Artificial perches for birds in deforested areas favour a seed rain similar to woodland remnants

The lack of seeds represents one of the highest difficulties to overcome for the ecological restoration of areas that have been deforested. This study evaluates the effectiveness of artificial perches in increasing the abundance and species richness of bird-dispersed seeds and the similarity of seed rain composition of deforested areas with and without artificial perches in relation to woodland remnants that serve as seeds source. We also tested for differences in seed abundance and species richness with different origins (native or non-native) as well as different type of habits (woody and non-woody). The experiment took place in two sites of the Espinal ecoregion, Argentina. We found that in deforested areas, perches increased seed abundance and species richness in the seed rain in comparison with deforested areas without artificial perches. The species composition under artificial perches was similar to the seed rain dispersed in the woodland. However, there was a decrease in the abundance and richness of native species under artificial perches, probably due to behavioral differences between opportunistic and obligate frugivorous. Seed of trees and shrubs species that can act as natural perches and nurses were well represented in the seed rain under artificial perches. We recommend using artificial perches in deforested areas with potential for recovery because it is an efficient technique to promote the entry of birds and increase seed rain, preserving features of the original environment. In places where native and non-native species show different fructification peaks, artificial perches could be used in certain periods of the year.


Introduction
The lack of seed supply is one of the main barriers to overcome for ecological restoration in areas that have been deforested (Shoo and Catterall 2013;Elgar et al. 2014;Karlin et al. 2020). In this sense, the role of many frugivorous birds species is essential in depositing seeds of the fleshy fruits consumed and in this way promote vegetation regeneration when they move towards the forest edges or degraded patches (Green and Dennis 2007). This contribution can be very significant in temperate regions where other groups of seed-dispersing animals (i.e., reptiles and medium and large mammals) are not present or show marked decreases in their richness and abundances (Karubian et al. 2012). However, frugivorous birds may find little incentive to move away or too far from the forest crossing open, deforested, or degraded areas, where fruit abundance is low and the chances of being caught by their predators are high (Mastrangelo 2014;Da Silveira et al. 2016). For this reason, deforested areas often receive negligible seed rain, even if they are surrounded by natural vegetation.
To overcome the resistance of birds entering deforested areas and dispersing seeds, some ecological restoration projects have implemented artificial perches for birds (Holl 1998;Zanini and Ganade 2005;Guidetti et al. 2016). Artificial perches increase structural complexity and stimulate birds to use open areas and stay there for longer times (Donald and Evans 2006;Pejchar et al. 2008). In addition, since seed deposition by defecation and regurgitation occurs more often when birds perch or immediately after they undertake flight (McDonnell and Stiles 1983), seeds tend to concentrate under artificial perches in the same way it occurs under isolated trees, live fences, or other remaining vegetation that has survived disturbances and it is used by birds as perch or rest points (Schlawin and Zahawi 2008;Pizo and dos Santos 2010;Cottee-Jones et al. 2015). The natural seed rain promoted by perches facilitates the arrival and establishment of small clumps of species in the open area (Guevara et al. 1992;Debussche and Isenniann 1994;Schleuning et al. 2011b). As these nuclei of species grow and eventually occupy the empty spaces merging with each other, the contrast between logged areas and forest diminish (Reis et al. 2010). In this way, artificial perches constitute a technique based in the concept of nucleation (Yarranton and Morrison 1974) that pretend to induce a process of secondary succession, as similar as possible to the mechanisms used by nature, towards a state of equilibrium (Schlawin and Zahawi 2008;Reis et al. 2010).
