Seed dispersal by frugivorous animals is important for plant mobility, regeneration, and persistence. Human-caused landscape change is thought to disrupt seed dispersal, but evidence is scarce. We performed a comprehensive meta-analysis on the effects of habitat spatial pattern on frugivory and seed dispersal. We found 233 effects from 71 studies. At a patch or local scale, altered habitat spatial pattern was measured as declining patch size, increasing patch isolation, or habitat edge (vs. interior). At a landscape scale it was measured as declining amount of habitat, increasing mean patch isolation, increasing number of patches, or increasing habitat edge in the landscape.

We found overall negative effects of altered habitat spatial pattern on: (i) the quantity of frugivory or seed dispersal, (ii) the number of species involved in a plant-frugivore interaction, and (iii) seed dispersal distance. Moderator variable analysis was only possible for the first of these. It revealed negative responses of the quantity of frugivory or seed dispersal to habitat loss at both the local scale (declining patch size), and the landscape scale (declining habitat amount), but little evidence for a response to habitat edge at either scale. In addition, altered habitat spatial pattern reduced the quantity of frugivory or seed dispersal more strongly in temperate than tropical areas. Finally, the few-recorded effects of landscape-scale fragmentation per se (increasing patch density or edge density) on the quantity of frugivory or seed dispersal were mixed and weak. Our meta-analysis reinforces the notion that habitat loss is a major threat to frugivory and seed dispersal by animals, and reveals an insufficiency of studies of the effects of habitat fragmentation per se. Thus, based on the current literature, we conclude that maintaining and increasing habitat amount is vital for maintaining seed dispersal by frugivorous animals.

We conducted a review of the scientific literature following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol (Moher et al., 2009). We included all records found through May 2020. We used the following sequence of search terms in the Web of Science search engine: "habitat fragm*”OR edge or isolation or connectivity OR “number of patches” OR “patch size” OR “forest cover” OR “forest loss” OR “deforestation” OR “habitat loss” AND "seed dispers* OR “seed removal” OR frugivor* OR “dispersal distance” OR “seed rain”.

The initial search resulted in 2141 published articles that were further screened to include only English-language studies that (i) evaluated the effects of habitat spatial pattern on any measure of frugivory or seed dispersal of zoochoric species; and (ii) provided sample sizes and quantitative measures of frugivory or seed dispersal, i.e. mean estimates and error measures for studies comparing treatments vs. controls, or correlation coefficients for gradient studies. After title and abstract evaluation, we excluded articles that did not meet these inclusion criteria, which left 99 articles. We completely read the methodologies of these 99 papers and excluded 28 that did not present complete information or did not evaluate the effects of habitat spatial pattern on frugivory or seed dispersal of zoochoric species. We ended up with 71 studies that measured the effects of habitat spatial pattern on frugivory or seed dispersal measure on different ways.

README.xlxs contains the following datasheets:

Cazetta&Fahrig_Dryad.csv: contains the information used to perform the meta-analysis. The information included:

-Location: temperate or tropical

-Habitat: forest, savanna, desert, semi-natural habitats

-Response level: community or species

- Spatial pattern: habitat amount, patch size, habitat edge, isolation, number of patches, edge density

-Scale: patch or landscape

-Yi= effect size

-Vi = corresponding sampling variance

References.xlxs: contains a list of the 71 published articles included in the meta-analyis, as supporting information.