2.1 Study area

This study was conducted in the Porto Ferreira State Park (PFSP), a Conservation Unity belonging to São Paulo State Forest Institute. Located in Porto Ferreira city (21° 49’ S and 47° 25’ W), the park covers an area of 611.55 hectares, of which about 400ha are represented by semideciduous forest (SF) and 180 ha by savanna. According to Köppen’s system the regional climate is Cwa, with two well defined seasons: dry winter and rainy summer (^{Bertoni et al., 2001}).

The experiment was installed in an abandoned pasture of 10.75ha (^{Tabanez et al., 2003}) located a minimum of 150m from the savanna and 230m from the forest edges. Although still dominated by exotic grasses, sparse trees and shrubs naturally developed in the area, among which are: Baccharis dracunculifolia DC. (Asteraceae), Byrsonima intermedia (Malpighiaceae), Myrsine coriacea (Primulaceae), Zanthoxylum rhoifolium (Rutaceae), Solanum spp. (Solanaceae), Trema micrantha (Cannabaceae) and Aegyphila sellowiana (Lamiaceae).

2.2 Procedures

For the evaluation of seed deposition, traps were installed in four experimental models: a) under natural perches (NPs); b) under simple artificial perches (SAPs); c) under elaborate artificial perches (EAPs) and d) in a control area (CA). In each sample set were allocated 10 seed traps, totaling 40 units. Traps were made of wood and PP-NWF (polypropylene non-woven-fabric), positioning the tissue about 30cm from the ground to reduce the chances of seed predation by terrestrial invertebrates as well as the conditions for its germination, as suggested by ^{Bocchese et al. (2008)}.

SAPs consisted of a 3m tall bamboo and a crossbar 1m long attached perpendicularly at the top; EAPs were built with Eucalyptus logs 7 m tall and 3 crossbars 1 m long, 1,5 m distant from each other counting from the top. The arrangement of artificial perches in the field is shown in Figure 1. NPs corresponded to 5 trees naturally developed in the abandoned pasture, randomly chosen, located around the set of artificial perches. Under the canopy of each tree were installed two seed traps. For the CA, traps were installed in the open field, adjacent to the artificial perches, according to Figure 1.

Figure 1 Arrangement of artificial perches and seed traps in the experimental area. 

This experiment was conducted between June 2012 and May 2013. Traps were emptied at intervals of 20 days (except when it was necessary to fix the seed traps: between August 20 and August 30, 2012; October 10 and October 19, 2012; December 19, 2012, and January 15, 2013). Seeds were sorted, counted and identified by comparison with the surrounding fertile material and consulting specific bibliography (^{Lorenzi, 2008}, ^2009a, ^b; ^{Kuhlmann, 2012}). Only bird-dispersed seeds were considered; unidentified species were classified as morphospecies, according to ^{Melo et al. (2000)}.

Plant species were ecologically classified according to: a) origin: native or exotic; and b) ecological group: pioneer, early secondary and late secondary. The climax species category was not included in successional classes because many typical understory plants occur in both stages, more mature and other succession phases, according to ^{Gandolfi et al. (1995)}. The inclusion of species in ecological groups followed descriptions of ^{Batalha and Mantovani (2001)}, ^{Paula et al. (2004)}, ^{Pinheiro and Monteiro (2009)} and field observations. Scientific nomenclature and family classification follow the ^{Reflora (2014)}.

The use of artificial perches by birds was also evaluated during one year, allowing recording the main visitor species and seasonal variations (^{Holl, 1998}). Visits were recorded for 120h, being 10h of monthly observations, using 8x40mm binoculars, from places where it was possible to view the SAPs and EAPs at the same time. Natural perches were not included in the systematic observations due to the impossibility to visualize all of them at the same time, invalidating data comparison. Recorded species were grouped into feeding guilds, according to ^{Sick (1997)} and field observations. Taxonomic classification follows the Brazilian Ornithological Records Committee (^{CBRO, 2014}).

