Spatial variation of seed rain in deciduous tropical forest

Seed rain is an ecological process and its functional attributes are essential for maintaining the dynamics of natural regeneration. The objective of this research was to evaluate the spatial variation of the seed rain in a toposequence of a Seasonal Deciduous Forest defined by three elevations: (Base 512 m; Slope: 534 m and Top: 559 m). 15 collectors of 1 m2 were installed at each elevation level. Data were collected monthly from September / 2017 to February / 2019. The seeds were classified according to the dispersion syndrome, habit and size. NonMetric Multidimensional Scaling (NMDS) was used to verify the variation in species composition and distribution. We identified 20,217 propagules, belonging to 65 species and 30 families, in addition to 4 morphospecies, which represents 449 seeds / m2. The families with the highest species richness were Fabaceae, Sapindaceae and Euphorbiaceae. Of the 65 species sampled, 71% were arboreal. Zoochoric species predominated (78%) and very small seeds corresponded to 53% of the sample. We demonstrate that, on a small spatial scale, the relief represents an important source of heterogeneity in the vegetation component, since the Research, Society and Development, v. 9, n. 9, e879997956, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i9.7956 2 topographic gradient influenced the composition and distribution of the functional attributes of the seed rain.

in the Chaco (Oliveira-Filho et al., 2006). In the South, it occurs between the Mixed Rainforest (Araucaria Forest) and the Campos Sulinos (Pampas), in an area of subtropical climate (IBGE, 2012). It is estimated that the SDFs occupies 4% to 6% of the Brazilian territory (Espírito-Santo et al., 2008). The Bodoquena plateau region, in Mato Grosso do Sul is one of the last and largest remnants of this phytophysiognomy in Brazil that is still relatively well-preserved (Pott & Pott, 2003;Salzo & Matos, 2013). Mato Grosso do Sul also has SDFs in some stretches of floodplains by the Paraguay River, in its alluvial formation (IBGE, 2012).
The SDFs are conditioned to regions with climatic seasonality (Pennington et al., 2006), occurring as forest disjunctions among other forest and/or savanna types, with arboreal strata formed by predominantly deciduous species (Veloso et al., 1991). They are called Dry Forest or Limestone Forest, in reference to the local climatic conditions or lithology. The variations found in the structure of dry tropical forest fragments can be attributed to the characteristics of the substrate, position, landform, location in relation to other savannahs, fire history, and land use. The influence of environmental factors, such as altitude, topography and soil types, on the composition and dynamics of forests is well known for tropical rainforests (Charles-Dominique, 1995;Franklin et al., 2012;Homeier et al., 2010;Xia et al., 2016). However, there is still a lack of understanding on the association of the Seasonal Deciduous Forests structure and dynamics with some abiotic components, such as topography for instance (Arruda et al., 2013). Studies on the SDFs in Brazil focus mainly on structural aspects of the vegetation, which include several quantitative inventories and physiognomicfloristic descriptions (Felfili et al., 2007;Ivanauskas & Rodrigues, 2000;Nascimento et al., 2004;Pedralli, 1997;Silva & Scariot, 2003;Scariot et al., 2005).
Seed rain is an important tool that allows information about species composition, availability and seed diversity (Piña- Rodrigues & Aoki, 2014). Since dispersion is characterized as a dynamic and fundamental transport for the reproduction of plant species, as it provides a favorable location for seed germination, away from the mother plant (Howe, 1993;Gomes, 2018). Seed production and dispersion are important functional attriutes for the maintenance of plant populations, influencing the spatial distribution of species and the composition of the plant community, in addition to affecting gene flow within and between populations and enabling the colonization of new sites and habitat restoration (Bacles et al., 2006;Howe & Smallwood, 1982;Kroiss & Hillers-Lambers, 2015;Mateo et al., 2016;McConkey et al., 2012;Merritt et al., 2010;Reid et al., 2015;Vellend, 2010).
The goal of this research was to evaluate the spatial variation of the seed rain in a toposequence of a SDF, to answer the following questions: is there variation in the seed rain in relation to the position occupied in this toposequence? What environmental factors influence this variation, if any? Are the dispersion syndromes and the propagules size related to the patterns detected?

Statistical analysis
The sample ordering was obtained by the Non-Metric Multidimensional Scaling (NMDS), based on the Bray-Curtis dissimilarity matrix using the Vegan package (Oksanen et al., 2018) of the statistical software R, version 3.5.1 (R Core Team, 2018). From species relative abundance, dispersion syndrome, and seed size we obtained the variation in species composition across environmental gradients. We used the Multivariate Analysis of Variance -MANOVA, according to Friedrich et al., (2016).

