Floristic and structural status of forests in permanent preservation areas of Moju river basin , Amazon region

The goal of this study is to analyze the floristic patterns and the structure of disturbed and undisturbed upland forests, in Permanent Preservation Areas (PPAs) along the Moju river, in the Brazilian state of Pará. Trees with a diameter equal to or larger than 10cm at 1.30m from the ground (DBH) ≥10cm were analyzed for the upper stratum. For the middle stratum, individuals with DBH between 4.99 and 9.99cm were sampled. Forty-five families and 221 species were found in disturbed forests, and 43 families and 208 species in undisturbed forests. Floristic similarity was high between strata and between forest types, with values above 50%. Similarity was highest between middle strata. The most species-abundant families in undisturbed forests were Fabaceae, Sapotaceae, Chrysobalanaceae and Myrtaceae; the species with the highest density there were Eschweilera grandiflora, Licania sclerophylla and Zygia cauliflora. In disturbed forests, the dominant families were Fabaceae, Sapotaceae, Lecythidaceae and Melastomataceae. The Shannon-Wiener diversity index was 3.21 for undisturbed forests and 2.85 for disturbed forests. Non-metric multidimensional scaling (MDS) analysis did not group the forests by their floristic composition in both upper and middle strata. Overall, the PPA forests along the Moju river, even if disturbed, did not show major floristic changes but substantially change their structural characteristics.


Introduction
Riparian forests in Amazonia are plant formations with a floristic composition and vegetation structure quite similar to those of upland forest areas, and are therefore considered as such.They are located along the edges of water courses and perform an important environmental role, mainly in the prevention of soil erosion and stilting of rivers, maintaining water quality and contributing to the preservation of biodiversity.Brazilian law considers them Areas of Permanent Preservation (PPAs, or APPs in Portuguese) (Brasil, 2002).
The concept of Permanent Preservation Areas in the Forest Law of 1965 emerged to ensure the integrity of the vegetation for maintaining the physical and biological equilibrium in Brazilian biomes.These are areas exclusively for preservation and cannot be utilized for agricultural or farming activities, forest extraction or recreational uses (Sparovek et al., 2011).Despite being protected by law, these forests are widely threatened in Brazil by deforestation and anthropogenic activities.In Moju municipality, Pará state, 29.3% of the PPAs were disturbed by land uses and the application of the new Brazilian Forest Code will result in the loss of ≈60% of forest vegetation from the PPAs in this county (Almeida and Vieira, 2014).The conservation role of PPAs presumably increases in highly deforested regions, such as the Belém Area of Endemism, that has reached at least 75% of the original extent and further extensive forest loss (Almeida and Vieira, 2010;Amaral et al., 2012).
It is therefore of foremost importance to develop studies describing the plant communities of those protected areas in order to provide a picture of the remaining biodiversity and its distribution in the landscape.Lack of studies on floristic composition, structure and conservation status of PPAs, allied to social and environmental problems in Amazonia have hindered the adoption of practices towards the restoration and conservation of those areas of permanent preservation.Within this context, this study aims at analyzing and comparing floristic diversity and structure of forests with different status of conservation in Permanent Preservation Areas in the municipality of Moju, state of Pará, and discuss their importance for the conservation and restoration of those areas.

Study area
The study was conducted in the municipality of Moju, eastern Pará, Brazil, in Permanent Preservation Areas of rural properties in the Moju river basin (01° 26' 31.7"-02° 24' 31.3"S and 048° 26' 54.2"-048° 59' 21.8"W).Local climate is of the type Ami, according to Köppen's classification (Nascimento and Homma, 1984).Average annual temperatures are high, varying between 25ºC and 27ºC.Annual rainfall is between 2000mm and 3000mm, irregularly distributed, being more concentrated (about 80%) from January to July; however, a short period of drought usually occurs between September and November.Relative air humidity is around 85% (Costa et al., 1998).Predominant soil types are Yellow Latosols of different textures, with Red-Yellow Podzols, Poorly Humic Gleysols and Plinthosols occurring as well (Santos et al., 1985;Costa et al., 1998).The original vegetation of this region of northeastern Pará, where the municipality of Moju is located, included extensive areas of ombrophilous dense forest (Rodrigues et al., 1997).The creation of the PA-150 highway led to intensive human occupation of the region with concomitant timber extraction and agricultural development such that by 2010 only 60% of its original vegetation cover remained, with much of this highly degraded (Almeida and Vieira, 2014).PPAs cover about 5% of the municipality of Moju, and are being subjected to strong pressures due to the expansion of agricultural activities and the cultivation of African oil palm (dendezeiro); some 28% of those PPAs are already in an illegal condition (Almeida and Vieira, 2014).
Location of PPA areas for this study was carried using satellite images followed by fieldwork.Area selection was conducted in the field, according to the following criteria: presence or absence of fire events and logging operations and availability of places authorised by community leaders in the municipality.
The selected areas were covered by upland ombrophilous dense forest (Pires, 1973), with a large number of 25-50m tall trees (Serrão et al., 2003) in unflooded terrain of the riparian zone of Moju river.Two types of forest were identified according to conservation status: disturbed forest and undisturbed forest.
Disturbed forests within PPAs, in illegal condition vis-à-vis the Brazilian Forest Law, were defined as a result of a process of degradation caused by a selective extraction of timber and regular occurrence of fire (Nepstad et al., 1999).Conversely, undisturbed forests within PPAs were defined by the presence of well-established forest strata and absence of visible disturbance, as the occurrence of stumps and fire scars.