2.3 Data analysis

The Spearman Correlation test was used to evaluate the degree of association between seed deposition and rainfall (data from Pirassununga region obtained in ^{Tabanez et al., 2003}) and between seed rain and records on perches. The Kruskal-Wallis test was performed to verify whether there was a significant variation in seed deposition among treatments. And the Mann-Whitney U test was used to determine whether there was significant difference between the rates of visitation by birds on artificial perches. All analyzes were performed using the statistical package BioEstat 5.0 (^{Ayres et al., 2007}).

3.1 Seed deposition

In one year, a total of 23,911 seeds was recorded: 74 (0.3%) in the CA, 1,642 (6.9%) under the SAPs, 5,816 (24.3%) under the EAPs and 16,379 (68.5%) under the NPs. The difference in seed deposition between the models was highly significant (Kruskal-Wallis H = 35.65, p<0.0001). Seeds of 70 bird-dispersed species belonging to 32 families were collected, of which eight were not identified. The species richness recorded in each experimental model was as follows: five species (7.1%) in the CA, 21 species (30.0%) under the SAPs, 37 species (52.9%) under EAPs and 60 species (85.7%) under the NPs (Table 1).

The most representative families in number of species were: Myrtaceae, with six species, followed by Melastomataceae and Rubiaceae, with five species each (Table 1). The species that showed greater seed amounts were, in decrescent order: Miconia affinis (41.0%, n = 9,799), Callicarpa reevesii (17.3%, n = 4,134), Solanum sp. 3 (13.4%, n = 3,215), Myrsine coriacea (10.6%, n = 2,530) and Cecropia pachystachya (6.7%, n=1,597), totalizing 89% of recorded seeds. Most of the recorded species belongs to the group of early secondary (40.5%, n = 17), followed by pioneers (35.7%, n = 15) and, finally, by late secondary (16.7%, n = 7) (Table 1).

There was a positive correlation between the average of monthly rainfall and monthly seed deposition (Spearman Correlation test: r_s = 0.61, p = 0.047). The highest number of seeds was recorded between November and March, with dispersal peak in the second half of February (n = 5,304 seeds), while the lowest total amount was recorded in September (n = 184) as shown in Figure 2.

Figure 2 Number of bird-dispersed seeds monthly collected in each experimental model between June 2012 and May 2013. 

3.2 Bird visitation to perches

During 120 hours of focal observations, 21 bird species belonging to 14 families were recorded using artificial perches. There were 280 visits, of which 11.8% (n = 33) occurred on the SAPs and 88.2% (n = 247) on the EAPs. Visitation rates differed significantly between the treatments: 0.3 visits/h on the SAPs and 2.1 visits/h on the EAPs (Mann-Whitney U test: p = 0.006). Tyrannidae was the most representative family in number of visitors and visitations, with 7 species (33.3%) who used the perches in 60.0% of the visits (n = 168). The species Tyrannus melancholicus made the highest number of visits (28.2%, n = 79), followed by Pitangus sulphuratus (24.6%, n = 69), both from Tyrannidae family (Table 2).

There was temporal variation in the number of visits, with a positive correlation between rainfall and number of monthly visits (Spearman Correlation test: r_s = 0.76, p = 0.007). During the rainy season occurred 81.8% of the visits, with a peak in February (26.8% of the visits), while in the dry season occurred 18.9% of the visits (Figure 3).

Figure 3 Temporal variation in the number of bird visits to the artificial perches during a 12-month period. 

The following migratory species were recorded using the perches: Myiodynastes maculatus, T. melancholicus, Tyrannus savana and Empidonomus varius, all from Tyrannidae family (Table 2). These species (except T. savana, recorded out of focal sessions) correspond to 14.3% of the total and were responsible for 34.3% of the visits, restricted to the period from September to March, coinciding with the period of greater visitation to perches.

The most representative feeding guilds were the omnivorous (38.1%) and insectivores (38.1%), followed by granivorous (14.4%) and frugivorous (9.4%). Although Patagioenas picazuro and Forpus xanthopterygius are classified as frugivorous, some authors consider them as seed predators (^{Howe and Estabrook, 1977}; ^{Moermond and Denslow, 1985}). However, 42.9% of the species recorded include fruits in the diet at least during scarcity of its preferential food. These potentially disperser birds were responsible for 64.3% of the visits (Table 2).

