Short-term impact of a hydroelectric... SHORT-TERM IMPACT OF A HYDROELECTRIC POWER PLANT’S RESERVOIR ON THE TREE COMPONENT IN AN ECOTONAL AREA IN SANTA CATARINA

1 Received on 03.03.2015 accepted for publication on 28.06.2017. 2 Universidade do Estado de Santa Catarina, Programa de Pós-Graduação em Engenharia Florestal, Lages,SantaCatarina,Brasil.Email:<carla_engflorestal@yahoo.com.br>,<amandaeng.f@gmail.com>,<juli_ajuri@hotmail.com>,<mariele.ferrer@hotmail.com>, <pamniederauer@gmail.com>,<vanessasoboleski@hotmail.com>,<rhodineli@hotmail.com>,<a.line_cruz@yahoo.com.br>, <karisouza@hotmail.com> , <chayanesouza8@gmail.com> and <jessica.oneda@hotmail.com.br>. 3 Universidade do Estado de Santa Catarina, Departamento em engenharia Florestal, Lages, Santa Catarina, Brasil. E-mail: <carol_sil4@yahoo.com.br> and <higuchip@gmail.com>. 4 Universidade do Estado de Santa Catarina, Programa de Pós-Graduação em Produção Vegetal, Santa Catarina, Brasil. Email: <karynaflorestal@yahoo.com.br>. *Corresponding author.


INTRODUCTION
The Atlantic Forest biome is known as a hotspot due to its great diversity of species and high proportion of endemic and endangered species (Myers et al., 2000).The state of Santa Catarina is located in this biome, with Ombrophilous formation at east and Deciduous formation at west.The area under study is located in a transition region between the Mixed Ombrophilous Forest (MOF) and the Deciduous Seasonal Forest (DSF), in the Upper Uruguay River basin, with high hydroelectric potential.Hydroelectric power plants (HPP) have been therefore built along its rivers, suppressing vegetation and reducing forest areas.
Their construction mainly affects riparian forests by causing the reduction and, sometimes, the disappearance of plant and animal species (Fearnside, 2001).By elevating the river's level and establishing the reservoir, new riparian forests are formed due to the water level fluctuation that causes a higher soil waterlogging of part of the vegetation that used to be distant from the water course, resulting in the selection of species adapted to this new environmental condition.These reservoir riparian environments are unique places ranging from mesic sites free from flooding to depletion areas where plants are partially or totally submerged during flood period (Davide et al., 1996).The seasonal change of aquatic/terrestrial habitats along the year creates stress over the plants, demanding them to adapt for surviving to long periods of submersion (Junk, 1989).Changing the species potentially able to colonize the reservoir banks afterwards might cause significant changes in the floristic-structural composition of remnant fragments (Ferreira et al., 2013).
This study aimed to identify the impact of filling the reservoir of the Garibaldi hydroelectric power plant on the floristic-structural composition of the riparian forest adjacent to the lake banks one year after the filling.Three sectors have been defined for the forest: the first one near the lake banks and the other two along the slope.As a hypothesis, the forest in the sector adjacent to the lake banks was expected to show higher tree mortality than the other two sectors since it is more influenced by water flooding.In addition, the highest number of dead trees on the reservoir banks was expected to result in the floristic-structural variation among sectors, a process that would represent the beginning of species substitution with the exit of those not tolerant to water stress.

