Long-term monitoring of diversity and structure of two stands of an Atlantic Tropical Forest

Abstract Background This study aimed to report the long-term monitoring of diversity and structure of the tree community in a protected semideciduous Atlantic Forest in the South of Minas Gerais State, Southeast Brazil. The study was conducted in two stands (B and C), each with 26 and 38 10 m x 30 m plots. Censuses of stand B were conducted in 2000, 2005 and 2011, and stand C in 2001, 2006 and 2011. In both stands, the most abundant and important species for biomass accumulation over the inventories were trees larger than 20 cm of diameter, which characterize advanced successional stage within the forest. New information The two surveyed stands within the studied forest presented differences in structure, diversity and species richness over the time.


Introduction
The formation of the structure of tropical forests is governed by a wide range of factors (e.g. abiotic, biotic, neutral and natural or anthropogenic disturbance) (Guariguata and Kattan 2002, Hubbell 2001, Phillips et al. 2004). These factors lead to the establishment of a mosaic of forest patches of different ages Kattan 2002, Oldemann 1990). These mosaics exhibit high environmental heterogeneity, which, associated with a combination of different floras, can lead to the formation of distinct local ecological patterns within a forest (Fortin and Dale 2005). However, the formation of these within-site differences is poorly understood, requiring further knowledge about community functioning. Tropical forests are both highly ecologically valuable and at great threat from disturbance. It is therefore vital to improve our knowledge of the mechanisms driving the formation of forest structure in order to protect and conserve the remaining areas of tropical forest.
Long-term research (community dynamics) is a vital tool in elucidating the causes of a wide range of ecological patterns and processes in plant communities (Phillips 1996). A better comprehension of events such as mortality, recruitment and ecological succession (Oliveira Filho et al. 2007, Saiter et al. 2011) will provide information about species' behavior and community turnover, which may prove crucial to the success of projects aiming to conserve and restore vulnerable biomes from Tropical Forests.
Due to its ecological complexity, its endemic biodiversity and its threats by anthropogenic activities (e.g. agriculture, mining), the Brazilian Atlantic Forest is considered one of 25 hotspots of biodiversity (Mittermeier et al. 2005). Compared with its original cover, just 11.73 % (approximately 16,377,472 hectares) remains today, and few of these areas are adequately protected (Ribeiro et al. 2009). However, these few remaining protected sites provide a rare opportunity to monitor plant community dynamics, and to compare intact forest with heavily disturbed and fragmented stands. Therefore, this study aimed to report the long-term monitoring of diversity and tree community structure in two stands of a protected Atlantic Forest in the South of Minas Gerais State, Southeast Brazil. In addition, this paper also aimed to make data regarding forest dynamics from the studied site publicly available in order to encourage further research about the composition, diversity and structure of Atlantic Forests over time, thus contributing to the preservation of this threatened Biome.
In 1999, 90% of the current area of the PEQRB was designated as protected in a decree from Lavras' city hall, which includes the assignment to FAK. Before it was designated as a protected area, the PEQRB was subject to various disturbances. Disturbances such as free movement of cattle, logging for charcoal production and construction of housing and recreational areas (Dalanesi et al. 2004, Oliveira Filho andFluminhan-Filho 1999). Long-term monitoring of diversity and structure of two stands of an Atlantic ...

