2.2 Data extraction
For each study, data were extracted that detailed: 1) year of study, 2) tracking method, 3) overall purpose of study, 4) country of study, 5) habitat type, 6) taxonomic group and 7) species tracked (see Supplementary Material 1). Quantitative information including the number of individuals tracked, the average number of tracking days and the average number of location points collected per individual tag were also collated (see Supplementary Material 1). This information was calculated from tracking summary results only when the total number of location points and deployment days were stated. Some articles did not provide sufficient information for these metrics to be calculated and these studies were omitted from further analyses. Tracking method included either radio transmitters (VHF), GPS or resource tracking with attached radio/GPS tags. Frugivore species data included taxonomic group at a species level, the number of species studied and the number of individuals per species per study.
The mean number of tag days was calculated for each study where data were provided. This was the mean number of days reported per study where data were collected across all tracked individuals per species. The mean number of tracking locations per species per study was also calculated when this information was provided (see Supplementary Material 1).
Mean and maximum seed dispersal distances were stated in 45 and 56 publications, respectively (see Supplementary Material 2). When publications presented estimates for different sized seeds or different seasons, we took an average across the different estimates as there was too much variability among studies to sub-divide data into different seed sizes, season or different sexes etc.
To explore what factors influenced dispersal distance, we extracted frugivore morphological traits and environmental variables from each study to be used as predictors for our models. Estimates of mean species body mass (g) for birds and mammals were extracted from Wilman et al (2014). We categorised species as either volant (i.e., capable of flying) or nonvolant. This is functionally informative, because in our data set a large proportion of the mammals studied were bat species, while birds included some flightless species such as cassowaries and emus.
If not mentioned in the publication, the geographic coordinates of each study were derived from Google Earth to determine whether the study was undertaken in a protected area. The study was defined as,such if the animal was recorded within a protected area (i.e., national park or reserve) at some point during the study. Each study site was categorised by biome; these data were taken from Ecoregion Snapshots: Descriptive Abstracts of the Terrestrial Ecoregions of the World. 2021, developed by One Earth and RESOLVE. Version 2021 (www.oneearth.org). The global distribution of studies was also mapped. Temperate climates are described as being > 35° or < -35° N and tropical climates between 23.4° and -23.4° N. The Human Footprint Index (HFI) value was extracted from Venter et al (2016) for each of the studies where GPS coordinates were included. These coordinates were either the coordinates given for the study area from the publications, or if not mentioned, the coordinates of where the animal was released. In addition to protected areas, HFI can be used as a proxy for identifying the impact of human-modified landscapes on seed dispersal. Two studies, Weir & Corlett (2007), and Wotton & Kelly (2012), were omitted from the seed dispersal analysis as there was no HFI data available for these locations. The HFI values extracted here use data from 2009 and do not align exactly with the dates of our studies estimating seed dispersal distances (ranging from 1987-2022), however we use these values as a broad-scale indicator of human disturbance.