MATERIAL AND METHODS
The study was carried out in a forest remnant located in the municipality of São José do Cerrito, SC, Brazil, in an area of the Garibaldi hydroelectric power plant reservoir built in 2013.It is placed in the Canoas River basin at 27°55'73'' South latitude and 50°95'60'' West longitude and from 700 to 800 m altitude with wavy to rugged topography.
According to Köppen's classification, it has a Cfb regional climate: mesothermic humid with a mild summer.Its mean annual temperature ranges from 15.8 to 17.9ºC with annual rainfall ranging from 1,460 to 1,820 mm (EPAGRI, 1999).IBGE (2012) classifies it as Mixed Ombrophilous Forest (MOF) in transition with Deciduous Seasonal Forest (DSF).The area is in advanced Short-term impact of a hydroelectric... successional stage, showing great specimens of the region's flora.Its current state of conservation is a result of the low exploitation of timber species in the forest remnant opted by its owner in the past and of the recently acquisition of part of the area by the company responsible for the Garibaldi hydroelectric power plant to compose the Permanent Preservation Area (PPA) of the lake.It is mainly formed by haplic Nitosols and humic Cambisols soils developed from basaltic rocks (Embrapa, 2004).
By August, 2014, one year after the reservoir filling, 48 200 m² (10x20m) permanent plots were systematically allocated 20 m apart from each other and 30 m from sector to sector (Figure 1): adjacent to the reservoir banks or lower sector (S1), in the middle of the slope or intermediate sector (S2) and at the top of the slope or upper sector (S3).All tree individuals with circumference at breast height (CBH) > 15.7cm were then measured, tagged and identified.Standing dead individuals were also measured and quantified.Identification was held by comparing herbaria and specialized literature; families were classified according to the APG III system (2009).Spatial variables corresponding to the x and y coordinates of each plot's center were obtained through GPS.
Sampling sufficiency was analyzed by the species accumulation curve through the randomization method with 1,000 permutations.Diversity and evenness were verified by Shannon and Pielou indexes, respectively (Brower and Zar, 1984).Tree component floristic-structural composition was evaluated according to the classic phytosociological estimates (Mueller-Dombois and Ellenberg, 1974).
The graphical representation of the variation of the number of dead individuals per plot in each sector was carried out by a box-plot graph.In order to verify if there was significant difference in the number of dead trees among sectors, a Generalized Linear Model (GLM) was applied with the Poisson distribution by considering each sector as a categorical explanatory variable.Because data variance is higher than the mean therefore resulting in overdispersion, standard errors were corrected through a quasi-GLM model, where variance is given by  × , where  is the mean and  is the parameter of dispersion (Zuur et al., 2009).The Mantel test was applied to determine whether or not the floristic-structural organization had spatial structure.The community's floristic-structural ordination was performed through the Non-Metric Multidimensional Scaling (NMDS).Generalized linear models were adjusted according to the methodology proposed by Wang et al. (2012) with species abundances as dependent variables and sectors as independent variables to verify the existence of floristic-structural variations among sectors.The negative binomial distribution was considered for the models since many species have just a few individuals.The species matrix of abundance per plot underwent a Hellinger transformation for reducing asymmetry among the most abundant and rare species (Legendre and Legendre, 2012).The suitability of all adjusted models was verified through the graphic analysis of the residues in order to verify the normality and heteroskedasticity of their distributions.
All analyzes were performed with the R statistical programming language (R Development Core Team, 2014) by using the vegan package (Oksanen et al., 2014) for the species accumulation curve, Mantel test and NMDS; and the mvabund package (Wang et al., 2014) for adjusting the generalized linear models.
According to the species accumulation curve, species richness has increased only 1.24% by including the last five plots (10% of the sampled area).Diversity and equability have reached the following values: H'= 3.10 and J' = 0.73.
We highlight the high density (152.08 ind.ha -1 ), dominance (3,772 m².ha -1 ) and absolute frequency (77.08%) of dead individuals, therefore ranking second in IV.The dead trees number was relatively high (146 individuals) with the highest value in the lower sector (81), which is closer to the reservoir and, consequently, to a soil more soaked, followed by the intermediate (39) and upper (26) sectors.The numerical variation of the dead trees among the three sectors (Figure 2) has resulted in a significant difference (p = 0.002) with the lower sector differing from the other two (p = 0.024 in comparison with the intermediate sector and p = 0.003 in comparison with the upper sector).
The Mantel test has resulted in r = 0.2997 (p = 0.001) thus indicating spatial structuring in the community organization.The NMDS ordering diagram (Figure 3) has indicated no variation in the floristic-structural composition among the three sectors, which has been confirmed by the mvabund function (p = 0,371).

