14. Patterns of Leaf-litter Amphibian Diversity in a Silvicultural Landscape of Southeastern Brazil

Amphibians are susceptible to environmental changes due to a combination of morphological, physiological and behavioral characteristics adapted to specific environmental conditions. Few studies focus on the patterns of amphibian diversity in disturbed habitats by human action. However, recent studies with other taxa suggest that agricultural landscapes can support considerable biodiversity. The main goal of this study was to evaluate the spatial-temporal distribution and abundance patterns of leaf-litter amphibians inhabiting a silvicultural landscape. This study was conducted at Três Lagoas and Arca farms, located in the Upper Paranapanema basin, state of São Paulo, in southeastern Brazil. The data collection was carried out from August 2007 to July 2009 in monthly sampling campaigns. A grid of sampling units (n = 30) formed by pitfall traps were distributed over Eucalyptus plantations (n = 18), native vegetation (n = 7) and abandoned pastures (n = 5). A total of 1047 leaf-litter individuals from 18 species were captured in the study area. The leaf-litter amphibian assemblage was composed predominantly by Physalaemus cuvieri (58.6%), Physalaemus nattereri (13.1%), and P. marmoratus (13.0%). There was a significant difference in species richness and abundance between environments, surprisingly with the lowest species richness and abundance in the native vegetation. However, there was no relationship between species richness and relative abundance considering the distance of water bodies or the native vegetation. The results of this study suggest that silvicultural landscapes may have some conservation value. In order to improve it, it is necessary to maintain conservation areas within the landscape, as well as to stimulate innovation in silvicultural wildlife-friendly management practices.


Introduction
The expansion of socioeconomic activities has proven to be a major driver of the alteration and destruction of natural habitats (Rocha et al., 2006). This process caused extensive impacts on biodiversity in different regions of the world due to the replacement of native vegetation by agricultural crops and livestock areas (Diamond, Introduction frequency resulting in population decline (Feder & Burggren, 1992;Young et al., 2000).
The socioeconomic development of the state of São Paulo in Southeastern Brazil reflects the global trend of natural resources utilization. The economic development model adopted by the state caused profound changes in its original vegetation, gradually replaced by agroecosystems (Victor et al., 2005). Such process resulted in the current critical environmental scenario, with only 17.5% of the native vegetation remaining (IF, 2009).
The state of São Paulo has approximately 1.2 Mha covered by forest plantations that correspond to the second largest in Brazil (ABRAF, 2013). Most of this area was formerly occupied by low productivity pastures (Vianna et al., 2007). In turn, environmental laws due to the relatively well established certification process (i.e. Forest Stweardship Council -FSC) in this sector provides the implementation of Areas of Permanent Protection (APP) and Legal Reserve (LR). Such change on the landscape structure can have direct and indirect consequences on local biodiversity.
A priori, amphibians can be considered good indicators of environmental change as they generally have relatively small home range and most depend on the availability of aquatic environments for reproduction and terrestrial environments for foraging, aestivation, migration and dispersion (Stebbins & Cohen, 1995;Zug et al., 2001;Pough et al., 2004). Habitat fragmentation, deforestation and water contamination by agrochemicals have been considered the main causes of global decline of amphibians (Gray et al., 2004;Young et al., 2004). Many species are also sensitive to changes in the vegetation structure near water bodies (Jim, 1980;Valan, 2002;Renken et al., 2004) and their dispersal patterns are intrinsically related to habitat quality (Duelman & Trueb, 1994;Peltzer et al., 2003). However, tolerant amphibians have been recorded in anthropic environments (e.g. Demaynadier & Hunter, 1998;Campos & Vaz-Silva, 2010;Machado et al., 2011).
The main goal of this study was to evaluate spatial-temporal variation in diversity of leaf-litter amphibians inhabiting a silvicultural landscape during the first three years (0-3 year) of Eucalyptus plantations. The following null-hypotheses have been tested: 1) There is no difference in leaf-litter amphibians diversity among the different components of silvicultural landscapes (Eucalyptus plantations, abandoned pastures and native vegetation remnants); 2) There is no relation between leaf-litter amphibians diversity and their distance to the nearest border of native vegetation inside the matrix (Eucalyptus plantations); 3) There is no time variation in leaf-litter amphibians diversity in silvicultural landscapes; and, 4) There is no intraspecific difference in body length in leaf-litter amphibians in relation to the silvicultural landscape components.

