The effects of the creation of a hydroelectric dam on small mammals’ communities in central Brazil

Hydroelectrics are proposed as a solution to the increasing energy demand in developing countries. However, dams interrupt river flow; emit greenhouse gases, and cause habitat loss and fragmentation. Here, we evaluated changes in the diversity of small mammals during the construction of the Luis Eduardo Magalhaes hydroelectric dam in central Brazil. Two sites, both located above the maximum filling of the dam, were sampled before (2001) and after (2002) the flooding. We used wire mesh traps and sherman traps for the total capture effort of 5,940 trap-nights. We captured 88 individuals from 13 species – 5 marsupials and 7 rodents. We expected an increase in the abundance of small mammals immediately after the flooding due to the fleeing of individuals from the flooded areas, with a following increase in density dependent mortality. However, our results indicated lower values of abundance and/or richness of small mammals after the flooding. In addition, live traps were frequently found overturned and with signs of attacks from potential predators of small mammals. Results suggested that predation pressure probably increased after the flooding, thereby causing a reduction in the abundance and/or richness of small mammals. Key words: reservoir, sustainability, flooding, marsupials, rodents, Cerrado.


Introduction
Hydroelectrics are proposed as a solution for the increasing electricity demand in developing countries (Yüksel, 2010).In Brazil, ca.70% of electric power is produced by hydroelectrics (ANEEL, 2010), and energy policies remain focused on hydropower.Dams contribute to global warming with the emission of greenhouse gases (Fearnside, 1995(Fearnside, , 2002)), they affect the diversity of aquatic communities by obstructing the river flow (Poff et al., 1997;Suzuki et al., 2011), and cause habitat loss by the permanent flooding of natural patches (Rosenberg et al., 1997;Cunha and Ferreira, 2012).However, their effects on terrestrial wildlife have been poorly investigated.The creation of a dam can change biotic interactions near the newly created edge (Marinho-Filho et al., 2000).The fleeing of individuals from the flooded patches can immediately overrun some species densities in the adjacent habitat patches.This can affect the regulation of vertebrate populations since territorial behavior is commonly seen in several taxa (Burt, 1949;López-Sepulcre and Kokko, 2005;Rodrigues, 2006), including small mammals (Cerboncini et al., 2011).For instance, female small mammals are usually aggressive against co-specific intruders within the limits of their territory (Wolff, 1993).Consequently, the increase in density can result in a counter effect of increasing density dependent-mortality.This is particularly important for wildlife conservation and management, since population densities are dependent upon quality and size of the habitat patches (Chiarello, 2000).Changes in small mammal species densities are a well-documented effect of habitat loss and fragmentation in the Neotropics (Bonvicino et al., 2002;Pardini, 2004;Pardini et al., 2005;Passamani and Fernandez, 2011).While some species become abundant in some habitat patches, others become rare or locally extinct (Laurance, 1994).This density compensation effect can be caused as a consequence of increased competition between similar species (Pardini, 2004).It can also occur due to changes in predation rates, as a decreased abundance of carnivores in forest fragments usually leads to an increased abundance of their prey (Fonseca and Robinson, 1990).Few studies have reported the effects of the construction of hydroelectric dams on terrestrial animals (Nilsson and Dynesius, 1994).Regarding the Neotropics, this knowledge is limited to academic grey literature and private technical reports (Sá, 1995;Carmignotto, 1999).However, scientific data on this subject can provide valuable information for decision makers and help when deciding upon efficient measures for wildlife conservation.The aim of this study was to determine whether small mammal diversity changed after the construction of a hydroelectric dam in central Brazil.We expect an increase in the abundance of small mammals immediately after the flooding, due to the fleeing of individuals from the flooded areas.This increase in density should lead to an increase in density-dependent mortality reducing small mammal diversity.We also gathered information on attacks of live traps made by the predators of small mammals, in order to determine if changes in predation rates could be affecting the abundance of small mammals after the filling of the dam.

