The Reticulated Flatwoods Salamander, Ambystoma bishopi, is an inhabitant of longleaf pine forests in the southeastern United States. Historically distributed across southern Alabama, Georgia, and the Florida panhandle west of the Apalachicola-Flint Rivers, the range of this species has been drastically reduced. It is currently listed as federally endangered under the Endangered Species Act (ESA). Population viability analyses (PVAs) represent a key component of many recovery plans for threatened and endangered species. Here we use 10 years of mark recapture data collected from two breeding populations of A. bishopi to construct a demographic model that can be used to evaluate future extinction risk. In chapter one, we quantify population sizes through time, and estimate the impact of annual variability in numbers on genetic viability. This species exists in small (< 500) breeding populations and exhibit annual fluctuations in abundance characteristic of pond-breeding amphibians. In chapter 2, we adopt a modified version of the von Bertalanffy equation to construct size-at-age curves for A. bishopi that include the metamorphic transition. Individuals exhibit rapid growth in the larval stage such that they emerge as metamorphs at 60% of their final body size. In chapter 3, we employ a Cormack-Jolly-Seber model, modified to include continuous covariates, to generate size-dependent survival curves. Survival of A bishopi exhibits dramatic annual and seasonal variability, but is always positively correlated with body size. Lastly, in chapter 4, we combine the elements of all previous chapters to construct an Integral Projection Model (IPM). Given the prevalence of complete recruitment failure in these populations, and their relatively small size, extinction probabilities under a business-as-usual model were high. Increasing the frequency of successful recruitment drastically reduces extinction risk; however, adult survival exerts the greatest influence on long-term population growth. To assure the recovery of A. bishopi, management must consider all elements of the life-history when allocating resources and effort. More generally, both aquatic and terrestrial habitats must be protected for amphibian conservation to be effective, making them ideal candidates for 'umbrella species' status. Amphibian conservation would also benefit from an increase in systematic, long-term data collection. / Doctor of Philosophy / The southeastern United States is the global salamander hotspot, representing a crucible for diversity. Longleaf pine forests, the predominant ecosystem in the southeast, have been reduced to 3% of their former range, with dire consequences for the animals that inhabit them. The Reticulated Flatwoods Salamander, Ambystoma bishopi, is endemic to the region, and currently listed as federally endangered owing to recent population declines. A recovery plan for the species therefore, is required by law, under the Endangered Species Act (ESA). A salient component of modern recovery plans are population forecasts that evaluate future extinction risk. Such forecasts can then be used to assess alternative management strategies proposed to improve the species' long-term prospects. By studying two of the last remaining populations of A. bishopi from 2010-2019, we were able to collect the data required to construct a demographic model that can be used to run population projections. In some regards, A. bishopi is a typical amphibian, in that their populations show dramatic fluctuations in numbers through time, and they exhibit rapid growth in the aquatic larval stage, achieving 60% of their maximum body size in the first three months of life. Flatwoods salamanders breed in ephemeral wetlands, that often dry before successful metamorphosis can occur. The frequency of pond-drying results in a high probability of extinction for a single population, but survival of breeding individuals was equally important when considering long-term persistence. To assure the recovery of A. bishopi, management must consider all elements of the life-history when allocating resources and effort. More generally, both aquatic and terrestrial habitats must be protected for amphibian conservation to be effective, making them ideal candidates for 'umbrella species' status.

