Skip to main content

Data from: Fruit traits of pioneer trees structure seed dispersal across distances on tropical deforested landscapes: implications for restoration

Cite this dataset

Camargo, Paulo H. S. A.; Pizo, Marco A.; Brancalion, Pedro H. S.; Carlo, Tomás A. (2020). Data from: Fruit traits of pioneer trees structure seed dispersal across distances on tropical deforested landscapes: implications for restoration [Dataset]. Dryad.


  1. Pioneer trees with fleshy fruits are typically planted in restoration projects to attract frugivores as a mean to increase dispersal and accelerate forest regeneration. However, differences in fruit traits of pioneer trees can potentially influence dispersal and their restoration outcomes.
  2. Here we investigated the effects of bird and plant traits, and distance to forest fragments, on the seed rain using a tree-planting experiment replicated in 12 deforested sites in Brazil. Factors were fruit traits of pioneer trees (wind-dispersed, bird-dispersed with lipids or with carbohydrates, and controls) and distance (10, 50, 300 m) from forest fragments.
  3. We found that density and richness of birds and seeds decreased exponentially with distance from fragments, yet these effects were minor compared to the effects of fruit traits on the structure of the seed rain.
  4. Overall, plots with fleshy-fruited pioneers attracted much greater bird activity and seed dispersal than plots with wind-dispersal pioneers and the controls. For instance, plots with carbohydrate-rich fruits received more than twice the average species richness and density of birds and seeds of plots with lipid-rich pioneer trees, surpassing wind-dispersed pioneers by more than 80%, and controls by over 90%. Furthermore, the fruit trait treatments resulted in morphological shifts in the average traits of visiting birds. Significant differences in bill gape and flight capacities (wing-loading) were associated with the differences in the seed rain associated with each treatments.
  5. Synthesis and applications. Understanding how trait-matching processes mediating mutualistic seed dispersal by frugivores interact with distance-dependent dispersal limitation on deforested tropical landscapes is critical for improving forest restoration efforts. This is especially relevant in the context of applied nucleation. As shown here, avian seed dispersal can thus be manipulated in restoration projects in order to increase connectivity and speed up forest recovery and the provision of the multiple ecosystem services that follow forest succession.


Study sites

We conducted the experiment in 12 sites in Paranapanema, São Paulo, Brazil (23°23'S, 48°43'W, Fig. 1). Paranapanema is located at ca. 600 m a.s.l. on the watershed region of the Alto Paranapanema river (Cielo-Filho et al., 2009). Average precipitation is 1.407,9 mm mostly during December-March, and a mean temperature of 18 °C (Cielo-Filho et al., 2009). Atlantic forest fragments cover ~6% of the Paranapanema landscape (Fundação SOS Mata Atlântica, 2013). Our study sites were located on private cattle pastures and fragments of primary and secondary semideciduous forest >30 years old and 12.2-98.8 ha in size.


Experiment set up

Between December 2016 and November 2018 we conducted a full-factorial experiment using two explanatory variables: distance to forest fragment at three levels: 10, 50, 300 m, and fruit traits of pioneer trees at three levels: Heliocarpus popayanensis Kunth with wind-dispersed fruits, Acnistus arborescens Schltdl, with bird-dispersed fruits, and Trema micrantha (L.) Blume also with bird-dispersed fruits, plus a control treatment without any tree. Thus, the 4 x 3 design yielded 12 types of plots that were replicated once in each of 12 localities in Paranapanema, totaling 144 (Fig. 1). The relative position of each Fruit trait treatment plot was randomly determined within each distance class, and spaced the others by a distance equal to their distance to the from fragments. The two bird-dispersed trees (Acnistus & Trema) differed in their fruit traits: Acnistus has larger carbohydrate-rich fruits, while Trema has small lipid-rich fruits (See Table S1). Plots measured 4.5 x 4.5 meters and were fenced with barbwire to keep cattle out (Fig. S1). Pioneer trees from all species had similar height, crown, and fruit crop (Tables S1-S2) and were grown nurseries until a height of ~1.5 m. A single tree was planted at the center of plots requiring trees, and fertilized with 300 g of 15-18-28 (NPK) every six months. Grasses were mechanically controlled during the entire experiment.


Bird activity

We recorded bird visits in all experimental plots using a combination of video recording and focal observation with effort equally distributed among plots. We used 28 video cameras from four different brands (ten Bushnell Trophy Cam, six Bushnell Trophy Cam HD, six Tigrinus, and six GoPro Hero 3). Camera traps operated 24 h/day recording 88,714.7 h, while the GoPro Hero 3 cameras filmed for 2 hours per day for, 288 h. To prevent sampling bias associated with camera models, all units were systematically rotated to sample for the same amount of time in each experimental plot across the twelve study sites. In addition, in each plot one observer (P. Camargo) recorded bird visits using a pair of binoculars from a distance of 50m for 20 minutes, during morning hours (0730-1030), once a month, for 5 months.


Seed rain

We sampled the seed rain of all plots using one 0.25-m² seed trap lined with a 0.2-mm nylon mesh and located at the center of the enclosure and covered with a 2.5 x 2.5 cm wire mesh (Fig. S1). The sticky paste Formifuu® was applied to the support posts of traps to exclude ants. Traps were collected each month and seeds counted and identified with the aid of a dissecting scope and reference books and seed collections. Seeds from grasses or from the species of the pioneer tree species in a plot were excluded from analyses.


Bird and seed traits

Bird traits included % fruit in diet (Wilman et al. 2014), wing-loading as a proxy of bird movement capacity, and gape-width as a proxy to their upper seed size dispersal potential. Wing-loading was calculated as:


where BM is the bird body mass and WL is wing length (species measurements were obtained from Rodrigues et al. 2019). As loading value increases, the movement capacity of a bird species decreases. We classified dispersed seeds by dispersal mode: bird-dispersed, wind-dispersed, and gravity-dispersed. In addition, we calculated seed mass and approximated lipid content of fruit in a rank scale (1 - 0 to 10% lipids, 2 - 10 to 20%, 3 - 20 to 30% and 4 - above 30%) adapted from Bello et al. (2017).


Coordenação de Aperfeicoamento de Pessoal de Nível Superior, Award: Finance Code 001

São Paulo Research Foundation, Award: 2016/17194-0

National Council for Scientific and Technological Development

National Science Foundation, Award: DEB-1556719