Sampling Methodology
The study design is based on the methodology proposed by the Biodiversity Research Program (PPBio) where sampling units form a grid with nodules distant one kilometer from each other. This methodology is appropriate for long-term ecological research because it allows rapid biota inventories (Magnusson et al., 2005). In the present study, 30 sampling units (located in the grid nodules) were implemented taking into account the proportion of the three landscape elements: Eucalyptus plantations (n = 18), native vegetation (n = 7) and abandoned pasture (n = 5).
The amphibian captures were conducted with drift fence pitfall traps (Corn, 1994;Block et al., 1998;Cechin & Martins, 2000;Dixo & Verdade, 2006;Bernarde & Macedo, 2008). Each sample unit was composed by four 100 L buckets buried to the surface of the ground connected by guide fence (plastic net) with 80 cm high (10 cm buried) in shape of "Y" with 15 m long segments. Buckets had small holes for rainwater drainage and remained capped while not in use. Small mammal studies were also carried out simultaneously (Martin et al., 2012;Rosalino et al., 2013).
Data collection occurred in 23 monthly campaigns from August 2007 to July 2009 (skipping November 2008 due to logistical impediments). The buckets remained open for two consecutive nights in each campaign and were checked every morning. Captured animals were identified, measured (body length), sexed and released at the capture site. Vouchers of amphibian species were deposited in the collection "Célio F. B. Haddad" (CFBH) of Instituto de Biociências -Universidade Estadual Paulista -Campus Rio Claro, under appropriate license from the Instituto Chico Mendes de Conservação da Biodiversidade -ICMBio (License n° 12623-1 and 12623-2). Species identification followed the Brazilian Society of Herpetology (Segalla et al., 2012).

Analytical Methodology
The sampling effort sufficiency to characterize the leaf-litter amphibian diversity was evaluated for each landscape unit and for the whole study area by species incidence curves (adapted from Colwell & Coddington, 1994) where the time-series sigmoid model bootstrapped (1000 randomizations by EstimateS software -Colwell, 2004) asymptote was considered the estimated number of species. The sampling effort (SE) of pitfall traps methodology was calculated according to the formula: where C is the number of sampling campaigns, D the number of sampling days, U the number of sampling units and PF the number of pit-fall stations. The capture success (CS) was calculated according to the formula: CS = (T / SE) x 100 where T is the total number of individuals captured and SE is the sampling effort. Abundance index (AI) was calculated according to the formula: Prior to statistical analysis we tested data distribution and homoscedasticity, respectively by Anderson-Darling and Levene's tests. One-way analysis of variance (ANOVA) (Zar, 2010) and Kruskal-Wallis test (Siegel, 1956) were used to respectively test whether leaf-litter amphibian species richness and relative abundance differ among the three distinct landscape units (Eucalyptus plantations, abandoned pastures and native vegetation). Analysis of means (ANOM) (Ott, 1983) was then applied as a post-hoc test. In addition, Correspondence Analysis (Manly, 1994) was used to verify the possible relation between the landscape unit type and species composition (Hypothesis 1). Linear regressions (Zar, 2010) were used to detect the possible relationship between species richness and relative abundance of leaf-litter amphibians and the distance of the sampling units to the nearest water body and native vegetation using 14 sampling units located within Eucalyptus plantations of Três Lagoas Farm (Hypothesis 2).
The temporal variation in species richness and overall relative abundance was evaluated by t-test (Zar, 2010) and Mann-Whitney test (Siegel, 1956), respectively (Hypothesis 3). April through September was considered dry season whereas October through March was considered wet season. In addition, temporal variation along this study period is presented. At last, one-way ANOVA (Zar, 2010) were applied to check possible differences in snout-vent length (SVL) of dominant species among environments (Hypothesis 4). All analyzes were performed in the statistical software Minitab 16.

Results
Eighteen leaf-litter amphibian species were detected in the study area, totaling 1047 captured individuals. These species are distributed in the families Bufonidae, Craugastoridae, Cycloramphidae, Leiuperidae, Leptodactylidae and Microhylidae (Tab. 14.1). The total sampling effort was 4977 trap-night (Eucalyptus plantations = 3054 trap-night; native vegetation = 1030 trap-night; abandoned pasture = 893 trapnight) and the overall capture success was 21.05%. Our sampling effort was enough to detect from 83.3 to 93.5% of the estimated leaf-litter amphibian species richness of this study area (Fig. 14.1, Tab. 14.2).

