Abiotic factors drives floristic variations of fern ’ s metacommunity in an Atlantic Forest remnant

We analyzed floristic variations in fern’s metacommunity at the local scale and their relationship with abiotic factors in an Atlantic Forest remnant of northeastern Brazil. Floristic and environmental variations were accessed on ten plots of 10 × 20 m. We performed cluster analyses, based on Bray-Curtis dissimilarity index to establish the floristic relationship. The influence of abiotic factors: luminosity, temperature, relative air humidity and relative soil moisture was evaluated from a redundancy analysis. We found 24 species belonging to 20 genera and 12 families. The fern’s flora showed high floristic heterogeneity (>75% for most of the plot’s associations). The fern’s metacommunity was structured along an abiotic gradient modulated by temperature, luminosity, and relative soil moisture.


Introduction
The metacommunity represents a group of species that potentially interact and that are spatially segregated into distinct patches connected by dispersal (Wilson, 1992).This concept highlights the importance of spatial scale affecting the role of the process that structures communities (Leibold et al., 2004).However, metacommunities have no clear boundaries, and different species respond and adapt across scales (Leibold et al., 2004).
At the local scale, the variations in species composition are often related to heterogeneity of environmental factors in space (Pyke et al., 2001;Zuquim et al., 2007).Specifically in tropical forests, the edaphic factors, luminosity, and water availability are associated with floristic dissimilarities across habitats (Tuomisto and Poulsen, 1996).Floristic variations may also be explained by historical factors, random dispersal or limitations species dispersal (Hubbell, 2001).Considering microhabitats, the floristic composition of metacommunities can be related to processes as patch-dynamic, which all patches are ecologically equaled and with dispersion playing the major role, or to species sorting from environmental gradients and niche (Leibold et al., 2004).
The distribution of tropical plant species has been intensely altered because of environmental degradation resulting in the creation of forests with new dynamics.Typically, tropical forests have become smaller and isolated, consequently, the dynamic of communities at local scale changes by a loss in the complexity of vegetation and biotic and abiotic homogenization (Murcia, 1995;Laurance, 1999;Lôbo et al., 2011).Unfortunately, this is a reality for Brazilian Atlantic Forest, intensively and widely deforested, remaining only about 8% of original cover, represented mostly small and isolated forests (Myers et al., 2000;Ribeiro et al., 2009), being one of the worst scenarios recorded in the northeast region.
In this sense, the aim of this study was to analyses the floristic variations in fern's metacommunity and their relationship with abiotic factors in an Atlantic Forest remnant of northeastern Brazil.Ferns offer several advantages as a focal community because variations in species composition are often related to environmental heterogeneity (e.g.Tuomisto and Poulsen, 1996;Tuomisto et al., 2003).In addition, the high ability of airborne ferns' spores indicates no dispersal limitation in terms of local distribution (Zobel et al., 2002).

Study area
The study was carried out in an Atlantic Forest remnant with Lowland Rain Forest vegetation (470 ha) surrounded by sugar cane matrix, situated in the district of Rio Formoso, Pernambuco, Northeastern of Brazil (8°38'58" S, 35°10'21" W, maximum: 100 m a.s.l.).According to Peel et al. (2007), the region presents hot and wet climate.The average annual temperature is 25.2 °C and the average annual rainfall is approximately 2300 mm.The short dry season is from October-December, with rainfall under 60 mm (Instituto de Tecnologia de Pernambuco, 2010).

Data collection
The fern community at the Atlantic forest of northeast of Brazil is spatially structured from specific microhabitats or "preferential habitats" (Ambrósio and Barros, 1997;Pereira et al., 2014).We walked the entire fragment and based on preferential habitats, delimited ten plots (10 × 20 m) that were at least 40 m apart from each other.Each plot was characterized according to the luminosity, measured with a light meter, as well as, the temperature, the relative air humidity, and the relative soil moisture, measured with a thermohygrometer (Table 1).Additionally, the plots were identified according to their occurrence in the interior or edge forest.We considered as edge forest the first 40 m of forest from the forest-matrix boundary (Silva et al., 2011).
The plot survey was carried out in June and July 2013 included all terrestrial ferns and epiphytes fixed until 2 m in phorophytes.For species identification, we utilized identification keys and specialized literature (e.g.Tryon and Tryon, 1982;Salino and Almeida, 2015).The species nomenclature was consulted from database Tropicos (2017), of Missouri Botanical Garden.The specimens were deposited in the Herbário UFP-Geraldo Mariz.