Data collection and analysis
Data was collected in 2011 in plots belonging to small farmers in six rural communities in the municipality of Moju: Vila Olho D'água, Vila Soledade, Vila São Jorge, Vila São Tomé, Vila São Pedro, Vila Braulande (Figure 1).In each rural community, one area of undisturbed forest and one of disturbed forest were selected along the riparian strip of the Moju river.
Two plots of 250m x 10m (5000m 2 ), located at least 40m away from each other and positioned lengthwise along the river course, were established for each type of forest (disturbed and undisturbed).In total 12 plots per forest type were sampled, totaling 3ha.The upper stratum, with trees of DBH ≥ 10cm, was thoroughly surveyed.For the middle stratum, trees with DBH ≥ 5.0cm and ≤ 9.99cm were inventoried in five subplots of 25m x 10m (1250m 2 ), randomly distributed within each plot.Palm trees forming a stool were considered as a single individual, of which the DBH of a single stipe was measured.Plant species were identified under field conditions by experienced parabotanists (Nelson Rosa and Carlos Alberto Silva) from the Museu Paraense Emílio Goeldi (MPEG).Voucher samples of specimens which defied confident field identification were collected and taken for comparison with reference material at the MG herbarium.Our taxonomy follow Haston et al. (2009).
To generate aerial biomass estimates (as Mg.ha -1 ) we used allometric equations for classes with 5cm ≤ DBH < 20cm and DBH ≥ 20cm.Each tree was measured with an ordinary measuring tape in order to obtain CBH values (circumference at 1.30m from the ground), later converted into DBH.Total tree height was estimated in the field using a 5m stick.for DBH ≥ 20cm, where: P = weight of fresh mass (kg); D = diameter at 1.30m from the ground (DBH); H = total height.Student t-tests were used to compare diversity and biomass for undisturbed and disturbed forests, at a 5% significance level.Distributions of DBH for each type of forest were tested for similarity with a Kolmogorov-Smirnov two-sample test (Sokal and Rohlf, 1995).
Floristic diversity for the upper and middle forest strata was calculated by Shannon-Wiener Diversity Index (H') and evenness (J') was calculated using Pielou Index (Magurran, 2004).The Sørensen diversity index was used to compare similarity between the strata (Mueller-Dumbois and Ellenber, 1974).The software Mata Nativa (Cientec, 2002) was used for analysis.
A multidimensional scaling analysis (MDS) was performed with the Systat 12.0 program to detect possible dissimilarities between the two forest types, from matrices of abundance of tree species in the upper and middle strata.The Bray-Curtis distance method was adopted for the quantitative data matrix (abundance).A similarity analysis (ANOSIM) was used, using the Bray Curtis distance with 999 permutations, to test the significance of the groups generated in the ordination.The ANOSIM was run in the program PRIMER v 6 (Clarke and Gorley, 2006).