Discussion
Although we carried out monthly sampling campaigns during two years the number of species detected in each sampled environment (from 83.3 to 93.5%) suggest that more intense sampling effort (possibly longer than only 2 night captures per month) should be applied in silvicultural landscapes. However, for logistic reasons, in order to do so it might be necessary to restrict the number of sampling units per sampled environment. Future studies should prioritize sampling design and sampling methodology concerning leaf-litter amphibians in silvicultural (and agricultural) landscapes. Such concerns have been raised for mammals (Lyra-Jorge et al., 2014) and birds (Penteado et al., 2014) as traditional methods for vertebrate surveys have been generally developed for pristine environments from temperate regions (Sutherland, 2006).
The pattern of variation in species richness and relative abundance of leaf-litter amphibians among environments found in this study is possibly due to the expansion of the dominant species in anthropic areas at the beginning of the Eucalyptus cycle (first three years of implantation), possibly in response to the sudden increase in food supply and lack of predators. This hypothesis is corroborated when we reanalyze the data without the dominant species with consequently no difference in species richness and relative abundance among environments (richness: F 2,29 = 2.02; p = 0.153; abundance: H 2,29 = 2.34; p = 0.311). The conversion of pasture on Eucalypts plantations promotes the appearance of an arboreal extract, an increase on the litter amount, a decrease in soil compaction and a decrease in light incidence . Such factors can be beneficial to some species of amphibians. Future studies should investigate whether such patterns persist along the productive cycles of the Eucalypus plantations (i.e. during a few decades).
The species composition of leaf-litter amphibians appears to be different between the native vegetation and the anthropic environments, with no distinction in this study between Eucalyptus plantations and abandoned pastures. Such patterns could be due to the fact that the initial phase of Eucalyptus plantations has structural characteristics intermediate between open and forest environment. In such a circumstance, Cerrado species like Physalaemus cuvieri, P. centralis, P. marmoratus and Physalaemus nattereri can be benefitted (Bastos, 2007).
Surprisingly, native vegetation showed just few exclusive species in relation to the anthropic environments. This pattern may be the historical result of local land use and exploitation of natural resources (e.g. deforestation). The study area is located in a transition between Cerrado and Semideciduous forest that suffered cycles of deforestation and revegetation since 1870 (Lisboa, 2008). Since the early 1970's, land use has been dominated by pastures for livestock production which possibly resulted in local extinction of many forest species of leaf-litter amphibians. Different species may be distinctly affected by land use changes (Swihart et al., 2003;Rubbo & Kiezecker, 2005). Revegetation of conservation areas (i.e. APP and LR) might result in an increase in leaf-litter amphibian diversity. A network of long-term biodiversity monitoring sites should be established in agricultural/silvicultural landscapes in order to evaluate the impacts of changes on land use and agricultural management practices on biodiversity (Verdade et al., 2014a) including leaf-litter amphibians.
The distance from water bodies or native vegetation can be decisive for amphibian survival. However, no such pattern has been detected in this study, which can be partially explained by the presence of small fragments of native vegetation of irregular shape with relatively small species richness possibly due to border effect (i.e., increase in temperature and decrease in relative humidity). In addition, in this study we considered only permanent water bodies due to the logistical difficulties to monitor temporary water bodies in large areas long term. However, the distribution of temporary water bodies may have affected the present results, since most leptodactylids of the Physalaemus group use temporary ponds and puddles for reproduction (Brasileiro et al., 2005). Such temporary water bodies can increase the carrying capacity of Eucalyptus plantations for leaf-litter amphibian at least in the early phase of its first productive cycle.
The species richness and overall relative abundance varied between dry and wet seasons. The decrease in abundance during the dry season may be related to lower activity of amphibians during this period due to decrease in temperature, humidity conditions and food supply (Gibbs, 1998;Alford & Richards, 1999;Pinheiro et al., 2002;Vasconcelos & Rosa-Feres, 2005). Humidity and/or temperature have been considered the main determining factors of amphibian diversity (Aichinger, 1987;Maffei et al., 2011). Most amphibian species which inhabit regions without marked seasonality can reproduce throughout the year; however, Cerrado species tend to concentrate their reproductive activities during rainy season (Brasileiro et al., 2005) as they depend on temporary aquatic environments to breed (Duellman & Trueb, 1986). The temporal distribution of the species in this study seems to be consistent with this pattern. Body size is related to age, gender, phylogeny, and environment (Calder, 1996;Morrison & Hero, 2003;Bidau et al., 2011). Habitat-related body-size variation can be an important indicator of a species' ability to adapt and survive to environmental change (Rosalino et al., 2013). As an example, the reduction in blacksnake head size has been associated with the consumption of smaller prey (Phillips & Shine, 2004). The relationship between snout-vent length and the environment found in this study for P. cuvieri may be due to a possible increase in environmental quality from the abandoned pasture to the native vegetation and the Eucalyptus plantation. However, this pattern can possibly be affected by silvicultural management practices like chemical weed control and mechanical harvest. Future studies should investigate the availability of food resources in the distinct environments of silvicultural landscapes along productive cycles of Eucalyptus plantations.
The results of this study suggest that silvicultural landscapes may have some conservation value, as long as natural water bodies are maintained along with conservation areas (i.e. APP/LR). In addition, the development of wildlife-friendly silvicultural management practices (i.e. the maintenance of sparse native trees inside Eucalyptus plantations and a relaxation in chemical weed control, as suggested by Athayde (2013) and Millan (2013), might possibly increase β-diversity in silvicultural landscapes (Verdade et al., 2014b). Such information should be incorporated in the certification process of Eucalyptus production (e.g. Forest Stewardship Council) in order to improve the conservation value of silvicultural landscapes.