Sampling methods
Sampling description: Investigation of floristic and structural composition of the tree community in PEQRB, its distribution over soil habitats and evaluation of the interaction of plants with environmental factors, were studied by Dalanesi et al. (2004). In the present study, the authors resurveyed two stands (called B and C), located 480 m apart, using plots of 300 m Stand B was surveyed with 26 plots in 2000 comprising 0.78 hectares, which were kept and resurveyed in 2005 and 2011constituing its long-term monitoring. Stand C was surveyed in 2001 with 38 plots (1.14 hectares), then again in 2006 and 2011. The total cover of sampling area was 1.92 hectares.
The stands were arranged as transects (Fig. 2) perpendicular to the watercourse which bisects the park, and extends into two adjacent slopes, following principles outlined by Causton (1988). The samplings plots were contiguously allocated on each stand (Fig. 2). This sampling design of the was previously planned aiming to catch more environmental heterogeneity in the transects of both stands in order to analyze the relations among trees distribution with soil and topograghy (Dalanesi et al. 2004). All trees with diameter at breast height (DBH) greater than or equal to 4.99 cm were measured and permanently tagged. In the subsequent samplings, the surviving trees in both stands were remeasured, as well as all individuals which reached the criterion of inclusion. Additionally, the DBH of multistemmed trees was calculated as the square root of the sum of all squared stem DBH's. Only multi-stemmed trees with DBH ≥ 4.99 cm were included in the survey as recruits. 2.

Figure 2.
Surface grid of the sampled transects of each stand (B and C) studied in the forest of the PEQRB, in the municipality of Lavras, Minas Gerais State, Souheast of Brazil. The size of each plot in both stands was 30 x 10 meters. The spacing between grid lines is 5 meters. Adapted from Dalanesi et al. 2004. Individuals were identified either in the field or through collection of samples of whole branches, leaves, and where possible, fruits. These samples were then compared with the existing collection present in herbarium "Herbário ESAL" of the Federal University of Lavras. In addition, samples were verified using appropriate literature and where necessary, specialists were consulted. The classification system followed that of the APG IV (Angiosperm Phylogeny Group) (2016) and we verified spellings and synonymous to the species through TNRS page (Taxonomic Name Resolution Service) Boyle et al. (2013).

Species diversity and richness
Species richness was compared between stands B and C stands using a Wilcoxon Rank Sum Test. The species diversity and evenness per stand were calculated by Shannon-Weaver (H) and Pielou (J) respectively, using the package "vegan" (Oksanen et al. 2017). To compare the total number of individuals and basal area per plot within each stand, a Nested ANOVA was carried out followed by a post-hoc test using the functions lmer and difflsmeans from the package lmerTest (Kuznetsova et al. 2016). The pairwise comparisons of number of individuals and basal area between B and C were conducted with a Two-Sample T-test of independent samples. All analyses mentioned above were carried out in R software version 3.3.0 (R Development Core Team 2016). Shannondiversity between B and C was compared in Past software (Hammer et al. 2001) with a Hutcheson T-test (Hutcheson 1970).

Tree Dynamics
The changes in the tree community over the time were determined for each stand by calculating the annual mean rates of mortality (M) and recruitment (R) of individuals (based on species abundance) and basal area loss (L) and gain (G) according to Sheil et al. (1995), Sheil andMay 1996 andSheil et al. (2000). The time series were 11 years to stand B (surveyed in 2000, 2005 and 2011) and 10 years to stand C (surveyed in 2001, 2006 and 2011) constituting two periods to the calculation of the rates (B: 2000 and 2005, 2005 and 2011; C: 2001 and 2006, 2006 and 2011). In addition, the net rates of change in number of individuals and basal area were also calculated, based on the relationship between the abundance and basal area recorded in the first and most recent inventories. The net rate of change between inventories was calculated considering both abundance of trees (Ch ) and their basal area (Ch ) Korning and Balslev 1994. To describe the rate of change in tree community, the turnover rates regarding both abundance (T ) and basal area (T ), were calculated Phillips et al. 2004, Phillips 1996 for each stand.The turnover rates to number of individuals and basal area are respectively calculates as the averages of mortality, recruitment, and the loss and gain of basal area rates. . The difference between the number of dead and recruited trees was calculated using Poisson Counting (Zar 2010