Study area
The Cerrado biome is exclusively within Central Brazil and its extension represents 25% of the Brazilian territory, with an area of approximately 1.5 million m 2 (Goodland, 1971).It extends specially through four Brazilian states (Eiten, 1972) making boundaries with the Amazonia to the north, Caatinga to the northeast, Pantanal to the northwest and Chaco to the south.Different types of plant formations can be found in the Cerrado biome -from open and shrub savannas to forest formations (Alho, 1982).It is one of the richest biomes on Earth and one of the world's most threatened hotspots (Myers et al., 2000).From the total of 711 different mammal species that occur in Brazil, 251 occur in the Cerrado (Paglia et al., 2012).The study sites are located in the central portion of the state of Tocantins, along the shores of the Tocantins River.Altitude in the region ranges from 200 m to 700 m and the climate is tropical.The rainy season extends from October to March and the dry season from April to September with an extremely dry climate between May and August.The main vegetation physiognomies that occur in the region are "cerradão" (a dense arborous savanna), "cerrado sensu strictu" (arborous open savanna), campo cerrado (shrub savanna with sparse trees) and gallery forests.The species of small mammals in this region are influenced by the biodiversity of other Brazilian biomes, such as Amazonia, Caatinga, Pantanal and Mata Atlântica, with a high diversity of small mammals (Passamani, 2004).Two sites -named Capivara and Jacinto -located near the margins of the Luís Eduardo Magalhães hydroelectric reservoir were sampled in this study.Vegetation in the Capivara site is typical of cerradão with gallery forests at a good level of conservation.The site is located near the Capivara streamone of the tributaries of the Tocantins river (UTM: 22L, 781192, 8877596).Vegetation in the Jacinto site is typical of cerrado sensu strictu (UTM: 22L, 785008, 8887216).Both sites were located above the maximum filling of the dam, which began in January 2002 and was finished 25 days later, flooding an area of 63,000 ha.

Sampling protocol
At each study site we surveyed for small mammals in a 0.56 ha grid with

Neotropical Biology and Conservation
The effects of the creation of a hydroelectric dam on small mammals' communities in central Brazil a total of 36 sampling points (6 x 6) placed at intervals of 15 m (Figure 1).Each sampling point received a wire mesh trap (42 x 21 x 21cm) or a sherman trap (29 x 13 x 13cm) which were alternately placed on the ground.In the Capivara site we also placed sherman traps in the medium strata (1 to 2m high) of each sampling point.The traps were baited with a mixture of bananas, peanut butter, cod-liver oil, and cassava flour or corn meal.The animals captured were marked with numbered ear tags and released back into the area of capture.Voucher specimens are housed in the Museu de História Natural do Centro Universitário Luterano de Palmas (CEULP/ ULBRA).A total of 10 sampling campaigns were conducted in each study site -four before the flooding (March, August, October and December 2001) and six afterwards (February, March, July, August, October and December 2002).Campaigns were bimonthly and each one composed of eight days of sampling.In the 5 th campaign carried out right after the flooding between February 22 and 30, 2002, we established a new sampling protocol near Jacinto's site.We placed six transects parallel to the dam's margin.This procedure was taken in order to evaluate if the flooding resulted in an accumulation of individuals near the margins of the dam.The positioning of the transects was changed daily, following the level of the reservoir.Ten traps were placed in 10m intervals in each transect.The animals captured in this sampling were not taken into account for data analysis.
During the sampling campaigns we looked for any signs of attacks to the live traps made by predators of small mammals.Medium to large size mammals -mainly carnivores, snakes and birds of prey are the main predators of small mammals (Palmuti et al., 2009;Silva-Pereira et al., 2011;Rocha et al., 2011).Disarmed traps, traps with marks or scratches, traps which had been rolled over, or even traps with small injured mammals captured inside were accounted in this sampling.

Data analyses
To test for differences in the richness and abundance of small mammals per sampling campaign per study site before and after the flooding, we used the Wilcoxon rank-sum test.The diversity of small mammals is variable throughout the year, but since we sampled periods distributed over two years -one before and the other after the flooding -differences in diversity should not be related to seasonality.To test for differences in the number of predator attacks to the live traps per campaign before and after the flooding, we used Wilcoxon rank-sum test (values reported are median, lowerupper quartiles).Similarities in species composition between the sampling campaigns for each study site were assessed by Jaccard's index and cluster dendrograms were constructed using Ward's agglomeration method.

