Woody flora of natural forest gaps in a bamboo-dominated forest remnant in southwestern Amazonia

1 Universidade Federal do Acre, Programa de Pós-Graduação em Ecologia e Manejo de Recursos Naturais, BR-364, Distrito Industrial, CEP 69920-900, Rio Branco, AC, Brazil 2 Universidade Estadual da Paraíba, Rua Horácio Trajano de Oliveira, CEP 58020-540, João Pessoa, PB, Brazil 3 Universidade Federal do Acre, Centro de Ciências Biológicas e da Natureza, BR-364, Distrito Industrial, CEP 69920-900, Rio Branco, AC, Brazil * Corresponding author. E-mail: alissonsobrinho@hotmail.com


INTRODUCTION
Tropical forests have the greatest plant diversity in our planet (Dirzo and Raven 2003).Recent estimates point that the Amazon is the home for approximately 16,000 tree species (DBH ≥ 10 cm), from which 227 are superdominant, because they are much more abundant the other species (Ter Steege et al. 2013).Despite this information and knowledge, few studies address species richness during regeneration in Amazonian forests.
Canopy gaps formed by one or more falling trees (Runkle 1992) are the most common and studied type of forest disturbance (Schliemann and Bockheim 2011) and are also thought to be one of the major drivers of species diversity at the local scale (Connell 1978).Nevertheless, recent studies have shown some divergence in the application of this hypothesis, suggesting that canopy gaps play a relative neutral role in the maintenance of diversity, mediating the limitation effect upon recruitment (Hubbell et al. 1999;Sheil and Burslem 2003;Obiri and Lawes 2004;Fox 2013).Other studies also associate gap area and heterogeneity with tree species composition (Brokaw 1985;Denslow 1987;Brokaw and Scheiner 1989).
Bamboo (Guadua spp.) dominated forests cover more than half of southwestern Amazonia (Carvalho et al. 2013).These are considered uncommon and differ structural and floristically from closed canopy forests in central and eastern Amazonia (Torezan and Silveira 2000;Griscom andAshton 2003, 2006;Griscom et al. 2007).Still, although these studies address regeneration patterns under the influence of bamboo, there is still little information about composition and diversity in natural gaps in these bamboo-dominated forests.
Our goal in this paper was to identify the floristic composition and to analyze the structure of regenerating woody plants in natural forest gaps and adjacent understory in a bamboo dominated forest remnant in southwestern Amazonia, Acre, Brazil.Our findings will increase the knowledge of the composition of species after gap opening in such forests, where there is scant literature about it.

MATERIALS AND METHODS
This study was carried out at the Catuaba Experimental Farm (FEC; Figure 1), a forest fragment with ca.1,200 ha located in the state of Acre, Brazil (10°04ʹ S, 067°37ʹ W).It has a gently rolling topography with predominance of oxisols and ultisols (Acre 2006); horizons A and B are predominantly sandy (62 and 47%, respectively); pH approximately 4.0 (Sousa et al. 2008) Plant identification was first made in the field with the aid of an experienced parataxonomist.Also, we sampled from all morphotypes identified in the field as a species for identification in the herbarium, based on the Angiosperm Phylogeny Group (APG III 2009).In order to check for proper species names spelling, we used the Brazilian Flora List (Flora do Brasil 2020, under construction, 2016).All fertile samples had their vouchers incorporated in the collection of the Botany and Plant Ecology Laboratory (LABEV) of the Federal University of Acre, Rio Branco, Acre, Brazil.Sterile specimens were not incorporated at the herbarium.
We surveyed forest gaps in this fragment and walked through 10 km of trails, following Runkle's method (1992).All gaps formed by the fall of one or more trees in a PPBio module (Programa de Pesquisas em Biodiversidade), established at FEC, following RAPELD protocols (Magnusson et al. 2005).Only gaps ≥ 100 m 2 were included in our study.
We applied Runkle's operational definition of a gap, which includes the soil area under the canopy opening, extending to the trunk of the adjacent trees.We surveyed 20 gaps.In each gap, we established eight subplots (2 × 4 m) following Brandani et al. (1988).For the understory, we randomly selected nine gaps (from the initial 20) and about 20 m away from each gap edge, we plotted a 2 × 32 m plot, divided in eight subplots of 2 × 4 m.
In each subplot (both gaps and understory) all woody plants ≥ 1 m tall and with DBH≤10 cm were sampled.Each plant was marked with a numbered tag.Species were also classified into conservation or threat status in agreement with the Brazilian Flora Red List (Martinelli and Moraes 2013).
To analyze the structure of regeneration in both environments abundance of species distribution curves were constructed (species abundance distribution: SAD), through the rank of the most abundant species for rarer (McGill et al. 2007;Matthews and Whittaker 2014).The length of the curves allows analyzing the species richness using the x-axis; and the slope allows an analysis of evenness among species, by reading the axis of ordinates.In this sense more inclined curves and smaller have fewer species and most dominant (Magurran 2005).To test whether there are differences between the SAD curves we used the Kolmogorov-Smirnov test.