Tree Dynamics per diameter classes
Diameter classes with increasing amplitudes were created (5-10 cm, >10-20 cm, >20-40 cm and >40-80 cm). This approach compensates for the sizeable decrease in abundance of the largest diameter individuals, which is a pattern commonly observed in tree diameter measurements and is characterized by the negative exponential distribution (Appolinário et al. 2005). To describe the temporal variations in each class, all individuals that underwent the following events: death, total ingrowth (inter-class imports through recruitment and growth) and total outgrowth (inter-class exports through growth and death) (Lieberman et al. 1985), were counted. The same procedure was used to calculate the absolute number of dead, absolute ingrowth (the absolute number of previously unrecorded individuals) and absolute outgrowth (the absolute number of individuals no longer present) per diameter class. To verify whether the frequency of the total number of dead trees within each stand (B: 2005 and 2011; C: 2006 and 2011) were dependent on diameter classes based on the frequencies expected from the second and last inventory diameter distribution, we used a G-Test of Goodness-of-Fit carried out with the package "DescTools" (Signorell 2016). The comparison between the total ingrowth and total outgrowth per class was carried out by calculating Exact Poisson Tests with the package "exactci" (Fay 2010). Both analyzes were carried out in R version 3.3.0.
To verify whether the frequency of dead trees within each stand (B: 2005 and 2011; C: 2006 and 2011) were dependent on diameter classes based on the frequencies expected from the second and last inventory diameter distribution

Dynamics of the most abundant species
The 10 most abundant species in each stand were selected and their mortality and recruitment rates were calculated. These species were also classified into regeneration guilds following the descriptions of Swaine and Whitmore (1988) and the field knowledge on the species of the authors of this study. This classification was used to aid in the deduction of the successional stage of the two forest stands. The difference between the absolute number of dead and recruits within the species was evaluated by Exact Poisson Tests in software R version 3.3.0 with the package "exactci" (Fay 2010).  In the stand B there was no significant difference in total basal area (F = 0.53, p = 0.593) or in total number of individuals (F = 0.23, p = 0.791) across all censuses. The same was evident in total basal area (F = 1.28, p = 0.279) and total number of individuals (F = 1.61, p = 0.201) in stand C ( Table 2). The total number of individuals did not differ between stands across all of the censuses. There were significant differences in total basal area between stands across all of the censuses ( Table 2).
The monitoring period of 11 years in stand B presented a decrease in the number of individuals per hectare, and an increase in basal area per hectare, from 2000 to 2011 ( Table 2). The increase of basal area per hectare was also observed in stand B from 2001 to 2011, but the number of individuals per hectare decreased from 2006 to 2011 (Table 2).

Tree Dynamics
There was a decrease in abundance and increases in basal area during both intervals in stand B (2000-2005 and 2005-2011; Table 3). This clear loss of individuals was confirmed by the negative net change in both intervals, which was higher in the second interval (-0.96 % year ), associated with an increase in mortality rate 3.02 % year ) and a reduction in recruitment rate (1.63 % year ). This was also reflected in a higher turnover a a a a a a N ind = total number of individuals; N Species = species richness; N (tree ha ) = number of individual per hectare; BA (m ha ) = Basal area per hectare;Total BA = total basal area; H' = Shannon-Weaver index (nats. indivídual ); J = Pielou's evenness index. A letter in the census of one stand followed by a different letter in the census of the another stand indicates significant difference in the comparisons. N ind was not significant different neither within nor among stands comparisons in all censuses. Total BA did not differ over censuses within each stand, but was different in all censuses comparisons among the two stands.

Tree Dynamics per diameter classes
Stand B showed higher total outgrowth (2005 to 2011: p = < 0.0001, Table 2) in the 5-10 cm diameter class in both intervals, which was the result of a higher absolute number of outgrowth in relation to absolute ingrowth. In the >10 to 20 cm class in 2011, the total outgrowth was significantly higher than total ingrowth (p = < 0.0001). This was due to an increase in the number of mortalities and a reduction in the raw ingrowth (Table 4). Conversely, between 2000 and 2011, the >20 to 40 cm class exhibited an increase in abundance, where 2005 showed a higher total ingrowth than total outgrowth (p = 0.0057    A progressive decrease of abundance in stand C was observed between 2001 and 2011 in the 5-10 cm class (Table 5). This was due to total outgrowth being significantly higher than total ingrowth in 2011 (p = < 0.0001) which was in turn caused by the increased mortality rate compared to 2006 (Table 3). In the >10 to 20 cm class the total outgrowth was higher than total ingrowth in both intervals ( Table 5.