Made available in DSpace on 2016-06-02T20:21:26Z (GMT). No. of bitstreams: 1 3856.pdf: 2448632 bytes, checksum: 1a20c453c771b7c694d6da5ae1f87f86 (MD5) Previous issue date: 2011-05-27 / Financiadora de Estudos e Projetos / Cerrado is one of the brazilian biomes with the highest concentration of species and endemism, and is considered a hotspot of global biodiversity. Only 20% of its original area remains intact in the country and has been permanently transformed by human actions, which makes this biome a priority area for conservation. Due to the great loss of habitat and biodiversity, it is necessary to increase knowledge of basic characteristics of the existing species. Conservation genetics makes use of the genetic variation to infer the taxonomic status and life history of a species, making possible the study of characteristics such as the pattern of social structure and dispersion. Thus, basic data from species to which we have difficult assess can be accessed and better conservation strategies can be adopted. Mitochondrial and nuclear markers are important tools for studies of genetic and social structure in animals. In this study we assessed the taxonomic status of Clyomys bishopi, a colonial and semi-fossorial rodent inhabiting the Cerrado of São Paulo, and also the social and spatial genetic structure of individuals collected in the Estação Ecológica de Itirapina- SP. To evaluate C. bishopi taxonomic status we analyzed the cytochrome b gene of 21 individuals belonging to the genus and from various localities of its distribution. Phylogenetic methods based on Bayesian Inference, Maximum Parsimony and Neighbor-Joining showed high values of branch support separating individuals from São Paulo with respect to the rest of the samples, and genetic distance of about 4% between these groups, suggesting a significant genetic separation between them. For population analysis, ten microsatellite loci were isolated and characterized within C. bishopi. Through the analysis of 40 individuals sampled at three areas within EE de Itirapina, we verified high levels of genetic diversity (NA = 10, average HO = 0,76 and averaged HE = 0,90). Nine microsatellite loci and the control region of the mitochondrial DNA were used to assess the spatial and social genetic structure of the EE de Itirapina population. The Bayesian clustering analysis indicated no genetic structure, suggesting the occurrence of gene flow between sampled areas. Relatedness analyses indicated, in general, low relatedness indices between analyzed individuals (0.05), with slightly larger values when only individuals in a small geographic scale are considered. The average relatedness by sex indicated no significant differences, suggesting similar dispersion between males and females. Results suggest that colonies of Clyomys bishopi are composed largely by unrelated individuals, and present genetic structure compatible to that showed by species with solitary habits. / O Cerrado é um dos biomas brasileiros com maior concentração de espécies e endemismo, sendo considerado um hotspot de biodiversidade mundial. Apenas 20% da sua área original continua intacta no país e tem sido continuamente transformada pelas ações humanas, o que faz deste bioma uma área prioritária para conservação. Devido à grande perda de habitat e biodiversidade, faz-se necessário a ampliação do conhecimento de características básicas das espécies existentes. A genética da conservação faz uso da variação genética para inferir sobre o status taxonômico e a história de vida de uma espécie, tornando possível a investigação de características como o padrão de estruturação social e de dispersão. Assim, dados básicos de espécies de difícil avaliação direta podem ser acessados e melhores estratégias de conservação podem ser adotadas. Os marcadores moleculares dos tipos mitocondriais e nucleares são importantes ferramentas para estudos de estrutura genética populacional e social em animais. Neste estudo, avaliamos o status taxonômico de Clyomys bishopi, um roedor colonial e semi-fossorial que habita fisionomias do Cerrado do estado de São Paulo, e ainda a estrutura genética espacial e social de indivíduos coletados na Estação Ecológica de Itirapina-SP. Para a avaliação do status taxonômico, analisamos o gene citocromo b de 21 indivíduos do gênero provenientes de várias localidades distribuídas por quase toda a sua área de ocorrência. Métodos filogenéticos baseados em Inferência Bayesiana, Máxima Parcimônia e Neighbor-Joining mostraram altos valores de suporte de ramos na separação dos indivíduos de São Paulo em relação ao restante das amostras, e uma distância genética de aproximadamente 4% entre esses grupos, indicando separação genética significativa entre eles. Para a realização das análises populacionais, dez locos microssatélites foram isolados e caracterizados para Clyomys bishopi. Analisando-se 40 indivíduos coletados em três áreas no interior da EE de Itirapina, verificamos elevados índices de diversidade genética (NA = 10, HO média = 0,76 e HE média = 0,90). Nove desses locos, mais a região controle do DNA mitocondrial, foram utilizados para avaliar a estrutura genética espacial e a estrutura social na população analisada. A análise bayesiana de agrupamento não indicou estruturação genética espacial, sugerindo a ocorrência de fluxo gênico entre as áreas analisadas. As avaliações de parentesco indicaram, em geral, baixos coeficientes de parentesco entre os indivíduos analisados (0,05), com valores levemente maiores quando se considera indivíduos amostrados em uma pequena escala geográfica. O parentesco médio por sexo não indicou diferenças significativas, o que sugere a existência de dispersão similar entre machos e fêmeas. Os resultados obtidos sugerem que as colônias de Clyomys bishopi são formadas por indivíduos em sua maioria não aparentados, apresentando estrutura genética compatível com a de espécies que apresentam hábitos solitários.