Results
65 species were sampled over 18 months, belonging to 30 families, in addition to four morphospecies. The most representative families were Fabaceae, Euphorbiaceae and Sapindaceae, while the species with the highest number of seeds were Myrsine umbellata (6.489 seeds), Terminalia mameluco (2.381 seeds), Erythroxylum sp. (1.884 seeds) and Cybianthus detergens (1.648 seeds), which represents 38% of the total We registered 20.217 seeds (449 seeds / m²), with 13.820 seeds at the base (68%), 3.898 on the slope (19%) and 2.499 at the top (13%) ( Table 1). The greatest number of seeds was found at the end of the dry season and beginning of the rainy season, with the period of highest production between November and February, with peaks in November and December and the lowest production in April and July. The seed rain was seasonal, varying among the elevation levels. Figure   Subtitle: Circle sizes represent the collection months from 1 to 18; the colors represent the gradient: black -base, light gray -top and dark gray -slope. The contribution of each species is given by the position in the graph: the further away from the "0.0" coordinates, the greater the contribution of the species to differentiate the samples. Source: Research authors (2019). Figure 4 shows the NMDS ranking in relation to the variation in seed size (stress = 0.13, with 98% of the total variance being recovered in the Bray-Curtis distance matrix); altimetric quotas and months of collection explained the variation in seed size distribution (Pillai = Locations 0.03296; gl 1 and 365; p 0.002207; Months = 0.32819; gl 17 and 732; p = 8,165e-14), reinforcing the differences between the sampled areas and between the months of fieldwork. Development, v. 9, n. 9, e879997956, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i9.7956  Very small seeds showed greater variation in abundance between the elevation levels, with 83% at the base, 14% on the slope, and only 3% at the top. At the base, the majority of seeds were very small, which corresponded to 53% of the total (10.637 seeds), represented by 14 species. Small seeds represented 16% (3.312 seeds and 18 species), medium seeds 11% (2.163 seeds and 19 species), large seeds 7% (1.461 seeds and 9 species), and very large seeds 13% (2.547 seeds and 5 species) (see Table 1).