Dynamics of the most abundant species
Among the 10 most abundant species in stand B (   recruitments. The first two were notable for high rates of mortality in 2011 (Z = 6.58. P = <0.0001 e Z = 7.92. P = <0.0001 respectively), whereas V. magnifica did not present recruitment and showed significant mortality in both intervals (2006: Z = 3.46. P = 0.0005 and 2011: Z = 3.68. P = 0.0002).
Eremanthus erythropappus also did not present recruitment in the second interval (between 2006 and 2011), in conjunction with higher rates of mortality (Z = 5.19. P = <0.0001) and only remained among the most abundant species in 2001. The shade tolerant E. acutata e R. jasminoides increased in basal area and abundance in both intervals (Table 6), but did not differ in mortality or recruitment. rights Information about who can access the resource or an indication of its security status.

Usage rights
rightsHolder A person or organization owning or managing rights over the resource.
bibliographicCitation A bibliographic reference for the resource as a statement indicating how this record should be cited (attributed) when used.
reference A related resource that is referenced, cited, or otherwise pointed to by the described resource. datasetName The name identifying the data set from which the record was derived. basisOfRecord The specific nature of the data record.
eventDate The date-time or interval during which an Event occurred. For occurrences, this is the date-time when the event was recorded. decimalLatitude The geographic latitude (in decimal degrees, using the spatial reference system given in geodeticDatum) of the geographic center of a Location. decimalLongitude The geographic longitude (in decimal degrees, using the spatial reference system given in geodeticDatum) of the geographic center of a Location.
acceptedNameUsageID An identifier for the name usage (documented meaning of the name according to a source) of the currently valid (zoological) or accepted (botanical) taxon.
parentNameUsageID An identifier for the name usage (documented meaning of the name according to a source) of the direct, most proximate higher-rank parent taxon (in a classification) of the most specific element of the scientificName.
nameAccordingToID An identifier for the source in which the specific taxon concept circumscription is defined or implied.
scientificName The full scientific name, with authorship and date information if known. When forming part of an Identification, this should be the name in lowest level taxonomic rank that can be determined. acceptedNameUsage The full name, with authorship and date information if known, of the currently valid (zoological) or accepted (botanical) taxon. parentNameUsage The full name, with authorship and date information if known, of the direct, most proximate higher-rank parent taxon (in a classification) of the most specific element of the scientificName. nameAccordingTo The reference to the source in which the specific taxon concept circumscription is defined or implied -traditionally signified by the Latin "sensu" or "sec." (from secundum, meaning "according to"). For taxa that result from identifications, a reference to the keys, monographs, experts and other sources should be given.
higherClassification A list (concatenated and separated) of taxa names terminating at the rank immediately superior to the taxon referenced in the taxon record. kingdom The full scientific name of the kingdom in which the taxon is classified. class The full scientific name of the class in which the taxon is classified. order The full scientific name of the order in which the taxon is classified. family The full scientific name of the family in which the taxon is classified. genus The full scientific name of the genus in which the taxon is classified. subgenus The full scientific name of the subgenus in which the taxon is classified. specificEpithet The name of the first or species epithet of the scientificName. intraspecificEpithet The name of the lowest or terminal infraspecific epithet of the scientificName, excluding any rank designation. taxonRank The taxonomic rank of the most specific name in the scientificName.
scientificNameAutorship The authorship information for the scientificName formatted according to the conventions of the applicable nomenclaturalCode.
nomenclaturalCode The nomenclatural code (or codes in the case of an ambiregnal name) under which the scientificName is constructed. Recommended best practice is to use a controlled vocabulary.
taxonomicStatus The status of the use of the scientificName as a label for a taxon. decimalLatitude The geographic latitude (in decimal degrees, using the spatial reference system given in geodeticDatum) of the geographic center of a Location. decimalLongitude The geographic longitude (in decimal degrees, using the spatial reference system given in geodeticDatum) of the geographic center of a Location. parentNameUsage The full name, with authorship and date information if known, of the direct, most proximate higher-rank parent taxon of the most specific element of the scientificName. nameAccordingTo The reference to the source in which the specific taxon concept circumscription is defined or implied.
higherClassification A list (concatenated and separated) of taxa names terminating at the rank immediately superior to the taxon referenced in the taxon record. kingdom The full scientific name of the kingdom in which the taxon is classified. class The full scientific name of the class in which the taxon is classified. order The full scientific name of the order in which the taxon is classified. family The full scientific name of the family in which the taxon is classified.