Considering that landscape transformation due to agricultural activities affects more than 40% of the planet's land area and is the main cause of biodiversity and ecosystem services decline (Foley et al. 2011), ecological restoration measures should be prioritized in areas that have potential for recovery. Deforested areas that do not receive active intervention can remain in intermediate successional states experiencing a very slow reversal towards the original ecosystem (Jones and Schmitz 2009). Under some circumstances, these areas can end up taking the form of a different ecosystem (like a grassland or shrubland) with a new species composition (Hobbs et al. 2006). The species best represented in deforested areas are mostly pioneer herbs and grasses dispersed by wind (Cubiña and Aide 2001;Zwiener et al. 2014). Late succession species rarely reach open areas or take too long to do so (Corlett and Hau 2000;Martínez-Garza and Howe 2003). Since seeds of tree or shrub species are rapidly predated, tend to immediately germinate or die after a certain latency time (Hardwick et al. 2004), it becomes unlikely that the pre-existing soil seed bank will contribute significantly to the regeneration of woodland vegetation (Nepstad et al. 1996;Ferri et al. 2009;Sione et al. 2016). On the other hand, active restoration actions such as direct seed planting in bare soil or reforestation plans, frequently fail in proper species selection (Griscom and Ashton 2011;Corbin and Holl 2012) or are performed with only one or a few species given the high costs involved (WWF International 2005;Corbin and Holl 2012). As an alternative, the introduction of artificial perches can increase the permeability of the deforested areas and favor the arrival of species by fostering the opening of a series of more or less stochastic events (Reis et al. 2010). Through an ecological flow in which nearby woodland remnants act as seeds sources, deforested areas with artificial perches for birds can recover species of the original environment, including both non-woody and woody plants (Corbin and Holl 2012;Rocha-Santos et al. 2017). The arrival of woody plants that can subsequently act as natural perches and nurses, facilitating the arrival and survival of more late-successional species, seems to be essential for the development of persistent woody plant communities (Barberis et al. 2002;Cabral et al. 2003;Tálamo et al. 2015;Keller et al. 2016).
Unfortunately, areas passively or actively recovering from disturbance may be also more susceptible than mature ecosystems to invasion by non-native species present in the landscape (Huebner and Tobin 2006;McCay et al. 2009;Cabido et al. 2018). This can be considered as one disadvantage of the perches. Given that woody non-native plants with fleshy fruits dispersed by avian frugivorous are usually more successful in increasing their distribution range than other non-native and native plants, their dispersal under artificial perches is a point that must be considered (Ponce et al. 2012).
This study evaluates the effectiveness of artificial perches for birds in deforested areas of the center of the Province of Entre Ríos (Espinal ecoregion), Argentina. Abandoned patches where agricultural activities have ceased, can be available for the recovery of the native woodland. Although artificial perches have generally shown high efficiency in promoting seed dispersal, this work aims to investigate the composition of the seed rain on artificial perches, recognizing differences concerning seed dispersal in forest remnants and deforested areas. Furthermore, it is of particular interest to study the effect that artificial perches have on the abundance and species richness of seeds of native and non-native plants dispersed, as well as on the arrival of woody plants that can subsequently act as natural perches and nurses. The questions to be answered are: 1-do artificial perches increase the abundance and species richness of the seed rain dispersed by birds?; 2-does the origin (native or non-native) of the seed composition differ between the woodland remnants and under the artificial perches in deforested areas along the year?; and 3-do artificial perches facilitate the arrival of woody species (trees and shrubs as opposed to herbs and vines)? Given that birds consume fleshy fruits mostly in the woodland, and perches are known to act as a focus of attraction for birds, the installation of perches in deforested areas facilitates the arrival of plants species. We predict that seed rain (abundance, species richness, origin and habit type) will be similar between the woodland and the artificial perches. Additionally, we predict a seed rain concentrated under the artificial perches in deforested areas, with greater abundance and species richness than open areas without perches.

