Floristic composition in ecotone forests in northern Brazilian Amazonia: preliminary data

Abstract Background Ecotone has been defined as “a multi-dimensional environmentally stochastic interaction zone between ecological systems with characteristics defined in space and time, and by the strength of the interaction” (Hufkens et al. 2009). This is a known concept to define transitional zones between two or more ecological communities, ecosystems or biotic regions. Ecotone forests, dispersed in northern Brazilian Amazonia, are natural formations which have been largely affected by anthropogenic impacts, such as deforestation and fire. Maracá Ecological Station, State of Roraima, Brazil, is a protected area with extensive representations of ecotone forests in this region of the Amazonia. Forest inventories and floristic surveys are important as they extend our knowledge (1) of forest structure and tree species composition and (2) of tree and palm species ecology in this region of the Amazonia. Both improve our ability to predict changes in plant diversity, considering the future scenarios of climate change in comparison with previous surveys performed in Maracá. New information We present a forest inventory carried out in 129 plots (10 m x 50 m; 6.45 ha in total) dispersed in a grid (5 km x 5 km) located in a forest zone ecotone in the eastern part of Maracá Ecological Station. All stems (tree + palm) with diameter at breast height ≥ 10 cm were recorded, identified and measured. A total of 3040 stems were recorded (tree = 2815; palm = 225), corresponding to 42 botanic families and 140 identified species. Seven families and 20 genera contained unidentified taxa (12.2%). Sapotaceae (735 stems; 10 species), Leguminosae (409; 24) and Rubiaceae (289; 12) were the most abundant families. Peltogyne gracilipes Ducke (Leguminosae), Pradosia surinamensis (Eyma) T.D.Penn. (Sapotaceae) and Ecclinusa guianensis Eyma (Sapotaceae) were the species with the highest importance value index (~ 25%). The dominance (m2 ha-1) of these species corresponds to > 36% of the total value observed in the forest inventory. Our dataset provides complementary floristic and structure information on tree and palm in Maracá, improving our knowledge of this Amazonian ecotone forest.


Introduction
The Pan-Amazon or Continental Amazon has the largest tropical forest area on the planet with > 6 million km (MapBiomas 2019, WWF 2019). Estimates based on > 1,100 permanent plots scattered throughout the region indicate that the tree richness ranges between 7-10,000 species occupying a great diversity of habitats (Cardoso et al. 2017. Brazilian Amazonia accounts for the largest physical area of this region (Salati and Vose 1984; > 5 x 10 km ). However, it has been long threatened by a series of anthropogenic impacts, such as the replacement of native forest by pasture and soybean planting (Almeida et al. 2016, Fearnside 2006, combined with an increase in burned areas of primary and secondary forests (Alencar et al. 2015, Aragão et al. 2018, Barni et al. 2015. Modern anthropogenic activities, associated with global warming, have a negative effect on tree diversity and reduce the conservation status of Amazonian forests (Barlow et al. 2016, Esquivel-Muelbert et al. 2019. Despite having large areas of ombrophilous forests, the ecotone forests are important ecological areas because they occupy the peripheral zone to the Amazon basin (Central Amazonia) and are considered as the forest formations most impacted by anthropogenic activities in the Southern (Marques et al. 2019, Nogueira et al. 2015 and Northern (Barni et al. 2015, Santos et al. 2013) Amazonian "arcs of deforestation". The accelerated increase of anthropogenic 2 -1 2 6 2 activities within the Amazonian ecotones has been a major contributor to the fact that the region as a whole is now approaching to the "tipping point", limiting forest growth and potentially favouring low-density environments when compared to those currently supported by the region (Lovejoy and Nobre 2018).
Several floristic surveys and forest inventories have been carried out in these peripheral regions, especially from 1980-1990(Nelson and Oliveira 2001. However, this period was insufficient to accumulate realistic information on forest structure and floristic in view of the continuous advance of anthropogenic activities. This impedes the reliable characterisation of plant diversity in these peripheral Amazonian regions. One such area is the State of Roraima, located in the northern part of the Brazilian Amazonia. This ~225,000 km area contains ecotone forest zones of great geoecological importance because they are located between Guiana Shield (highland savannas and tropical dry forests) and Central Amazonia (lowland tropical forests) Bacelar-Lima 2008, Oliveira et al. 2017), which gives them a unique phytogeographical and ecoevolutionary history, where a high degree of endemism is observed and unique vegetation sets exist Proctor 1997, Milliken andRatter 1998). However, the region has received few forest inventories and floristic surveys (Suppl. material 1); main of them are associated with the monodominant forests , campinas and campinaranas Ferreira 2004, Damasco et al. 2013) and forest fragments dispersed in savannas (Sette-Silva 1993, Santos et al. 2013, Jaramillo 2015.
In an attempt to expand studies on plant diversity in this part of the Amazon, two PPBio (Biodiversity Research Program, https://ppbio.inpa.gov.br) 25-km research grids were installed in areas defined as ecotone forests in Roraima taking into account the Brazilian Vegetation Classification System (Brazil-IBGE 2012): Maracá Ecological Station and Viruá National Park. Both are protected areas under Brazilian government management. Surveys of structure (vertical and horizontal) and tree species composition of the main forest types of these research grids have been expanded in Maracá , Villacorta 2017 and Viruá (Damasco et al. 2013. Accordingly, the current study provides preliminary data from a forest inventory carried out in the Maracá research grid. These data expand the scale of floristic and structural observations in this northern Amazonian ecotone zone from those initiated by Milliken and Ratter (1998) and .  , Barbosa and Campos 2011). Although it was inhabited until the mid-1970s (Proctor and Miller 1998) and there has been an increase in anthropogenic pressures in adjacent regions (Couto-Santos et al. 2014), the conservation status of Maracá remains excellent and the area may be considered an important representative of mature forest tree species diversity for the region. Eastern Maracá soil classes are directly related to drainage and relief (51-99 m a.s.l.). These local constraints determine different forest types that occupy such seasonal flooding-free areas as moist lowlands and rocky slopes (Villacorta 2017). Well-drained soils are situated in areas of higher reliefs (Ultisols) or on slopes dominated by base-rich rocky soils. Soils occupying the lower-lying areas are poorly drained with a dominance of typical hydromorphic sandy soils (Nortcliff and Robison 1998). Regional climate is characterised as a transition between Aw/Am climate subtypes by the Köppen classification (Barbosa 1997). Average annual rainfall is ~ 1900 mm , with a rainy season between April and September (140-420 mm month ) and a dry season between October and March (40-130 mm month ) (Couto- .