genus
The full scientific name of the genus in which the taxon is classified. subgenus The full scientific name of the subgenus in which the taxon is classified.
specificEpithet The name of the first or species epithet of the scientificName. intraspecificEpithet The name of the lowest or terminal infraspecific epithet of the scientificName, excluding any rank designation. taxonRank The taxonomic rank of the most specific name in the scientificName.

Structure, diversity and species richness
The diversity observed in stand C is amongst the highest observed in the region of this study (Pereira et al. 2006) and increased in the last census ( Table 2) (Table 2), indicates the increase of dominance by a few species, which were found to increase in basal area and number of individuals (Brower andZar 1990, Gurevitch et al. 2009). Thus, the increase in dominance reduces diversity and richness (Thorpe et al. 2011) demonstrating the natural successional advance which commonly promotes the exclusion of heliophiles and short life-span species (Swaine and Whitmore 1988).

Tree dynamics
The prevalence of basal area accumulation, in conjunction with reduction of abundance in stands B (in both intervals) and C (just in the first interval), gives rise to a mature forest. This is indicative of an advanced successional stage in both stands of the studied forest (Oldemann 1990, Oldeman 1989. In sites where clearcutting has taken place, successional change inevitably leads to a phase of self-thinning, whereby overcrowding creates high levels of mortality (Gomes et al. 2003, Oldemann 1990, Oliveira-Filho et al. 1997. Although PEQRB has a history of clearcutting and burning, there is no precise data or records of the exact location relative to the monitored stands, prior to the most recent survey carried out for this study in 2011. However, this pattern of biomass gain and local abundance decrease in the community is common in tropical forests which are protected from intense man-made disturbance, thus allowing the natural advance of succession (Chazdon 2008, Guariguata and Ostertag 2001, Letcher and Chazdon 2009, Letcher 2010, Muscarella et al. 2016. When natural succession occurs, tree interactions increase and events, such as competitive exclusion from the larger and taller canopy trees suppressing the smaller ones present in the understory, achieve higher importance in community assembly (Chazdon 2008, Letcher et al. 2012, Letcher 2010).
The transition of dynamic patterns observed in stand C (1° survey: simultaneous increase of abundance and basal area; 2° survey: increase of basal area and decrease of abundance) is possibly caused by gap formation (Whitmore 1988, Silva et al. 2011) and the dynamic balance between mortality and recruitment (Phillips 1996). These are some of the main factors responsible for the change in tree community structure in conserved and protected sites such as PEQRB (Oliveira-Filho et al. 1997, Letcher 2010, Phillips et al. 2004, Stephenson and Mantgem 2005. Another possibility could be that events of mortality and recruitment may appear to act at different intensities at a local level. It is possible that these incidents may be just part of the natural dynamics of a stable community (Felfili 1995, Korning and Balslev 1994, Phillips and Gentry 1994, Saiter et al. 2011, Sheil et al. 2000. However, the events of mortality and recruitment are often associated with stochastic factors (Hubbell 2001, Laurance et al. 2011) and therefore difficult to interpret in short-term temporal scales (Oldeman 1989).
The location of the stands in two different locations within the forest in PQERB can be also considered relevant as an explanatory factor for the observed rates of mortality and recruitment, because these stands encompass various ecological units of vegetation with distinct ages of formation, leading to a heterogeneous forest with multiple successional stages (Oldemann 1990, Phillips 1996.