Study area
The experiment was conducted in the subtropical woodland of the Espinal ecoregion. We selected two sites in the Paraná Department, in the center of Entre Ríos Province, Argentina, to perform the experiments. The two sites were more than 15 km apart. The first site (PSM) was located in the "Parque General San Martín" (31°43′30′′ S, 60°20′00′′ W), a natural protected area with a size of 400 ha. The second site (REA) was located in the "Reserva de Uso Múltiple Escuela J.B. Alberdi" (31°50′12′′ S, 60°31′25′′ W), with a size of 20 ha (Online Resource 1- Fig. S1). The climate is warm and humid, with an average annual temperature of 20 °C and annual rainfall exceeding 1200 mm. The precipitation is concentrated from the austral spring through summer to early fall (from October to April); however, there is usually a water deficit in the soil (due to high temperatures) during this period of the year (Matteucci 2012). The woodlands of the region are low and xerophytic and vary from dense to open. There can be distinguished two tree strata (a low and continuous one, 6 to 10 m high, and another discontinuous with isolated individuals exceptionally exceeding 12 m in height), a relatively poor shrub stratum, and a relatively well developed herbaceous stratum (Matteucci 2012). The dominant species in the upper tree stratum are Prosopis nigra and, to a lesser extent, Aspidosperma quebrachoblanco. In the lower tree stratum, there are Prosopis affinis, Vachellia caven, Celtis ehrenbergiana, Geoffroea decorticans and Scutia buxifolia. Other tree species that accompany in more closed forests or close to water courses are Myrcianthes cisplatensis, Schinus longifolia, Schinus molle, Sapium haematospermum, Zanthoxylum fagara, Jodina rhombifolia, Achatocarpus praecox and Phytolacca dioica (Lewis and Collantes, 1973;Cabrera, 1976;Matteucci, 2012). The shrub stratum is composed mainly of Vachellia astringens, Senegalia bonariensis, Senegalia praecox, Castela tweedii, Ephedra tweediana and Berberis ruscifolia. The forest also presents a high diversity of lianas, vines and epiphytes, such as Passiflora caerulea and Smilax campestris, among others.
Historically these woodlands have experienced different alterations, being subjected to logging, selective extraction of dominant species for poles, wood, firewood, and charcoal, as well as overgrazing by livestock and the replacement of native trees with fast-growing exotic species plantations (Johnson and Zuleta 2013). Moreover, in recent decades, intensive agriculture has advanced on the remaining woodland, producing woodland loss and fragmentation (Arturi 2006;Matteucci 2012). In the case of the selected sites, both present a mixture of native woodland remnants and deforested areas that have undergone diverse uses in the past (extraction of dominant species, extensive cattle ranching, small-scale agriculture, and firewalls). Nowadays, productive activities are no longer allowed, since they are currently inside protected areas. Besides, the deforested areas in both sites have well conserved soil (covered by native grass and herbaceous species, not bare, compacted, under limited drainage) and are close to woodland remnants that can act as sources of propagules. All this makes them deforested areas with potential for recovery in the future.
The seed dispersal interaction network in the woodland of the study area is composed of at least 27 species of birds that can disperse fleshy fruit seeds (Guidetti, 2020). Taking into account their relative abundance in the bird assemblage, the amount of seeds they disperse and the richness of plant species they consume, the main frugivorous species are Saltator aurantiirostris, Elaenia parvirostris, Thraupis sayaca, Turdus amaurochalinus and Turdus rufiventris (Guidetti, 2020).

Experimental design
We set up 24 sampling blocks: 14 blocks set on the PSM between December 2014 and May 2016, and another 10 blocks set in the REA between May 2015 and May 2016. Each block consisted of a native woodland area and an adjacent deforested area. We set seed traps consisting of 0.5 × 0.5 m "mosquito net" plastic fabric (mesh size of 1.5 mm), raised from the ground 0.25 m with galvanized wire structures to prevent predation of the seeds by rodents (García et al. 2010). In each sampling block, two traps separated from each other by two meters were placed within the native woodland area ("woodland" treatment, n = 48), and two traps in the deforested area. Within the woodland area, seed traps were placed under the canopy of trees selected at random. Within the deforested area, one of the traps was placed under an artificial perch built with a 2 m-high bamboo pole and two wooden rods one meter long and 5 mm in diameter, crossed together and tied to the top end of the pole ("perch" treatment, n = 24). The other trap was placed in the open area ("open" treatment, n = 24), two meters away from the axis of the pole of the artificial perch. Artificial perches were located at a distance between 10 and 40 m from the edge of the native woodland area and separated by at least 25 m away from the artificial perches of the contiguous sampling blocks (Fig. 1). In each block, the traps designated "open" and the traps designated "perch" were always located at the same distance away from the woodland edge.