Results
Considering all sampled plots and the upper and middle strata of the two types of forests, a total of 4227 individuals belonging to 47 families and 270 species were identified.Undisturbed forests accounted for 2060 individuals of 43 families and 208 species, and disturbed forests for 2167 individuals of 45 families and 221 species.
In the upper stratum of undisturbed forests the family Fabaceae stood out in terms of species richness, as well as Lecythidaceae and Chrysobalanaceae, which also fared well in the upper stratum of disturbed forests.Fabaceae showed high values of species richness in any forest type.In the middle stratum, the families Fabaceae, Annonaceae and Melastomataceae made up 31.41% of its number of species.As for the environment where they occur, Fabaceae stood out in more preserved forests and again proved to be indifferent to disturbance.
Species exclusive to undisturbed forests added up to 49 (30 of them rare, that is, with a single individual).Conversely, 62 species were exclusive to disturbed forests, with a higher density of Cochlospermum orinocense (Kunth) Steud.
The upper and middle strata of undisturbed forests had higher values for richness and floristic diversity index than those of disturbed forests.However, considering both strata, undisturbed forests exhibited 2060 individuals, 43 families and 208 species, in contrast to, respectively, 2167, 45 and 221 in disturbed forests (Table 2).Comparing forest types, there is higher species richness and significance The similarity indices between the upper strata of undisturbed forests and disturbed forests, and between the middle strata of each type, were the highest found (Table 3); no index was lower than 50%, suggesting a certain floristic resemblance between forest types.
Considering both undisturbed and disturbed forests, the mean density of individuals in the upper stratum was 468 ind/ha, and the mean basal area, 25.98m 2 /ha.In the middle stratum, the mean density was 470 ind/ha and the mean basal area, 1.90m 2 /ha (Table 4).
Distribution of the number of individuals per diameter class tended to follow a negative exponential function (inverted J), with many small-diameter individuals and few large-diameter ones (Figure 2).Swartzia acuminata Willd.exVogel and Taralea oppositifolia Aubl.were the species with the highest number of individuals with diameters above 70cm in undisturbed forests; in disturbed forests, the species with the most numerous large-diameter individuals were Eschweilera coriacea (DC.)S.A. Mori and Bertholletia excelsa Bonpl.We observed that in the DBH ≤ 10cm class, the number of individuals per hectare was higher in disturbed (494.67)than in undisturbed forests (451.33);also, the 10cm ≤ DBH ≤ 20cm class exhibited the widest difference between the two forest types, with 225 individuals per hectare recorded in undisturbed forests and 298.67 individuals/ha in disturbed forests (Figure 2).Nº Ind. = number of individuals, H' = Shannon-Wiener Diversity Index, J' = Pielou Equability Index, Nf = number of families, Ng = number of genera, Ns = number of species.Those pattern of diameter distribution in both upper and middle strata did not present differences between forest types as showed by Kolmogorov-smirnov test (Ks max = 0.250, p= 0.906).
Aerial biomass was 353.35 Mg.ha -1 for undisturbed forests and 242.98 Mg.ha -1 for disturbed forests.Vila Braulande had the highest biomass values for both strata (392.20 Mg.ha -1 ), a major contribution for the high value found in undisturbed forests; this locality also had the highest mean biomass value (257.60 Mg.ha -1 ).Differences between forest types were statistically significant (t = 4.263 p <0.05) (Figure 3; Table 3).The mean basal area of the upper stratum of disturbed forests was smaller (20.95 m 2 /ha) than the value found for undisturbed forests (31.02 m 2 /ha).
The first and second axes of the NMDS in disturbed and undisturbed forest plots (stress value = 0.15 and 0.21 for upper and middle strata) did not group the forests by their floristic composition (Figure 4).This result was confirmed by ANOSIM analysis (R =0.026, p= 0,27, 999 permutations) which suggest that the disturbance observed in the forest plots is not interfering in species composition of tree communities studied for trees with dbh ≥ 10 cm and for individuals between 5 and 10 cm DBH.
However we observed the formation of two groups in the middle stratum of both disturbed and undisturbed plots (Figure 4b).The first group was formed by the localities of Vila Olho D'água, Vila São Tomé and Vila Braulande; the second by Vila Soledade, Vila São Jorge and Vila São Pedro.The distribution of species richness, abundance and basal area for each forest type of these rural villages were showed in Figure 5.