Analysis:
We calculated abundance (number of stems) and richness (number of species) for all arboreal stems ≥ 10 cm in diameter (tree and palm) recorded in the floristic inventory. Frequency and dominance (absolute and relative) were also calculated to estimate the importance value indexes for family (FIV) and species (IVI). All stems were classified by diameter size and total height classes to analyse the horizontal-vertical structure. We tabulated all floristic composition and diameter data (n = 129 plots) using a multiple interacting spreadsheet programme (Excel Office 365) and applied the vegetation analysis methods described in Kent and Coker (1994) to calculate frequency, dominance and importance value indexes (family and species). . Family-level delineations followed APG-IV (2016).
Step description: The floristic survey described here was done in two periods (March and December 2017).

Geographic coverage
Description: Data was collected in 129 plots across PPBio-Maracá research grid located on the eastern end of Maracá Ecological Station (see Fig. 1  2), all of these being encountered at low abundance. Thirty-four species were represented by a single stem.

Column label Column description
basisOfRecord The specific nature of the data record.
language Language of the resource.
institutionCode Institution that has custody of the object or information about its registration. collectionCode The name or acronym of the collection or dataset from which the record is derived. occurenceID A identifier for the occurrence (unique).
catalogNumber An identifier (preferably unique) for the record within the dataset or collection.
habitat Description of the habitat in which the event occurred. The geographical latitude in decimal degrees of the geographical centre of a location.
decimalLongitude The geographical longitude in decimal degrees of the geographical centre of a location.
geodeticDatum The ellipsoid, geodetic datum or spatial reference system (SRS) in which the geographical coordinates given in decimalLatitude and decimalLongitude are based.
kingdom Full scientific name of the kingdom in which the taxon is classified.
family Full scientific name of the family in which the taxon is classified.
genus Full scientific name of the genus in which the taxon is classified. specificEpithet The name of the species-specific epithet. intraspecificEpithet The name of the terminal or lower-level infraspecific epithet of the scientific name. scientificName The full scientific name. It must be the name of lowest level taxonomic rank that was determined.
vernacularName Common or vernacular name.
taxonRemarks Comments or notes about the taxon or name.