We collected seeds in the traps every 21 days. Whole fruits were discarded, and only seeds that had been dispersed by birds were analyzed. Since there are no frugivorous bats confirmed in the study area (Barquez and Díaz 2020), the possibility of confusing bird feces with bat feces was ruled out. The samples were dried on a stove for 60 min at 80 °C to remove moisture and stored in paper envelopes. Seeds found in the traps were identified at the species level under a stereomicroscope (Nikon SMZ645, with magnification up to 50 X). For this purpose, a reference collection was made with seeds of fleshy fruits species collected within the study area. We additionally reviewed specimens of the Instituto de Botánica Darwinion herbarium and consulted bibliographic material (Alzugaray and Carnevale 2009;Abraham de Noir and Bravo 2014). For each species of seed, we recorded the origin (native and non-native) and habit (tree, shrub, vine, or herb). Further, two types of habit were considered: woody type included trees and shrubs with erect stems (species that can act as natural perches), while non-woody type included vines and herbs. Seeds that could not be identified were considered as morphospecies only if seemed to come from fruits dispersed by birds (showing no signs of adaptation to anemochory or epizoochory).

Data analysis
To assess the effect of artificial perches on the abundance and species richness of seed rain reaching deforested areas (considering differences between arrival of native and non-native, woody and nonwoody species), generalized linear mixed models (GLMMs) were used. Both, the abundance model and the species richness model, included negative binomial distribution and log link function (Crawley 2007;Zuur et al. 2009;Logan 2010). Treatment (woodland, perch, and open), origin of plants (native or non-native), type of habit (woody and non-woody), date (as numeric value by month), and all the interactions were included in both models as fixed factors. Sampling blocks (n = 10 and 14) nested within the sites (PSM and REA) were included as random factors in both models. The GLMMs were performed using the "glmmTMB" package (Fournier et al. 2012;Skaug et al. 2013;Magnusson et al. 2016) in R (R Development Core Team 2018). All models were evaluated according to the BIC (Burnham and Anderson 2002). Model assumptions were evaluated using the "DHARMa" package (Hartig, 2021). ANOVA (type III Wald chi-square tests) were performed using the "car" package (Fox and Weisberg, 2019). Conditional and marginal coefficients of determination for GLMMs (R-GLMM 2 ) were calculated using the "MuMIn" package. Post-hoc differences were analyzed with Tukey tests using the "emmeans" package (Russell, 2021). Additionally, we analyzed the temporal variation in seed abundance of native and nonnative seeds under artificial perches using the lineal model in R. We carefully structured mixed models analysis to control for any possible problems with pseudoreplication (Arnqvist 2020).
Finally, we compared seed composition in the different treatments (woodland, perch, and open) analyzing the relative abundance of each species per sample. We conducted an ordination with Euclidean dissimilarity values calculated by a non-metric multidimensional scaling (NMDS) and a permutational multivariate analysis of variance, using the metaMDS and adonis functions of the "vegan" package (Oksanen et al. 2017) in R.