4.1 Seed deposition

Perches increased the seed deposition in the old pasture as observed in other studies developed in temperate and tropical areas (^{McDonnell and Stiles, 1983}; ^{McClanahan and Wolfe, 1993}; ^{Wunderle, 1997}; ^{Holl, 1998}; ^{Shiels and Walker, 2003}; ^{Zanini and Ganade, 2005}; ^{Carlo and Yang, 2011}, ^{Cavallero et al., 2013}).

According to ^{Wunderle (1997)} and ^{Jordano et al. (2006)}, seed deposition can be significantly higher under perches than in nearby places where there are no perching sites, as in the control area of this study, because birds regurgitate and defecate more when perching than during flight. In the same way, perches with more crossbars or branches are more likely to be visited (^{Guevara and Laborde, 1993}; ^{Wunderle, 1997}; ^{Holl, 1998}; ^{Zanini and Ganade, 2005}).

Furthermore, the natural perches used in the PFSP offered additional resources for birds such as fruits, insects and shelter, which also seems to have contributed to the greater seed deposition in their traps. On artificial perches birds became more exposed, besides not having food readily available, which probably inhibited the visitation by some species sporadically observed on natural perches, as the tanagers Ramphocelus carbo (Pallas, 1764), Tangara sayaca (Linnaeus, 1766), Tangara cayana (Linnaeus, 1766) and Dacnis cayana (Linnaeus, 1766), among others. According to ^{Zahawi et al. (2012)}, the structural complexity of perches is an important factor for attracting bird-seed dispersers in degraded areas, since it is directly related to the availability of food resources and habitats.

Only five plant species were responsible for almost 90% of the seed deposition. These species are characterized by a generalist dispersal system, since they produce large amounts of small, colorful and juicy fruits, attracting a wide variety of opportunistic birds. Dispersal of these plants is highly efficient near forest edges and secondary vegetation, where they usually occur (^{Snow, 1981}; ^{Howe and Smallwood, 1982}; ^{Fleming et al., 1987}; ^{Zahawi et al., 2012}).

M. affinis, species with the highest rate of seed deposition, belongs to the family Melastomataceae, considered one of the most attractive to birds in the New World. These plants have quite small fruits with a high proportion of pulp, bearing a large number of small seeds (^{Snow, 1981}). It is one of the most abundant bird-dispersed species in the PFSP (pers. obs.).

The second species with the highest deposition rate was the exotic Callicarpa reevesii, originating from China (^{Potascheff et al., 2010}). Similar results were obtained by ^{Athiê and Dias (2012)} in a study of frugivory by birds in a ​​reforestation area in Rio Claro, São Paulo State. The authors found that the species with the highest visitation rates and consumption by birds was the exotic Melia azedarach L. (Meliaceae), which produced fruits during all dry season, as well as C. reevesii. For ^{Gosper et al. (2005)}, invasive alien plants usually bear fruit in times of food scarcity to reduce competition for dispersers with native species, thus facilitating its propagation and colonization of new sites. However, in the PFSP only two individuals of C. reevesii were observed, which does not corroborate the hypothesis that it acts as invasive species.

In all experimental models were predominant, in richness and number of seeds, pioneer and secondary initial species. These results reflect the regeneration stage of vegetation and the floristic composition of surroundings: vegetation nearest perches is mainly composed of early successional plants, common at forest edges and in areas in natural regeneration process, including the old pasture.

Equally important is the effect of distance between perches and surrounding remnants, the main seed sources of late secondary species to the old pasture. Most of bird-dispersed seeds is dropped in less than 100m away from the parent plant (^{Wunderle, 1997}), thus patterns of dispersal distance per animal (seed shadow) are typically a blend of high dispersal frequency near the maternal plant and a low occurrence of these events over long distances (^{Jordano et al., 2006}; ^{Jordano, 2014}). Knowing the minimum distances between the perches and the remnants of savanna and SF in PFSP are, respectively, 150 and 230m (see Material and Methods), these findings help to explain the low frequency of seeds from later stage species in the experiment.

