Dataset DOI: 10.5061/dryad.g1jwstr3h

Description of the data and file structure

Four datasets are provided, including different columns with information.

Files and variables

File: Dataset_S1_Network_Plant_Frugivore_Complete.xlsx

Description: Plant-frugivore network data: Dataset S1. 

We sourced the plant-frugivore interaction data from a comprehensive global dataset compiled by Fricke and Svenning (2020). This dataset consists of 410 local interaction networks, each constructed based on direct observations of animals interacting with fruits or seeds or from seeds retrieved from captured animals. These networks are bipartite, representing interactions between animal and plant species. While some networks recorded interaction frequencies, we used them in a binary format (presence or absence) due to inconsistencies in how frequencies were recorded across the dataset. This study focused on a subset of networks from tropical regions, defined geographically as the area between the Tropic of Cancer and the Tropic of Capricorn. This focus was chosen due to the key role of frugivorous vertebrates in seed dispersal within tropical ecosystems. The initial subset contained 251 spatially and temporally distinct tropical interaction networks (see Dataset S1). To account for repeated sampling at the same location over time, we aggregated unique plant-frugivore interactions from networks with the same coordinates into single composite networks, resulting in a final dataset of 181 geographically distinct networks (n = 145 non-composite networks; n = 36 composite networks formed by merging 2–11 original networks, mean ± SD: 2.94 ± 1.67). Prior to data analysis, the nomenclature of animal and plant species was updated to reflect taxonomic revisions made since the compilation of the original dataset. Animal species names were standardised following HBW and BirdLife International for birds, the International Union for Conservation of Nature (IUCN 2022) for mammals, and Oskyrko et al. (2024) for reptiles. Plant species names were standardised using the World Checklist of Vascular Plants (WCVP).

Variables

File: Dataset_S2_Fruit_and_Seed_Traits.xlsx

Description: Plant functional trait data: Dataset S2

We compiled data on frugivory-related plant traits, including fruit and seed dimensions (length and width), fruit type (e.g., berry, drupe, capsule; see Dataset S1), dehiscence (presence or absence), and seed fleshy appendages (presence or absence) for all 2,406 plant species that were present in the 181 ecological networks (see supplementary methods in Text S1 for details). Trait data were complete for 94.4% of the species (n = 2,272) and were primarily sourced from floras and monographs (79%), supplemented by scientific articles (19%) and photographs (2%) (see Dataset S2 for data and sources). Plant trait data were preferred under the accepted species name; however, when unavailable, they were retrieved from taxonomical homotypic synonyms following the WCVP. Methods for measuring fruit size vary considerably across studies. Some approaches consider fruit length or width, while others use a composite measure that sums fruit length and width. Both approaches present limitations: fruit length or width may not always correspond to the fruit’s longest or shortest axis, and combining length and width into a single metric may fail to capture the constraints frugivores face during consumption. To address these inconsistencies, we defined fruit size as the smallest reported axis of the dispersal unit (either fruit or seed), as this dimension is most likely to represent the key constraint for frugivores during ingestion. For indehiscent fruits, frugivores generally interact with the entire fruit, as the fleshy pericarp remains intact. In contrast, for dehiscent fruits, frugivores often interact with the seeds rather than the usually dry pericarp, especially when fleshy structures such as arils, sarcotestae, or elaiosomes are present. Accordingly, for the trait matching analyses, we selected the smallest axis of the dispersal unit, whether it was fruit length or width for indehiscent fruits or seed length or width for dehiscent fruits. Throughout the study, we refer to this measurement as ‘fruit size’.