Results
The model that best explained the variation in seed abundance included treatment (woodland, perch, and open) and the interaction between type of habit (woody and non-woody), origin (native and non-native) and date (as numeric value by month) as explanatory variables (Table 1) (Online Resource 1-Table S1) (conditional R 2 = 0.99). The abundance of seeds that reached artificial perches in the deforested area was higher compared to open area treatment without perches (where the seed abundances decreased significantly), but slightly lower in comparison with the woodland treatment (Fig. 2, Online Resource 1- Table S2). Results of the post-hoc test also indicate that native plants were more abundant than non-native over the year in all the treatments (ratio = 1.53, SE = 0.172, df = 4435, t ratio = 3.797, p value = 0.0021, Online Resource 1- Fig. S2). These difference in abundance according to treatment and plant origin is significant in the case of non-woody species (ratio = 1.85, SE = 0.337, df = 4435, t ratio = 3.388, p value = 0.0093), but not for woody species (ratio = 1.26, SE = 0.166, df = 4435, t ratio = 1.793, p value = 0.4701). Results of the post-hoc test also indicated that the seed abundance of non-woody species is significantly lower than for woody species over the year for all the treatments (ratio = 0.39, SE = 0.043, df = 4435, t ratio = − 8.437, p value < 0.0001, Online Resource 1- Fig. S3). Particularly, in the case of artificial perches (but similar to what happens in the woodland), seeds of non-woody plants show a similar abundance patron over the year (Online Resource 1- Fig. S4). Instead, seeds of woody native plants showed a constant linear pattern along the year, although have their maximum peak from January to June, while seeds of woody non-native plants increase their abundance along the year with a maximum peak from October to December (Fig. 3). The relative abundances of plant species within the different treatments, origins and type of habits are included in Online Resource 1, Tables S3-5. The model that best explained the variation in species richness in seed rain included treatment (woodland, perch, and open) and the interaction between type of habit (woody and non-woody), origin of plants (native or non-native) and date (as numeric value by month) as explanatory variables (Table 2) (Online Resource 1- Table S6) (conditional R 2 = 0.61). The species richness in the seed rain was similar in the perches and woodland treatment, while in the open area treatment was lowest (Fig. 4, Online Resource 1- Table S7). In the woodland, there were a total of 23 native species of seeds while only five were non-native species. In the deforested area under artificial perches, there were 13 native species and six non-native species. In the open area treatment seeds from only two species were identified, one was native and the other non-native. Non-native species richness was lower than native species richness  S5). This trend in the species richness of seeds rain according to treatments and plant origin is significant for non-woody (ratio = 1.87, SE = 0.333, df = 4435, t ratio = 3.511, p = 0.0060), but not for woody species (ratio = 1.39, SE = 0.170, df = 4435, t ratio = 2.732, p = 0.0692). Concerning type of habit, the post-hoc tests indicates that the species richness of woody plants is significantly higher in comparison with nonwoody in all the treatments (ratio = 0.40, SE = 0.043, df = 4435, t ratio = − 8.516, p value < 0.0001, Online Resource 1 - Fig. S6).
Seed rain in the woodland and the artificial perches were similar in terms of their composition (Table 3), as it is shown in the ordering of NMDS points (Fig. 5). A very small percentage of variation in the composition in seed rain was explained by the different treatments (R 2 = 0.03, P < 0.001, Online Resource 1- Table S8).

Discussion
Artificial perches produced a significant increase in the abundance in the seed rain that reach deforested areas surrounding the Espinal woodland remnants. The abundance of seeds found under artificial perches indicates that these structures act as a focus of attraction for birds and promote their arrival, facilitating the entry of plant propagules and giving rise to seed nucleation points in areas that have lost woody cover (Zwiener et al. 2014). These results agree with those obtained by most restoration programs that have used artificial perches around the world (Guidetti et al. 2016). In the few studies that did not find a significant increase in seeds under artificial perches, the duration of sampling may have been insufficient, considering that birds require a certain period of adaptation to start using these structures (Shiels and Walker 2003). Predation of dispersed seeds by other animals was also considered another possible cause (Holl 1998). Increasing abundance in seed rain in deforested areas is crucial, especially given that having a nearby source of propagules does not always imply a bigger seed rain and, in addition to that, a large proportion of the seeds arriving in the area do not survive until germination (Graham and Page 2012;Reid and Holl 2013). Therefore, the use of artificial perches then results in higher plant densities than projects based on passive restorations (Tres and Reis 2009;Schorn et al. 2010).
