well as factors that vary over longer timescales including the birds’ immunological state (Hicks et al, 2018), and physical factors such as wind speed, turbulence and air density (Bishop et al, 2015; Furness & Bryant, 1996; Sapir et al, 2010). High-frequency data from animal-attached loggers have proved powerful in this regard, as the signal from onboard accelerometers can be used to quantify second-by-second changes in wingbeat frequency (Cochran et al, 2008; Sato et al, 2008; Van Walsum et al, 2020), and potentially other kinematic parameters (Taylor et al, 2019).…”