1. Animal-borne logging devices are now commonly used to record and monitor the movements, physiology and behaviours of free-living animals. It is imperative that the impacts these devices have on the animals themselves is minimised.

2. One important consideration is the interaction between the body mass of the animal, and the mass of the device.

3. Using captive homing pigeons, we demonstrate that birds lose the equivalent amount of body mass compared to that of the logging device attached. With our experiments, we calculated that the compensatory mass loss because of the logging device equates to a total loss of 1,140 kJ of energy to the bird, over the 25-day period. This equates to 32% per day of their total daily energy budget.

4. We suggest that practitioners of biologging give due consideration to the possibility of a device-induced decrease in body mass when making decisions regarding device size, and when considering the period of the time of the year at which devices are attached.

5. It appears, based on the results of the present study, that device attachment is likely to be most disruptive during periods of regulated mass change, especially when periods of mass gain precede periods in which stored energy reserves are extensively utilised.

6. These findings have significant consequences for anyone using biologging technology on both wild and captive volant animals. Further studies utilising captive birds are now needed to fully understand how context- and species-dependent physiological responses to externally attached devices are.

Birds and housing

A group of 18 homing pigeons (hereonin referred to as pigeons) were housed at Royal Holloway University of London (Egham, UK). All birds were a minimum of 15 months old and had lived together since hatching. The sex of the birds was a 55/45% split (males/females) (mean body mass at start of the study, 590 ± 75.6 (SD) g). Birds were kept in a pigeon loft (dimensions 3.6 m (long), 2.4 m (wide)) with ad libitum access to food and water (see Portugal et al., 2017a; Portugal et al. 2017b for further husbandry details). Wooden perches (n = 20) were attached to the sides of the loft, in arrangements of six perches in horizontal rows at three heights (1 m, 1.30 m, 1.60 m), plus two additional single perches (1.30 m).

Body mass measurements and artificial mass attachment

Birds were weighed twice weekly for four weeks, by being placed in a cotton bag and weighed using digital scales (0.1 g accuracy, Scales and Balances, Thetford, UK), before each pigeon was fitted with artificial mass in week five. The artificial devices mimicked the size and mass of typical logging devices (30 g, 3.7 x 2.4 x 0.8 cm), both those that are generally used for studies with birds and those which have been used previously with homing pigeons [Usherwood et al. 2014; Pettit et al. 2016; Taylor et al. 2019].

The artificial mass was attached to the back of the pigeons using Velcro strips and epoxy glue (total package mass 30 – 31 g, 5% of the mean body mass of the pigeon group at start of the study; see Sankey and Portugal (2019) and Sankey et al. (2021) for further details on logger attachment to the pigeons). Birds were then weighed twice weekly for the next 18 days (see Figure 1). On day 18 the artificial loggers were removed, and birds were again weighed twice weekly for a four-week period. Following the four-week period, 10 birds were randomly selected to have the artificial loggers reattached, with the remaining 8 birds having no devices attached. All birds were then weighed twice weekly for a further 21 days (Figure 1).

Statistical analysis

Data were analysed using linear mixed models implemented in the lme4 [Bates et al. 2015] package of R v3.2.2 [R Core Team 2016]. The significance of random effects in mixed models was determined using likelihood ratio tests, and the significance of fixed effects in mixed models was determined using t-tests with Satterthwaite approximations to degrees of freedom implemented in the lmerTest package [Kuznetsova et al. 2016]. The model included a random intercept for individual identity and a random slope for the effect of time, which varied among measurement blocks. Data for time and mass were mean centred and scaled to unit variance prior to analysis. Full statistical findings can found in Supporting Information: Statistical Outputs.