Data supporting "Airflows redefine the cost of aerial transport in the wild"

These files support the findings of the study "Airflows redefine the cost of aerial transport in the wild". The aim of this study was to update existing equations defining the cost of transport for flying animals, mainly developed using data extracted from wind tunnels under laboratory settings, with detailed information about flight behaviour in the wild obtained from biologging technology and radar. Estimating the proportion of time flying animals spent in flapping flight allowed us to redefine how energetic costs vary with body mass and with airflows, and the selective benefits of different forms of aerial locomotion. These datasets gather flight segments from 50 bird and bat species collected using biologgers and radar. The biologging dataset includes information about >194000 flight segments from 850 individuals and 38 species of birds and bats (sized 250 g to 11 kg) collected using GPS and acceleration (ACC) sensors over 17 years (2007 - 2024). The radar dataset contains information about > 5000 segments from passerines species sized 12.5 - 21.2 g, collected during 4 years (2018-2021). Where there is interest in using any of these datasets, we strongly encourage people to contact the researchers who collected them (individuals are listed as authors of this study and contact information is provided in the table along with associations to each dataset). Each data owner has deep knowledge of the specific study species and knows the details and context in which datasets were collected. For extensive details about the aim and results of the study and the methodology underlying the datasets shared in this folder, please refer to the full text and material and methods of the paper.All the scripts used to produce and analyse these datasets are written in R, deposited on Github (https://github.com/mscacco/COT_publ/tree/main) and linked to the version published on Zenodo (10.5281/zenodo.21915973). The codes follow the files structure below and refer to the specific file names needed to run the analysis. ----------------File structure:----- The table "Table_allDatasets_contacts_permits.pdf" contains the contact information of the data owner, ethical permits and study id associated to all datasets (biologging and radar). ----The folder ModelData contains 3 files: - "RADAR_finalSummaryDataset_perEcho_COTvariables_WFF-month_echoDurFilter.csv" segment-level radar data used for the models in the last steps of the radar scripts. - "BIOLOGGING_finalSummaryDataset_perSegment_fromFix+COTvariables_Feb2025.csv" segment-level biologging data, produced in script 7 and used in the models (scripts 8 and 9) - "MRCtree_DendroPy_from1000_Ericson_Feb2025_pruned.tre" bird phylogenetic tree, used in the phylogenetic models (scripts 8 and 9) The RADAR and BIOLOGGING summary datasets contain summary values per segment, each associated to:- morphological information (at the species level, obtained from published datasets), - flight metrics and behavioural information obtained from the GPS trajectories and radar echoes, - eCOT (effective cost of transport), defined specifically for this study and expressed in joules per kg of body mass and metre travelled,- environmental information (wind and uplift potential) obtained from ERA5 hourly reanalyses, interpolated in space and time at each location and later summarised per segment. ----The folder BiologgingData contains 2 files: -"allStudies_allTags_allFlightSegments_binded_birdsBats_thresholdClass_transfGs_March2024_noDupl.rds" 38 species; 853 individuals; 195'715 segments; 1'524'121 observations (GPS points).This dataset contains all GPS data actually used for this study, that is, the commuting segments extracted from the thinned/subsampled original tracks deposited on Movebank. This dataset is produced at line 330 of script 3 (see Github repo), and contains all studies after duplicated individuals across studies have been removed. - "FinalDf_perPoint_VedbaGs_flappingProbs_ENV.rds" 38 species; 853 individuals; 195'715 segments; 1'524'040 observations (81 GPS points less than the file above, removed at line 27 of script 4, where the tail of VeDBA values larger than 2 g were removed). This dataset is produced at the end of script 5B and imported in script 6 in the COT_publ github repo, and in addition to the previous file, contains all the environmental variables annotated to each GPS point. At line 81 of script 6, this dataset gets summarised per segment, and segments lasting longer than 10 hours get removed (1157 segments in total, see SM for details). This leads to a total of 194'558 segments, 850 individuals and 38 species, as reported in the SM of the paper. ----The folder RadarData contains 2 files, all used to run script "1-9_RadarData_processing-analysis.R": - "SpeciesList_midjuly_midseptember_Desert_Med_Species.csv" This dataset contains a list of species that were assigned to each "body mass group" (labelled WFF group) based on flapping frequency, body mass, phenology and expert opinion. - "Extended_Radar_data_midjuly_midseptember_no_bats.csv" Raw radar data echo per echo, used in the workflow to produce the results. ----The folder InputData contains 2 files: - "wingMorphology_perSpecies_Feb2025.csv" this file contains wing measurements gathered by the authors from the literature and used in script 0B to calculate wing loading for the species used in the study. - "FlappingPowerValues_Guigueno2019.csv" this file contains metabolic measurements supporting the flapping metabolic rate model in script 0C, applied in scripts 7 to 9 for the calculation of effective COT. This dataset was compiled by, and used in, Guigueno et al 2019 "Flight costs in volant vertebrates: A phylogenetically-controlled meta-analysis of birds and bats." Comp. Biochem. Physiol. A. Mol. Integr. Physiol. 235, 193–201