Network metrics are widely used to infer the roles of mutualistic animals in plant communities and to predict the effect of species' loss. However, their empirical validation is scarce. Here we parameterized a joint species model of frugivory and seed dispersal with bird movement and foraging data from tropical and temperate communities. With this model we investigate the effect of frugivore loss on seed rain, and compare our predictions to those of standard coextinction models and network metrics. Topological coextinction models underestimated species loss after the removal of highly-linked frugivores with unique foraging behaviors. Network metrics informed about changes in seed rain quantity after frugivore loss. However, changes in seed rain composition were only predicted by partner diversity. Nestedness, closeness, and d’ specialization could not anticipate the effects of rearrangements in plant-frugivore communities following species loss. Accounting for behavioral differences among mutualists is critical to improve predictions from network models.

See methods section in : "Can network metrics predict vulnerability and species roles in bird-dispersed plant communities? Not without behavior" by  Morán López, Teresa; Espíndola, Walter; Vizzachero, Benjamin S.; Fontanella, Antonio; Salinas, Letty; Arana, César; Amico, Guillermo; Pizo, Marco; Carlo, Tomas A.; Morales, Juan Manuel Morales. Ecology Letters in press.

Data consists of one Excel file from each study site. Each file contains a tab for (1) a diagram of the each grid and the arrangement of sampling cells (2) plant transect data around the plot, (3) foraging observations of individual birds with information of movements across grid cells (10 x 10 meters) of the study sites, (4) fruit availability data per week for each of the grid cells of study cites (only ripe fruits ready to be eated included in counts).

Foraging observations: individual birds were followed (focal individuals) and the location (grid cell) were foraging ocurred was noted, including the plant species consumed and the number of fruits ingested (or the number of fruit pieces taken in the case of large fruits eaten in piece-meal fashion). Observations lasted 3-5 hours in the mornings or afternoons and included 2-4 observers. Frugivory were also recodedfor non-focal individuals and noted as "incidental".

Plant transects: four transects at each of three distance classes (100, 200, 300 meters x 2 m wide) from study plots were established to sample the abundance of ornithocorous plant species beyond the plots. Transects increased in length with distance to maintain sampling proportional. At 100 meters they were 100 m in length, at 200 meters they were 125 m, and at 300 m they measure 175 m in length. The height and canopy diameter of each ornithochorous plant species was measured for those plants intersected by the transect areas.

Plot phenology: ripe fruits were estimated within each cell grid once per week. Unripe fruits were not counted.

Phenology outside plot: ripe fruits were tallied in a sample of tagged plant species round the plot every week.

Weekly data (5-10 weeks) of the ornitochorous (bird-dispersed) plant species' ripe fruit avaialbility, bird fruit foraging sequences and movements, and seed rain collected within six 1.0 - 1.6 ha study sites: Powdermill Nature reserve (Rector PA); Finca Montaña (University of Puerto Rico, Aguadilla, Puerto Rico); Morro Calzada (Moyobamba, Peru); Oxapampa -Perú (CDS https://www.cds-peru.org/); Rio Claro, Sao Paulo, Brazil; Yao-Yao National Forest, Bariloche, Argentina.