Discussion
The results in the present study revealed that the species richness and abundance of Permanent Preservation Areas forests of Moju river were comparable to other terra firme forests of eastern Amazonia (Salomão et al., 1988;Amaral et al., 2012).The most of the individual stems belong to a relatively small number of species and many of these species are hyperdominants (Ter Steege et al., 2013).Also the floristic richness observed in the upper and middle strata in the PPAs surveyed (208 species in undisturbed forests, 221 in disturbed ones) was considered high in comparison with other PPAs areas studied in the same region (Pinheiro et al., 2007), where 158 were sampled.
This study found higher species richness in disturbed than undisturbed forests (Table 3).Of the 270 species sampled in the two forest types of Moju, only 96 (36%) were common to both, 62 exclusive to disturbed forests in PPAs and 49 found only in undisturbed forests.The cause of such high richness should be attributed to the higher environmental heterogeneity resulting from disturbances in the uniform structure of the forest.In the case of the forests of Moju, it should be noted that, from the 270 species sampled, only 19 accounted for about 50% of the total number of individuals found: those are hyperdominant species, as defined by Ter Steege et al. (2013).In the forests near Moju, Rodrigues et al. (1997) found about 51% of individuals belonging to only nine species, a situation frequent in throughout the Amazonian forest region.
The family Fabaceae predominate in terms of species richness in both undisturbed and disturbed forests.It is highly frequent in the Neotropics (Ribeiro et al., 1999), and its richness in vegetation formations in Amazonia is mentioned by many authors who verified its wide geographic    distribution and high ecological plasticity (Pires, 1973).In areas dominated by upland forests in Amazonia, the families Sapotaceae, Moraceae and Burseraceae are also referred to as the most species-rich in other floristic inventories in the region (Pires, 1973;Prance et al., 1976;Dantas et al., 1980;Amaral et al., 2012).Other families found in the present study were Chrysobalanaceae and Lecythidaceae, which in Amazonia exhibit the highest densities of individuals and species, along with the Fabaceae (Oliveira and Amaral, 2004).
Overall, the disturbed forests surveyed had lower basal area values (20.95m 2 ) than those recorded for upland forests in Amazonia, which are of about 30-40 m 2 (Salomão et al., 1988;Pitman et al., 2001).The aboveground biomass was higher in undisturbed forests than in disturbed ones, which had undergone selective logging common practice in the Moju river basin.Vila Braulande had the highest biomass value (392.20 Mg.ha -1 ), evidencing its high degree of protection in comparison to other undisturbed forests, and especially to those classified as disturbed.That community collects only non-wood forest resources, such as oils and resins, which could have had an effect on this level of protection.In general, the response of the forests to disturbance is different in each rural communities (Figure 5).The mosaic of preserved and degraded forests present in an anthropogenic landscape such as these of Moju is the result of both the rural land use dynamics and the Forest Code, by requiring that 80-50% of the properties must be maintained as legal reserves (reserva legal) does not provide guidance as to its conservation status and format.
This study suggests that the impacts caused by humans on the Moju river communities did not lead to great floristic changes.Also, there were no significant differences between the analysed strata with respect to floristics (richness, Shannon-Wiener diversity, density).However, man-made impacts were perceptible in the tree community structure.As for the floristic composition, multivariate analysis evidenced there are no clear differences between forests respectively considered undisturbed and disturbed, which could be due to the different stages of preservation in PPAs or to their low level of exploitation by local communities -which have not seriously impacted the forest under their domain.

Conclusions
Floristic data on richness, abundance and tree species composition did not show any association with the conservation status of the PPAs, indicating that even the disturbed PPAs support viable populations of many tree species, therefore they should be maintained within rural properties.Moreover, the disturbance of forests within the PPAs seems to have been slight, not having significantly jeopardized the community of tree plants with respect to floristics.This favors the use of the most abundant species in the recovery of disturbed PPAs.The impact in the structure of the tree assembly was perceptible in such a way that disturbed forests exhibited a smaller basal area and biomass.

Figure 2 .
Figure 2. Number of individuals/ha distributed in different DBH classes in disturbed and undisturbed forests within PPAs in the municipality of Moju, Pará.

Figure 3 .
Figure 3. Student's t-test for mean biomass in undisturbed and disturbed forests within PPAs in the municipality of Moju, Pará.

Figure 5 .
Figure 5. Box-plot showing the distribution of richness, density and basal area of disturbed and undisturbed forests in PPAs of six rural communities studied in Moju, Pará.

Table 1 .
List of the 30 most abundant species, with density values (ind/ha) recorded in plots in the upper and middle strata in undisturbed and disturbed forests in Permanet Preservation Areas -PPAs within the municipality of Moju, Pará.

Table 2 .
Floristic information on tree species in undisturbed and disturbed forests within PPAs in the municipality of Moju, Pará.

Table 3 .
Sørensen Similarity Indices for undisturbed and disturbed riparian forests within PPAs in the municipality of Moju, Pará, considering two strata.