Results
The capture efforts were made up of 3,060 trap-nights in the Capivara site and 2,880 trap-nights in the Jacinto site, making a total effort of 5,940 trap-nights.Total captures summed 141, with 94 captures in the Capivara site and 47 in the Jacinto site, which represents a capture success rate of 3.07% and 1.63%, respectively.During the third campaign, in May 2001, no captures occurred in Jacinto site -probably because of a fire that had partially burned the area two days prior to sampling.Thus, we decided to leave this sample out of our analysis.A total of 13 small mammal species and 88 individuals were captured during the study period (Table 1).Sixty two individuals were captured in Capivara site and 26 in Jacinto site.The abundance and richness of small mammals per sampling campaign decreased after the flooding in the Capivara site (Table 2; Figure 2).Abundance decreased but richness was not affected by the flooding in the Jacinto site (Table 2; Figure 3).Only one individual of Cerradomys subflavus (Wagner, 1842) was captured in the transects set along the margins of the reservoir during February 2002.In general, the sampling campaigns before the flooding were more similar to each other in species composition (Figure 4).Only the first campaign after the flooding in the Jacinto site clustered within the campaigns before the flooding in this site.Traps with signs of attacks were less frequent before (3.5%, 0-8.5%) than after the flooding (39.7%, 27.4-92%; W = 0; p = 0.01).In the two sampling campaigns after the flooding more than 90% of the traps showed signs of predator attacks (Figure 5).Foot-prints of the crab-eating fox Cerdocyon thous (Linnaeus, 1766), the crabeating raccoon Procyon cancrivorus (G.Cuvier, 1798), and ocelots Leopardus spp.were commonly seeing next to the overturned traps.We also captured five medium size mammals after the flooding -one crab-eating raccoon Procyon cancrivorus, one ring-tailed coati Nasua nasua (Linnaeus, 1766) and three capuchin monkeys Sapajus libidinosus (Spix, 1823).In the Capivara site two snakes Boa constrictor (Linnaeus, 1758) were found wound around traps with small mammals inside during the two campaigns that occurred immediately after the flooding.