Discussion
The number of species recorded in the seed rain was similar to that of Sccoti et al., (2011) in an SDF area in Rio Grande do Sul, while Battilani (2010), in the Planalto da Bodoquena registered 117 species in three years of sampling. Our results differed a bit from some floristic studies on the shrub-tree component carried out in that same region, however, they were carried out in areas with riverside forests (Baptista-Maria et al., 2009) or in ecotonous areas of forest formations (Zavala et al., 2017), that has greater richness than the SDFs. Nevertheless, the large number of propagules registered in this research shows the potential that the remaining SDFs have to support the restoration of degraded areas, aiming to maintain biodiversity, considering the different reproductive strategies of the species that Development, v. 9, n. 9, e879997956, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i9.7956 14 compose this type of vegetation. In Deciduous Seasonal Forest, natural regeneration develops satisfactorily, presenting floristic composition and high diversity, with a clear increase in the dominance and density of some key species (Scipioni et al., 2009).
Fabaceae, which stood out as the most representative families in species richness, is considered a typical family of the Brazilian forests (Oliveira-Filho & Fontes, 2000;Oliveira et al., 2016), with great importance for environmental restoration due to the ability to fix nitrogen that several species have and, consequently, improvement of soil conditions. Therefore, its species are considered facilitators of ecological succession (Canosa et al., 2012). Sapindaceae is a very diverse family, especially in the North -Amazon Forest -and Southeast -Atlantic Forest regions of Brazil, with emphasis on the species of Serjania, the richest genus in Brazil (Somner et al., 2010). Composed almost exclusively by lianas, species of this genus play a role in environmental dynamics, forming true "biological corridors" in the forest canopy (Aschoff, 2012). Euphorbiaceae, which also stood out in this study, has fast growth as one of its main characteristics, participating in the initial stages of succession and contributing to the rapid plant densification in areas undergoing restoration (Amaral et al., 2013).
Intra and interspecific variations in seed production and dispersion syndromes influenced the spatial distribution of seeds during the fieldwork period. Such variations may be related to the species phenological and reproductive characteristics, as pointed out in other studies of seed rain for both individual and community levels (Araújo et al., 2004;Armesto et al., 2001;Au et al., 2006;Hampe et al., 2004;Masaki et al., 2007;Shen et al., 2007).
Variations in the composition of seed rain among collectors and between altimetric levels were influenced by the occurrence of exclusive species of specific habitats and by variations of seed abundance through space. The species composition in the seed rain, in general, is related to the local plant community (Penhalber & Vani, 1997) but tends to be different between areas (Au et al., 2006;Pivello et al., 2006;Rother et al., 2009), between habitats (Martini & Santos, 2007), under trees with different forms of dispersion (Clark et al., 2004) and among seed collectors in the same stretch of the forest (Hardesty & Parker, 2002).
Such variations are related to several factors, such as environmental heterogeneity, floristic composition, phenological patterns of the local community, dispersal agent activities, and topographic gradients. The topographic gradient is an environmental variable that indirectly influences a series of other environmental factors closely related to the patterns of plant distribution in the Atlantic Forest Meireles et al., 2008;Oliveira-Filho & Fontes, 2000;Oliveira et al., 2013;Pedroni et al., 2013). The results of this study confirm that for the SDFs, the topographic gradient is also important in determining functional patterns and affects the seed rain composition. The differences in richness and abundance of the seed rain between the elevation levels can be a consequence of several factors, such as soil conditions, topography and microclimate characteristics, that change along the elevational quotas (Ferreira-Júnior et al., 2007;Ferreira-Júnior et al., 2012;Jones et al., 2011;Klein, 1980Klein, , 1984Marangon et al., 2013;Veloso & Klein, 1959). Wolf et al., (2012) observed that greater species richness is expected where topographic variations on a local scale result in greater availability of water in the soil.
Some studies carried out in wetland and riverside areas have shown that some species are indicative of each portion of the slope (Borghi et al., 2004;Moro et al., 2001). Considering that several other studies report the influence of topography on the organization of plant species communities (Cardoso & Schiavini, 2002;Higuchi et al., 2012;Higuchi et al., 2013;Rodrigues et al., 2007), the occurrence of indicator species on the slope and top can be partially explained by the ecological requirements of these species, associated with distinct ecological niche. Along the altitudinal gradient studied, environmental variations were observed, which are associated with soil type, slope, humidity, and microclimate, as also registered by Oliveira-Filho et al., (1998), Thomas & Winner (2002) and Homeier et al., (2010), Webb & Peart (1999).
The large amount of small seeds recorded at the base of the slope is related to the presence of species associated with more humid areas, such as Maclura tinctoria, Myrsine umbellata, and Cissus erosa, that are typical of these environments. Generally, species that produce smaller seeds, under favorable environmental conditions, have a higher growth rate compared to larger-seed species. These species may also have the dispersal process optimized since the small size increases the number of seeds ingested by the dispersers (Graham et al., 1995;Larson et al., 2015;Moles & Westoby, 2006;Muller-Landau, 2010;Pereira et al., 2013). On the other hand, large seeds have a greater amount of stored resources, and thus are more likely to germinate (Pesendorfer et al., 2016), and also has a better ability to cope with adverse environmental conditions and competition when in seedling stage (Fenner & Thompson, 2005;Lebrija-Trejos et al., 2016). However, larger seeds are more susceptible to post-dispersal predation due to their greater nutritional value (Jansen et al., 2004).
The diversity in the propagules size is an interesting aspect in the evolution of the plants since it has already been demonstrated that the seed mass strongly influences other important characteristics, in addition to the shape and type of fruit dehiscence, implying in the geographic species distribution and interactions (Barroso et al., 1999). At the base of the Development, v. 9, n. 9, e879997956, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i9.7956 16 slope, where more zoochorous individuals were recorded, there is greater proximity to a watercourse, where Myrsine umbellata is abundant, which has attractive fruits to the fauna and can bring other zoochoric diaspores, confirming the importance of biotic agents in the gene flow in forest formations. Junior et al., (2012) pointed out that the relationship between plants and frugivores is essential for the conservation and maintenance of ecosystems, and the proportion of zoochoric species recorded in this study follows the dispersion pattern described for tropical forests (Howe & Smallwood, 1982;Stefanello et al., 2010).
Anemochorous diaspores occurred in a smaller amount, being generally associated with pioneer species, in dry environments, and less frequent than zoochorous in tropical forests (Wilkander, 1984). Pioneer species play an important role in environmental restoration, as they assist in the forest regeneration, guaranteeing their resilience and facilitating the process of forest succession after natural or anthropic disturbances (Capellesso et al., 2015;Mantovani & Martins, 1988;Martins et al., 2012;Vieira & Scariot, 2006)

Conclusion
We demonstrated that, on a small spatial scale, the landform represents an important source of heterogeneity from the vegetational component, determined by the fact that the study area is located in a region formed by geologically complex land, with varied lithologies and a mosaic of phytophysiognomies shaped by the combination of topography, microclimate, and altitude.