The installation of artificial perches also produced a significant increase in species richness compared to  seed rain in deforested areas without perches. Perches give temporal and spatial continuity to the diverse interactions between frugivorous birds and plants in the forest, enabling, for example, the dispersal of rare species seeds by anti-apostatic selection, an important mechanism that structures the diversity of forests in recovery (Karubian et al. 2012;Carlo and Morales 2016). Thus, the seed rain under perches seems to reflect better the reproductive potential of endozochoric plants (Guimarães et al. 2008). Nevertheless, the process of seed dispersal from the woodland to the open area involves not only fruit consumption but also the movement of the birds. In this sense, there are some forest visitors or habitat generalists species more adapted to the conditions in deforested areas, but forest specialist species may refuse to cross the edge of the woodland, dispersing seeds mostly in the forest interior or flying straight ahead directly to another forest remnant (Bennun et al. 1996;García et al. 2010;Pizo and dos Santos 2010;Carlo et al. 2013). Generally, frugivorous bird species with nonspecialized diets (species that consume fruits based on their availability in the environment) are the ones that are habitat generalists, making the most important seed dispersal among different environments (Purificação et al. 2014). This characteristic of the frugivorous bird assemblage can be the reason for the difference of the seed species richness observed in the artificial perches in comparison to the woodland treatment. It is expected that as the development of secondary forest patches progresses in the deforested areas, more forest specialists and obligate frugivorous species will begin to move within the patches, increasing much more the species richness in the seed Seed composition under artificial perches was similar to that of the woodland. Since seed flow is established from the neighboring woodland remnants, perches can promote vegetation similar to the original ecosystem (Schorn et al. 2010). Considering that the identity of the seed rain species has decisive effects on the natural succession of the sites, the control in the seeds' dispersal of non-native species could be one of the factors that seriously condition the use of artificial perches. It is quite common for invasive species to produce large amounts of fruits (Gleditsch and Carlo 2011). Here, the dispersal of non-native species was high in all treatments, even within the woodland. The noticeable difference between native and nonnative species abundance was strongly influenced by the quantity of Morus alba (Common Mulberry; Moraceae) dispersed seeds. This species has a large number of seeds per fruit (56.8 ± 15.7) and a large number of fruits per adult tree (1000-10,000, SD, pers. obs.), ripen at the studied area from October to early December (SD, pers. obs.). Again, the reason for the difference observed between the species richness in the artificial perches and the woodland treatment could be the movement of occasional or opportunistic frugivorous being attracted to the woodland when the amount of exotic fruits increases abruptly or tracking the presence of others across the foraging landscape (Carlo et al. 2007;Purificação et al. 2014). Instead, obligate frugivorous species get involved in much more interactions to sustain their basic metabolic requirements and have more selective foraging behaviors (Schleuning et al. 2011a;Sebastian-Gonzalez 2017;Bomfim et al. 2018), but do not leave the forest as frequently (Purificação et al. 2014), increasing the seed rain species richness of native plants inside the woodland. However, thinking about the possibilities of ecological restoration, it is interesting  to state that several non-native species differ on the fruiting season with native (Gosper 2004;White and Vivian-Smith 2011). In the study area, the fruiting peak for non-native fleshy fruit species was detected from October to December, while for the native species the peak is between January and February and another occurs between May and June (Scarpa 2013). So, installation of artificial perches could be carried out if species with fleshy fruits in nearby woodland remnants are well recognized, avoiding sites where non-native species are present or placing them only at the time of the year when the invasive species have fewer fruits (Prather et al. 2017). A large proportion of the seeds that reached the site favored by the artificial perches in the experimental area were woody plant species, as it has also been found by Prather et al. (2017) in Houston, USA. In this sense, artificial perches act in a similar way to isolated trees and remnant shrubs that persist to disturbance, accelerating the colonization of woody species below and around their canopy (Slocum 2001;Schlawin and Zahawi 2008). Trees or erect shrubs established in the nuclei come to play a dominant role, acting themselves as natural perches for birds and improving conditions in the patches (providing shade, food, and nesting sites). Regeneration in deforested areas appears to be irregular and dependent on the formation of vegetation groups or nuclei that facilitate the recruitment of seeds from the forest, the establishment of new seedlings, and the expansion of woody plant cover (Barberis et al. 2002;Cabral et al. 2003;Carlo and Morales 2016). The influence of these nuclei continues even in later successional stages, showing a higher density of trees, basal area, and species richness in their vicinity (Slocum 2001;Schlawin and Zahawi 2008).