Data analyses
We calculated the Bray-Curtis index between plots to cluster analyses, using unweighted pair group method with arithmetic mean (UPGMA).A Multi-Response Permutation Procedure (MRPP) was carried out to test whether the groups were significant and not explained by chance.The Indicator Species Analysis (ISA) was performed to detect indicators species, with statistical significance assessed via Monte Carlo test, 1000 randomizations.We considered robust an indicator value (IndVal) ≥ 25% (Dufrêne and Legendre, 1997).A redundancy analysis (RDA) using linear combination "LC" scores was performed to evaluate the relationship between floristic variations and abiotic factors.The significance of RDA axis was tested applying Monte Carlo randomizations (1000).For all statistical analyses p values ≤ 0.05 were considered significant.The analyses were carried out in Fitopac 2.1 (Shepherd, 2010) and Pcord 4.0 (McCune and Mefford, 1999).
The RDA showed 86.63% of total cumulative variance explained in the two first axis: 53.38% and 33.25% for the first and second axis, respectively (Figure 2), both significative (p = 0.0100).The first axis was positively correlated with relative soil moisture (0.873), and negatively with luminosity (-0.819) and temperature (-0.837).The second axis was negatively correlated with relative air humidity (-0.976).This results indicated the presence of an abiotic gradient, represented in the first part by sites with high luminosity and temperature and low relative soil moisture (plots 3, 7 and 8), and in the second part for shaded sites, with high relative air humidity and high relative soil moisture (Figure 2).

Discussion
Our data support a strong floristic dissimilarity pattern for ferns at a local scale.The species richness data of the study area denotes that an accentuated floristic variation occurs even with few species (see Pereira et al., 2013 to access information about species richness in northeastern Atlantic Forest).Therefore, this pattern does not seem to be related to species richness.
The floristic variation in ferns' metacommunity at local scale drove by abiotic factors has been indicated for other herbaceous groups or tree species in tropical forests (Poulsen et al., 2006;John et al., 2007).Specifically for ferns, abiotic factors as water availability and shading are key-factors affecting ecological patterns (Ferrer-Castán and Vetaas, 2005;Karst et al., 2005).These factors are directly related to the biology of ferns for both the gametophyte and sporophyte phases (Windisch, 1990).The higher water availability and lower luminosity certainly favor spore germination and the establishment of gametophytes (Page, 2002).Additionally, these factors are fundamental for the physiological maintenance of the sporophytes, which typically has low evaporative control and high intolerance to abiotic fluctuations (Page, 2002).Therefore, even ferns being a group with high ability of airborne, the establishment of individuals and species distribution does not occur from a random pattern.
The species N. brasiliense was an indicator of areas with low water availability as edge plots, with low abundances or absences in water-saturated plots.This fern is widely distributed in the Brazilian Atlantic forest, occurring in several habitat types (Dittrich, 2005).The high abundance values  observed seems to be the result of vegetative propagation that can be favored by edge conditions (Murcia, 1995), and is typically recorded in the Blechnaceae family (Dittrich, 2005).Indeed, edges can favor species dominance (Murcia, 1995), even for some ferns species (e.g.Silva et al., 2011).
We showed that ferns' metacommunity has an accentuated floristic variation at the local scale, which was modulated by an environmental gradient, despite the low species richness of the forest remnant.

Figure 2 .
Figure 2. Ordination diagram of the first two axes of Redundancy Analysis for the fern's flora of 10 plots surveyed in a Lowland Atlantic Forest Remnant in the Northeast of Brazil (Pernambuco, Brazil).RSM= Relative soil moisture; RAH= Relative air humidity.Blue circles indicate edge forest and red circles interior forest.

Table 1 .
Plots description and environmental conditions of ten plots established along a Lowland Atlantic Forest remnant in Northeaster of Brazil.

Table 2 .
List of ferns species surveyed in 10 plots in a Lowland Atlantic Forest Remnant in the Northeast of Brazil (Pernambuco, Brazil).