At the same time, disperser birds visiting abandoned pasture and other disturbed areas, often explore few or no resources in forest interior, where those plant species are more frequent, even when these areas are surrounded by remaining forests (^{Holl, 1998}). Thus, forest species such as the Euterpe edulis palm, had low relative abundance in the seed traps. On the other hand, species from early ecological groups are essential for the natural regeneration process because they can act as “facilitators”, helping the establishment of later stages species in the recovery area (^{Avendaño-Yáñez et al., 2014}). Moreover, they are indispensible for attracting and/or maintaining pollinators and seed dispersers, because in general they produce large amounts of flowers and fruits attractive to animals (^{Tomazi et al., 2010}).

Regarding temporal patterns of dispersal, the seed deposition was higher in the rainy season. In the tropics, animal-seed dispersal is more common during wet season, relating to pattern of plant fruiting (^{Howe and Smallwood, 1982}; ^{Fleming et al., 1987}; ^{Wunderle, 1997}; ^{Zanini and Ganade, 2005}). The peak of seed deposition occurred between December and February, coinciding too with birds breeding season, when increases the exploitation of food resources (^{Pizo, 1997}; ^{Jordano, 2014}).

4.2 Bird visitation to perches

The species observed using artificial perches are typically generalists. This result was expected, since most of birds that frequent disturbed areas have opportunistic habits, with diet based on insects and fruits of pioneer and ruderal plants, common in forest edges and secondary environments (^{Guedes et al., 1997}; ^{Jordano et al., 2006}). Similar result was found by ^{Holl (1998)}, while evaluating seed deposition by birds under artificial perches in an abandoned pasture in Costa Rica. According to this author, who also developed the work in a field surrounded by native vegetation, forest bird species did not visit perches because they are not used to exploiting resources in open areas and usually find the necessary resources for their own survival inside forest.

On the other hand, regeneration process does not necessarily depend on specialized, large and medium-sized frugivores, often absent in areas to be restored. Bird-seed dispersers that can tolerate disturbed landscapes are the most important in the early stages of forest succession, since they are responsible for disseminating propagules that will begin this process (^{Guedes et al., 1997}; ^{Jordano et al., 2006}; ^{Francisco et al., 2007}).

^{McKey (1975)} and ^{Bocchese et al. (2008)} emphasize the importance of Tyrannidae, the most representative family in this study, for the forest succession: these birds feed both in forest edges and secondary and isolated plants. While feeding on insects in open areas, where they can find and capture them more easily, they carry seeds from pioneer species to degraded areas, accelerating the natural regeneration. In fact, many flycatchers were observed capturing preys in flight, while they were using the artificial perches of this experiment.

According to ^{Bocchese et al. (2008)}, perches, besides providing better airspace visibility for birds that capture insects in flight, such as flycatchers, improve the field of view of potential predators. These authors observed in an experiment with artificial perches, near a savanna remnant in Mato Grosso do Sul, that some birds used them for hunting and foraging. In the PFSP experiment, bones of small animals were found in seed traps under the SAPs, possibly leftover food of predatory birds such as owls, hawks or falcons. For ^{McDonnell (1986)}, perches higher than the surrounding vegetation, such as the elaborate perches of this experiment, are attractive to predatory birds because it act as an observatory for these species.

There was a seasonal pattern in bird visits to the perches, intensified during the rainy season. In this period, in addition to having the increase of resources exploitation due to greater availability of fruits and reproductive activity in birds, there is also the arrival of migratory birds, many of which participate actively of seed dispersal (^{Pizo, 1997}; ^{Alves et al., 2008}; ^{Zahawi et al., 2012}). ^{Zanini and Ganade (2005)} obtained similar results in a study about efficiency of perches in restoring Araucaria Forests in Brazil. According to ^{Jordano (2014)} the fruit ripening of some bird-dispersed plants is synchronized to the arrival of migrant birds.