Variables

Dataset S2: Plant trait data, fruit and seed traits

File: Dataset_S3_Frugivore_Traits.xlsx

Description: Frugivore functional trait data: Dataset S3

We compiled data on frugivory-related animal traits, including body mass, degree of frugivory (percentage of the diet composed of fruit), beak dimensions, and gape size for all 1,226 animal species that were present in the 181 ecological networks. Trait data were complete for 98.9% of the species (n = 1,213) and were primarily sourced from established databases: AVONET, COMBINE, EltonTraits, ReptTraits (see Dataset S3 for data and sources). Animal trait data were preferred under the accepted species name; however, when unavailable, they were retrieved from taxonomic synonyms, following HBW and BirdLife International for birds and Soria et al. (2021) for mammals. In cases of discrepancies between accepted names and synonyms, the mean trait values were used. Dietary information was available for 1,102 species, with species-specific data for 93.7% of them; for the remaining 6.3%, diet was inferred at the genus or family level (see Wilman et al. 2014 for details). We selected body mass as the primary trait because it provides a comparable metric across birds, mammals and reptiles. While bird-specific traits such as beak volume and gape size are often used in trait matching studies, body mass was chosen for its broader applicability across taxa. This decision was supported by the strong correlations observed between body mass and beak volume (r = 0.81) and gape size (r = 0.85) in birds. The final dataset was predominantly composed of birds (75.8%), followed by mammals (23.3%) and reptiles (0.8%).

Variables

Dataset S3: Animal trait data, frugivore traits

File: Dataset_S4_Network_TraitMatch_Env_v2.xlsx

Description: 

Human disturbance and environmental data: Dataset S4

Human disturbance and environmental data were assembled and extracted for each plant-frugivore network at spatial resolutions of 1 km and 5 km using ArcGIS Pro 3.3. Human disturbance data were derived from the Global Human Modification (gHM) dataset (Kennedy et al. 2020), which provides an integrated (i.e., cumulative) measure of human alteration of terrestrial landscapes at 1 km resolution. This continuous metric (ranging from 0 to 1) represents the proportion of land modified by human activities and integrates 13 indicators of disturbance, including human population density, built-up areas, cropland, livestock, major roads, minor roads, two-track roads, railroads, mining, oil wells, wind turbines, powerlines, and night-time lights, based on spatially explicit global datasets with a median reference year of 2016 (Kennedy et al. 2020). Environmental data were sourced from CHELSA V2.1, which provided annual temperature (°C), precipitation (mm), and NPP (g C/m²) at a 1 km resolution, averaged over the period 1981–2010. Although precipitation and NPP showed a strong correlation (r = 0.71), this remained below the commonly accepted multicollinearity threshold of r = 0.80. Consequently, both variables were retained, as their influence on frugivory may act through distinct ecological mechanisms. Additionally, AET (mm) was extracted as the annual average across the period 1982–2019 at a 1 km resolution. All network-level data is provided in Dataset S4.

Variables

Dataset S4: Environmental and human impact data per network

Code/software

Analytical workflow to reproduce main results from Structural Equation Models and Figure 3 are provided in the following files, including the R script and the R data to be loaded into R.

• Analytical_workflow_Guerra_et_al._Ecol_Let.R
• Data_Analytical_workflow_Guerra_et_al._Ecol_Let.RData

Access information

Other publicly accessible locations of the data:

• In the supporting information of the article.

Data was derived from the following sources:

• Fricke, E.C. & Svenning, J.-C. (2020). Accelerating homogenization of the global plant–frugivore meta-network. Nature, 585, 74–78.

• Karger, D.N., Conrad, O., Böhner, J., Kawohl, T., Kreft, H., Soria-Auza, R.W., et al. (2017). Climatologies at high resolution for the earth’s land surface areas. Sci Data, 4, 170122.

• Kennedy, C.M., Oakleaf, J.R., Theobald, D.M., Baruch-Mordo, S. & Kiesecker, J. (2020). Global Human Modification of Terrestrial Systems.

• Oskyrko, O., Mi, C., Meiri, S. & Du, W. (2024). ReptTraits: a comprehensive dataset of ecological traits in reptiles. Sci Data, 11, 243.

• Soria, C.D., Pacifici, M., Di Marco, M., Stephen, S.M. & Rondinini, C. (2021). COMBINE: a coalesced mammal database of intrinsic and extrinsic traits. Ecology, 102, e03344.

• Wilman, H., Belmaker, J., Simpson, J., de la Rosa, C., Rivadeneira, M.M. & Jetz, W. (2014). EltonTraits 1.0: Species-level foraging attributes of the world’s birds and mammals. Ecology, 95, 2027–2027.