Discussion
Our floristic composition results for ecotone forests on eastern Maracá Island complement previous investigations carried out at the macro (Milliken and Ratter 1998) and micro (Thompson et al. 1992 sampling scale at this location. Although our study was not conducted to differentiate the forest types comprising this ecotone region, the broad dispersion of 129 small plots (0.05 ha each) over a wide area (25 km ), aids understanding of the various forest types present in the area. This occurred since small plots were better suited to the environmental variability scale of this ecotone region, because they covered specific sampling areas of each forest type comprising the ecotonal mosaic. Use of plots smaller than those commonly used in tropical forest inventories (0.5-1.0 ha; e.g. Phillips et al. 2016) may be an alternative for floristic surveys or forest inventories in regions with high environmental variability. However, sampling, using small plots in Maracá, followed basic rules: (i) sampling design maintained the independence of each sampling unit, (ii) large number of samples (> 100) to adequately represent the environment and (iii) annual tree censuses. All these recommendations are important to reduce the coefficient of variation between samples and to avoid temporal measurement problems associated with the descriptors of floristic composition, dynamics and forest structure (Keller et al. 2001, Wagner et al. 2010 Ratter 1998, Nascimento et al. 1997), retained the proportional representation they had in surveys two decades earlier -indicating the site had little anthropic impact in the intervening time period. Additionally, the similarity of our larger sample to smaller-scale efforts of the past indicates that compositional units repeat, underscoring the near-fractal nature of the vegetation mosaic. The main families observed in Maracá are common throughout the Amazon, are always present in forest inventories and floristic surveys and almost always have the largest number of species, so that they are considered to be hyperdominant in the region . For example, Leguminosae and Sapotaceae (Condé and Tonini 2013) and Burseraceae, Chrysobalanaceae and Leguminosae (Alarcón and Peixoto 2007) were also families with high importance value indices in other Roraima forest types. We emphasise that the importance of each family in these surveys differs from Maracá, indicating that the ecotone forests of this region have their own composition and dynamics, so differing from mosaics or forest types observed elsewhere in the State of Roraima.
As with the families, most plant species, described in past inventories, are also present in our list, especially those with higher IVI (P. gracilipes, P. surinamensis, E. guianensis), besides Lecythis corrugata subsp. rosea (Lecythidaceae), Attalea maripa (Arecaceae) and Licania discolor (Chrysobalanaceae), all of which were strongly represented in previous surveys. The case of L. discolor Pilg. is the most interesting because the individuals attributed to this species in our work were largely attributed to L. kunthiana Hook.f. in previous surveys, a very similar taxonomic species, but were far less abundant at Maracá. The proportion and spatial distribution of species inventoried in Maracá ecotone forests are conditioned by a variety of environmental filters. For example, P. gracilipes is a deciduous species endemic to this area of the northern Amazonia that can form monodominant agglomerations . However, the monodominance of this species occurs only in seasonally flooded areas in bottom lands, with high Fe concentration in the soil . Such conditions are non-existent or rare in fertile and flood-free soils (Villacorta 2017). These distinct P. gracilipes-associated environmental characteristics reveal an ecotone region where forest types are intertwined with topographic, edaphic and hydrological constraints.
The species P. gracilipes plays an important ecological role (IVI = 10.4%) and it has been used to delimit forest types on Maracá. For example, Milliken and Ratter (1998) used this species to define forests monodominant with Peltogyne as "Peltogyne gracilipes forest". Similarly,  used this species to divide the Maracá ecotone zone into three forest types: (i) PRF (Peltogyne-rich Forest) or forests monodominate with P. gracilipes, (ii) FWP (Forest without Peltogyne) or types where this species do not occur and (iii) PPF (Peltogyne-poor Forest), which are mixed types with low P. gracilipes abundance. The three types correspond analogously to the Deciduous Seasonal Forest (= PRF), Semideciduous Seasonal Forest (= PPF) and Open Ombrophilous Forest (= FWP) 2 -1 -1 2 -1

+2
The results of this study agree with data from previous investigations, indicating that the environmental heterogeneity of the ecotone forests of eastern Maracá Island influences floristic richness and structural distinctions, with P. gracilipes abundance acting as a descriptor variable for forest types. Consequently, the floristic survey conducted by this study is important because it expands our knowledge of forest structure and tree species composition in ecotone zones of the northern Brazilian Amazonia, improving our ability to predict changes in species composition and plant diversity when we take into account comparisons between previous forest inventories performed in Maracá. Finally, this study contributes to the local floristic knowledge, complements the herbarium collections with new collections, allows the development of similar studies and also enables the elaboration of management plans for the conservation of the local biota.
of the Brazilian Vegetation Classification System (Brazil-IBGE 2012), all with distinct hydroedaphic and topographic characteristics (Carvalho et al. 2018). These forest types definitions have enhanced the understanding of variation in biomass/carbon stocks estimates for the Maracá Island (Nascimento et al. 2007.