The experience states that artificial perches for birds are easy to implement, require little labor for manufacture and installation, and have substantially low costs, taking into account that they can be manufactured from materials recycled or found in the site (Holl 1998;Reis et al. 2010;Graham and Page 2012).
Here only the simplest model with bamboo pole has been tested due to its availability and the possibility of replication in the study. In the future, it would be interesting to analyze what happens with more complex perches, branches of dead native trees or with a grouped arrangement of perches, since large nuclei are proven to be more attractive to birds (Holl 1998;Toh et al. 1999;Corbin and Holl 2012). Perches seem to be a possible alternative when there are remnants of woodland in the proximity of the open area. It would also be interesting to study the optimal distances relative to the edge of the remnant for which perches are most effective in the Espinal (Pizo and dos Santos 2010). Nevertheless, while a big challenge of restoration in deforested areas is to attract birds to deposit seeds, there can be other barriers to plant regeneration and the validity of the predictions depends a lot on the success or mortality of these seeds during initial recruitment stages (Holl et al. 2000;Reid and Holl 2013). For that reason, we considered that artificial perches may be much more successful in ecosystems that are sufficiently resilient and just need minimal intervention. Subtropical and temperate environments with disturbance matrices similar to the ones considered here have shown a significant increase in seedlings established under artificial perches (Guidetti et al. 2016). The soil and environmental conditions (low elevation, warm temperatures, and abundant rainfall) during and immediately after the period of highest native seeds dispersal at the experimental sites are expected to favor germination and seedling growth (Holl 2013). Furthermore, apart from the most common herbaceous plants (e.g., Brassica rapa, Solanum sisymbriifolium and Trifolium repens), many woody species identified under the perches are also described as shade intolerant, pioneer or early successional woody species (e.g., Achatocarpus praecox, Celtis ehrenbergiana, Schinus longifolia, Lantana montenvidensis and Sapium haematospermum) that germinate and can grow rapidly in disturbed areas that are not occupied by woody vegetation (Barberis et al. 2002;González and Cadenazzi 2015;Keller et al. 2016;Peirone-Cappri et al. 2020). Also, both sites were located in protected areas, making the commitment of stopped disturbances and ensuring that care can be prolonged in time. In other contexts, additional measures and multiple approaches (such as livestock exclusion, soil transposition or nurse plants) may be necessary (Bevilacqua Marcuzzo et al. 2013). Finally, longer studies may be required to properly monitor the seedlings establishment and the functional nuclei development, achieving a better recognition of the real potential that perches have for forest regeneration in open areas.
In conclusion, artificial perches function as a nucleation technique to increase seeds dispersal by birds in deforested areas around Espinal woodland remnants. Artificial perches increased the abundance and species richness in the seed rain, with a species composition similar to that of the woodland. Using artificial perches should be avoided in sites with potentially invasive non-native species. However, in places where native and non-native species show different fructification peaks, artificial perches could be used in certain periods of the year, avoiding the dispersal of undesirable seeds of invasive species. In addition, artificial perches principally facilitate the arrival of woody species that then also serve as natural perches; this can progressively improve the conditions for secondary succession in the deforested area, having a positive impact even in later stages.