Discussion
Small mammals are not able to travel large distances in the Cerrado (Nitikman and Mares, 1987;Vieira and Marinho-Filho, 1998) or to distant habitat patches after the creation of dams (Carmignotto, 1999), so we expected that the abundance of small mammals would increase immediately after the flooding.However, this pattern was not observed.Small mammal abundance decreased and remained low in the Capivara and Jacinto sites, and only one individual was captured during the sampling in the extra transects set parallel along the dam's margin immediately after the flooding in February 2002.These results may suggest that most small mammal individuals were not able to escape from the flooding.There were few signs of predator attacks to the live traps before the filling of the dam.On the other hand, the frequency of predator attacks to the live traps was high after the flooding.Small mammals are key components of the carnivore diet (Gatti et al., 2006;Moreno et al., 2006;Silva-Pereira et al., 2011) and footprints from carnivore species were frequently found near overturned traps after the flooding.These findings suggest that increased predation can be the responsible for the decrease in small mammal diversity after the flooding.In fact, hydroelectrics usually create land islands in which medium and larger mammals are absent.In these islands the richness and abundance of small mammals tend to increase immediately after the flooding (Marinho-Filho et al., 2000;Terborgh et al.,   2001).Nevertheless, in the study sites the densities of predators probably increased near the dam after flooding, resulting in a reduced abundance and in the local extinction of small mammal species.
A standard procedure taken during the construction of a hydroelectric is wildlife rescue (Vié, 1999).However, density compensation effects should be affecting vertebrate populations after the disturbances that cause habitat loss (Pardini, 2004;Pardini et al., 2005;Rodrigues, 2006;Passamani and Fernandez, 2011).Thus, wildlife rescue could result in harmful cascade effects in the adjacent habitat patches.In fact, if we consider our basic theoretical knowledge about animal behavior, territoriality and the regulation of natural populations, the actions currently taken according to governmental laws in Brazil are not logical when we consider the objectives of wildlife conservation (Rodrigues, 2006).Instead of contributing to the damage, we should prioritize actions such as increasing the protection and connectivity of nearby natural habitat patches.
Long term studies are necessary in order to evaluate how animal densities change during the several years after the building of a dam, and if these effects are permanent, resulting in some species being made locally extinct.Scientific data (Marinho-Filho et al., 2000;Terborgh et al., 2001) indicates that some populations tend to increase immediately after the flooding, changing intra and interspecific competition and causing density dependent mortality.Here we propose that predation can also affect wildlife after the flooding of a dam, just as it probably affected the abundance and richness of small mammals near the Luís Eduardo Magalhães hydroelectric.Actions planned by authorities to minimize the disturbances caused to wildlife by hydroelectric dams should consider the scientific data available to avoid unwanted effects and to act in accordance with wildlife conservation.Regarding at least small mammal species in areas to be flooded, we believe that the best procedure would be the capture, collecting and housing them as voucher specimens in scientific collections, as proposed by Rodrigues (2006).Our current knowledge of biodiversity is still insignificant, we can only estimate the number of species that there may be on the planet (Mora et al., 2011).Several taxa of small mammals were recently discovered in Brazil (Bonvicino et al., 2003;Voss et al., 2005;Weksler et al., 2006;Percequillo et al., 2011), and we believe that many rare or unknown species are being released during wildlife rescues.Understanding the components and the distribution of biodiversity can allow a more accurate identification of priority areas for conservation (Alho, 2000;Castro-Mello, 2003).The creation of conservation units is usually proposed as a compensatory action when a hydroelectric is built (McNeely, 1987).However, most of the dams flooded gallery forests in the Cerrado, which are regarded as the most diverse areas of the biome (Fonseca and Redford, 1984;Mares et al., 1986;Marinho-Filho et al., 1994;Johnson et al., 1999).Therefore, it is important that new areas composed of similar habitat type (gallery forest) are contemplated for the establishment of conservation units, as linear parks along rivers, for example.

Figure 1 .
Figure 1.Location of the study sites in Central Brazil (A), and the sampling protocol used for the capture of small mammals in two sites (Capivara and Jacinto) near the Luís Eduardo Magalhães hydroelectric dam (B).Each circle represents a sampling point at 15 m intervals in a 6 x 6 grid with 0.56 ha where a wire mesh trap or a sherman trap was placed alternately on the ground.In the Capivara site sherman traps were also placed on trees (1 m -2 m above ground) at each sampling point.

Figure 2 .
Figure 2. Abundance and richness of small mammals in the Capivara site per sampling campaign during the study period in the state of Tocantins, central Brazil.The dashed line represents the flooding of the Luís Eduardo Magalhães hydroelectric dam.

Figure 3 .
Figure 3. Abundance and richness of small mammals in the Jacinto site per sampling campaign during the study period in the state of Tocantins, central Brazil.The dashed line represents the flooding of the Luís Eduardo Magalhães hydroelectric dam.

Figure 4 .
Figure 4. Cluster dendrograms of the sampling campaigns regarding small mammal species composition in the Capivara and Jacinto study sites in Central Brazil.Outlined numbers represent the sampling campaigns that occurred before the flooding of the Luís Eduardo Magalhães hydroelectric dam.During the third campaign, no captures occurred in Jacinto site.

Figure 5 .
Figure 5. Percentage of live traps with signs of predator attacks per sampling campaign.The campaigns are ordered according to date, and the dashed line indicates the flooding of the Luís Eduardo Magalhães hydroelectric dam, in Tocantins, central Brazil.

Table 1 .
List of small mammal species and the number of individuals captured in the Capivara and Jacinto sites near the Luís Eduardo Magalhães hydroelectric dam inTocantins, Brazil.

Table 2 .
Comparisons of the abundance and richness of small mammals before and after the flooding of the Luís Eduardo Magalhães hydroelectric dam in Brazil.Reported values are median (lower-upper quartiles).
Note: (*) statistically significant values based on Wilcoxon rank-sum test.