Tree Dynamics per diameter classes
The J-inverted distribution of individuals per diameter class is typical of many tropical forests and has been reported to be the case in other forests in the region of the studied site (Appolinário et al. 2005, Guimarães et al. 2008, Oliveira Filho et al. 2007). This study showed that the forests of the PEQRB also conform to this distribution. The accentuated reduction of abundance in the smaller diameter class (5-10 cm) is implicit in J-inverted distribution. This happens because in this class the individuals are very size and density dependent (Farrior et al. 2016), meaning individuals are less competitive (Weiner 1990) and are more sensitive to natural disturbances, such as the fall of a big tree (Lieberman et al. 1985, Swaine et al. 1987, Appolinário et al. 2005, Guariguata and Ostertag 2001, Letcher 2010. Conversely, the increase in abundance of individuals > 10 cm in stand B and > 20 cm in stand C corroborate the maturing status of the tree component of the forest in PQERB (Swaine et al. 1987, Oldeman 1989, Phillips et al. 2004).

Dynamics of the most abundant species
Shade-tolerant species are able to develop under a closed canopy and require little sunlight. Conversely, the light-demanding species need higher sunlight incidence in order to develop and establish in a site. Both types of species occur in mature tropical forests in ongoing successional advance (Swaine and Whitmore 1988) and were more abundant compared to the pioneers, which were not among the most abundant ones in all inventories. One such example is Croton echinocarpus, which presented mostly bigger trees (> 10 cm) in the first two inventories. In a similar way, Myrsine umbellata was only among the most abundant species in the survey in 2000, probably as a result of the decreasing abundance of the individuals from 5-10 cm and similar abundance of individuals > 10 cm, in both intervals. During successional advance and in the exclusion stem phase (Letcher 2010), few trees survive the intensification of competition, caused by the biomass accumulation and higher crown of more competitive species, for light (Farrior et al. 2016). Smaller, less competitive trees (Coomes and R.B 2007) such as halophytic species with short life cycles die (Oldeman 1989, Chazdon et al. 2010).
The importance of light-demanding and shade-tolerant species in the community is demonstrated by their progressive increase in abundance and basal area. This confirms their important role in the formation of community structure and successional advance (Felfili 1995, Oliveira Filho et al. 2004. Stand B differs from stand C in that there was a larger number of light-demanding species and fewer shade-tolerant species. This finding possibly reflects the presence of more gaps in stand C than B, allowing halophytic species to remain in the forest interior (Lieberman et al. 1985, Pereira et al. 2006) as the forest develops towards the most advanced successional stages (Oldeman 1989, Ostertag 2001, Condit et al. 2002).

General conclusions on forest dynamics and conservation
The higher diversity in stand C, plus the diversity in stand B, indicates high alpha diversity within the PEQRB forest. This highlights the importance of the protection of biotic resources, and also supports the demand for further research to understand underlying determinants of this diversity. Despite the differences found in the structural dynamics within the studied forest, the basal area increased in both stands indicating biomass accumulation. This is a key factor in ecosystem services, such as the amplification of carbon stock through biomass gain. These two points show the importance of the protection of Atlantic Forests and that studies like this are important so that we may better understand the drivers of forest dynamics. Thus, the continued monitoring of this study site is necessary to further refine the mechanisms underlying tree dynamics.