DISCUSSION
The basal area value we have found ( 26  Plots from 1 to 16 are in the lower sector, from 17 to 32 in the intermediate sector and from 33 to 48 in the upper sector.Figura 1 -Localização das unidades amostrais no trecho de floresta ciliar avaliado em São José do Cerrito, SC.Parcelas de 1 a 16 estão no setor inferior, de 17 a 32 no setor intermediário e de 33 a 48 no setor superior. Short-term impact of a hydroelectric...The richest families we have registered are in agreement with other studies carried out in the region under study, such as Herrera et al. (2009) and Klauberg et al. (2010), where the highest richness have been observed in the families Myrtaceae and Lauraceae.The majority of rich families found by Souza et al. (2015) in a transitional area, excepting Rutaceae, have shown a close number: Myrtaceae (11), Rutaceae (6), Fabaceae (5), Lauraceae (4), and Sapindaceae (4).Results show that the floristics of the area is as expected for such region despite structural changes.
Species accumulation curve demonstrated that sampling effort was sufficient for the floristic characterization since it tended to stability and species have increased less than 5% after 10% increase in sampling area.According to the criteria of Kersten Regarding the Shannon index diversity value we have found (H'= 3.10), diversity can be considered from intermediate to high when compared to values found in other studies.According to Jarenkow and Budke (2009), several studies carried out in MOF areas from the southern and southeastern Brazil have indicated H' estimates ranging from 2.25 to 3.54 nats.ind - .Souza et al. (2015) have found a lower value (H'= 2.78) in a MOF and DSF transitional area.As for the result obtained for the Pielou evenness index (J = 0.73), one can verify that the species ecological dominance is relatively intermediate.Jarenkow and Waechter (2001) have obtained a lower evenness value (0.56) in DSF therefore indicating high ecological dominance.In a transitional area, Souza et al. (2015) have obtained a similar value to the one we have found (0.67).
Allophylus edulis has stood out in the community, which may be mainly explained by the species' high density and frequency.However, the group of dead individuals and the large-sized Nectandra megapotamica and Araucaria angustifolia have shown the highest dominance, with a high basal area.The high importance of few species in contrast with the high occurrence of rare species is a common pattern in natural forests, as can be observed in other studies carried out in this region, such as the study in MOF areas by Silva et al. (2012) and the study in DSF areas by Jarenkow and Waechter (2001).
Considering that riparian areas are particularly sensitive to changes in the hydrological cycle being considered good indicators of environmental changes caused by dam reservoirs (Nilsson and Bergger, 2000), the high mortality of individuals, especially in the lower sector, is probably associated with its proximity to the dam water reservoir.These contact areas are exposed to new conditions of soil flooding with low availability of oxygen to the roots, therefore generating stress to non-adapted tree species (Nilsson and Bergger, 2000).Thus, it is inferred that one year after filling the reservoir it is possible to observe its interference in the community structure through the higher mortality in the lower sector.
As far as the spatial structuring of community organization is concerned, the plots spatially closer are also more similar from the floristic-structural point of view.However, such spatial structuring could not be associated with the reservoir distance sectors since the NMDS and the mvabund function have found no floristic-structural variations.There are possibly other ecological conditions that cause this spatial dependence pattern.
Thus, one can verify that structural changes have begun to occur one year after the reservoir filling and have been caused by variations in mortality among sectors.Yet there was not enough time for floristic changes to occur as they require a process over mortality time, which has already begun, along with the colonization of new species adapted to the new environmental conditions.Such results are not in agreement with a similar study (Souza et al., 2015) carried out on the banks of a hydroelectric power plant's reservoir in an ecotonal forest area between MOF and DSF where variation in the floristic-structural composition along a topossequence adjacent to the reservoir has been observed.This differentiated pattern noticed by Souza et al. (2015) is probably due to a longer filling time of the reservoir, since their inventory was carried out eight years after reservoir filling.

CONCLUSION
It is worth noting the high representation of dead individuals, especially in the lower sector closer to the dam bed.This condition suggests an impact on the forest structure under the direct influence of the reservoir.There was no variation of the floristic-structural composition among sectors, though.Such condition may be due to the short period of time to the reservoir filling, where the environmental filter caused by the impact has not induced the expected changes therefore making it necessary to continue monitoring the vegetation dynamics over time in order to gauge possible changes in the community.

ACKNOWLEDGEMENTS
We thank to FAPESC for granting a scholarship to the first author of this paper and for the financial support granted through the PAP/UDESC; to CNPq for providing a research productivity grant to both the second and third authors; to Rio Canoas Energia for making the area available for study and for providing logistical support.We also thank the support offered by the landowners Marcos e Raquel Albuquerque.
.828 m².ha -1 excluding dead individuals) is below from what is expected in this region for forest areas in advanced successional stages.Such value is lower than what have been observed both in MOF areas by Herrera et al. (2009) (31.4 m².ha -1 ), Higuchi et al. (2012) (36.45 m².ha-1), Silva et al. (2012) (35.54 m².ha -1 ) and Higuchi et al. (2013) (34.8 m².ha -1 ) and in MOF and DSF transition areas by Souza et al. (2015) (32.32 m².ha -1 ).Results thus suggest the impact of the hydroelectric power plant's lake on the forest structure.Species and families richness registered for the area under study can be considered as intermediate in relation to other studies carried out in the same region.Jarenkow and Budke (2009) have compiled and analyzed the floristic and structural patterns of studies related to 38 areas carried out by several authors in the Mixed

Figure 1 -
Figure 1 -Location of sampling units in the riparian forest area in the municipality of São José do Cerrito, SC, Brazil.Plots from 1 to 16 are in the lower sector, from 17 to 32 in the intermediate sector and from 33 to 48 in the upper sector.Figura 1 -Localização das unidades amostrais no trecho de floresta ciliar avaliado em São José do Cerrito, SC.Parcelas de 1 a 16 estão no setor inferior, de 17 a 32 no setor intermediário e de 33 a 48 no setor superior.

Figure 2 -
Figure 2 -Variation in the number of dead individuals per plots in each sector (Lower, Intermediate and Upper) in an ecotonal area between Mixed Ombrophilous Forest and Deciduous Seasonal Forest in the municipality of São José do Cerrito, SC, Brazil.Figura 2 -Variação do número de indivíduos mortos por parcela de cada setor (Inferior, Intermediário e Superior) em uma área ecotonal de Floresta Ombrófila Mista e Floresta Estacional Decidual, São José do Cerrito, SC.

Table 1 -
Floristic-Structural composition of a forest remnant from an ecotonal area between Mixed Ombrophilous Forest and Deciduous Seasonal Forest, in the municipality of São José do Cerrito, SC, Brazil, ranked by importance value (IV, %) [AD= absolute density (ind.
Short-term impact of a hydroelectric...