RESULTS
The average gap area was 521 ± 347 m², ranging from 108 to 1,413 m², median 353 m².Total gap area was 10,429 m².The average canopy openness was 49%, while for small, medium and large gaps it was 50%, 47% and 52% respectively.Forest understory showed an average of 14% of canopy openness and was significantly (F=11.05;p<0.001) different from gaps, which did not differ among each other.
We sampled 1,656 shrubs and trees, 159 species, 116 genera and 45 families in both gaps and understory (Table 1).Sixty-two species were found only in gaps and 14 only in the understory.Eighty-six species co-occurred in gaps and understory.Eighty-three percent of the species are trees, 11% are shrubs and 6% were not classified in any habit due to lack of species identification.Many species were rare, with 41 sampled only once and 23 just twice.
The most specious families were Fabaceae (27 species, from all three subfamilies) comprising 17% of species richness; Rubiaceae (13); Moraceae (12); Lauraceae, Malvaceae and Sapotaceae (7).Four families were represented by two species and 19 families had only one species each.Piperaceae showed the highest number of individuals, comprising 15% of total abundance.Together, Piperaceae, Fabaceae, Moraceae and Rubiaceae hold 49% of all plants sampled.
For the forest as a whole (gaps and understory), the species with highest relative abundance were Piper sp. 1 (13.0%),Faramea capillipes Müll.Arg. ( 5 (2.1%).From these most abundant species, three were also found to be the most frequent genera regenerating in the same forest fragment: Neea, Guarea and Celtis (Silva 2011).Some of the species showed gap dependence for regeneration, showing a much greater abundance in gaps.Piper sp. 1 had 93% of its individuals in gaps, Faramea capillipes 83%, Tachigali setifera and Brosimum guianense with 69% of their individuals in gaps as well.On the other hand, only two species were found mainly or solely in the understory, Compsoneura ulei Warb.(63%) and Randia armata (Sw.)DC. (100%).Nevertheless, most species did not present any pattern or preference for either gaps or understory (Table 1).
Table 1 shows also the threat status of each species sampled at FEC.Only three species are classified as "vulnerable VU", which according to the Red List (Martinelli and Moraes 2013), face a high risk of extinction in the wild.Six other species were classified as "least concern LC", which means there is lack of information available now, but could be included in VU with further studies (Martinelli and Moraes 2013).
Figure 2 shows the species abundance distribution (SAD) within the gaps and the understory.The Kolmogorov-Smirnov test showed that there is a structural difference between the two environments (p<0.001).

DISCUSSION
The results shown here for Fabaceae are characteristic of Amazonian forests, where the family presents the highest diversity and abundance (Steege et al. 2013).For the genus Piper, such high relative abundance in gaps was expected since species from this genus are known to be light dependent, have higher growth rates and more abundant in natural gaps (Denslow et al. 1990;Daws et al. 2002;Bernades and Costa 2011).
Differences in abundance and dominance of species rank (Figure 2) will indicate structural differences, which are modified on the environment and their colonization by different species, caused by disturbance when the formation of natural gaps (Connel 1978;Connel and Green 2000;Denslow 1987Denslow , 1995)).The curve of species distribution that occurred in clearings is steeper than the abundance curve of the understory.This shows that the dominance is higher in gaps, caused by the abundance of the species Piper sp. 1 and Faramea capillipes.In understory only Eugenia sp. 2 has mild dominance.In this sense, the distribution of abundances of species of understory is more evenness compared with clearings.
The three vulnerable species (     (Vogel), Hymenaea parvifolia Huber and Mezilaurus itauba (Meisn.)Taub.ex Mez) are commercially exploited in the Amazon, mainly for timber.Such a status for these species mean that more attention is needed when harvesting permits are granted from government.
and also by the use of a functional group list, developed by the Botany and Plant Ecology Laboratory of the Federal Humb.& Bonpl.(Lecythidaceae), Tetragastris altissima (Aubl.)Swart (Burseraceae) and Carapa guianensis Aubl.(Meliaceae).Canopy height varies from 20 to 40 m, with emergent trees up to 45 m

Figure 1 .
Figure 1.PPBio's plots following RAPELD protocol, in the Fazenda Experimental Catuaba.Each black dot identifies a 250m trail/plot (following RAPELD's protocol for PPBio) and addition trails are yellow dot between the black dots.

Continued
Maranho et al.  | Woody flora of treefall gaps in bamboo-dominated Amazonian forest = gaps and U = understory; ²Functional groups: P = pioneers species, ES = early secondary species, LS = late secondary species, and UN = "unclassified; ³Status available only for the species present in the Brazilian Flora Red List, where: VU = vulnerable, LC = least concern, DD = data deficient.

Table 1 .
Tree and Shrub flora and abundance, occurring in natural gaps and understory in a forest fragment in the Fazenda Experimental Catuaba, Senador Guiomard, Acre Brazil.