The southeastern United States is a global biodiversity hotspot but has experienced severe declines of natural ecosystems. The southeast is currently facing widespread change, particularly from an increasing human population and climate change, that is likely to impact all remaining natural areas to some degree. In this work, I examine some of the challenges currently facing imperiled species of reptiles and amphibians in this region. The work is focused on two species, the Reticulated Flatwoods Salamander (Ambystoma bishopi) and the Eastern Indigo Snake (Drymarchon couperi) both of which are listed on the U.S. Endangered Species List. Chapter 1 used Light Detection and Ranging (LiDAR) data to measure wetland bathymetry (basin shape) in flatwoods salamander breeding wetlands. Bathymetry data were used to construct stage–area relationships for each wetland, and a history of water level monitoring data were applied to these relationships to build multi-year time series of flooded area metrics. These metrics were then combined with an assessment of vegetation characteristics to map potentially suitable habitat for flatwoods salamander breeding within each wetland. Chapter 2 focused on flatwoods salamander phenology (the timing of life history events) in response to climate change. I quantified flatwoods salamander movements into and out of breeding wetlands over a 10-year period (2010–2020), identifying temperature and precipitation patterns that were conducive to salamander movements. I then used future climate projections to forecast movement opportunities for flatwoods salamander from 2030–2099 and used an existing hydrologic model built on the same climate data to understand how phenology may interact with hydrology. Overall, only a small number of years are likely to have an ideal intersection of phenology and hydrology as has been observed during recent breeding seasons. Chapter 3 described the construction of a stochastic Integral Projection Model for flatwoods salamanders. I integrated the projections from Chapter 2 with the population model to estimate the viability of two flatwoods salamander populations from 2030–2099 under multiple climate change scenarios. The results indicated that approximately half of the examined scenarios resulted in a high probability (>0.5) of extinction when considering both wetland hydrology and salamander phenology. In Chapter 4, I described the creation of a stochastic Integral Projection Model for indigo snakes. I then demonstrated the utility of this model by examining the effects of initial population size, road density, and removal of individuals to support a captive colony on indigo snake populations. I found that high road densities and high collection rates would likely lead to population declines, although the rate of declines and extinction risk varied across scenarios. Taken together, these projects highlight some of the challenges currently facing herpetofauna in the southeastern United States, demonstrate the difficulty in conserving these often-overlooked species, and provide useful tools for ongoing conservation efforts focusing on these two imperiled species. / Doctor of Philosophy / We are in the midst of a global biodiversity crisis, with rates of species extinction far exceeding normal levels. Species loss is largely driven by global change attributable to human activities. A rapidly changing world can make it challenging to effectively conserve and manage imperiled species. In this work, I studied two species found only in the southeastern United States that are listed on the U.S. Endangered Species List. Chapters 1–3 focused on the Reticulated Flatwoods Salamander (Ambystoma bishopi), while Chapter 4 focused on the Eastern Indigo Snake (Drymarchon couperi). In Chapter 1, I used high resolution elevation data to map the shape of flatwoods salamander breeding wetlands. These data were then used to estimate flooded areas across multiple years. Flooded area metrics were combined with vegetation measurements to map potential flatwoods salamander breeding habitat. In Chapter 2, I examined how flatwoods salamander movements may respond to climate change. I identified time periods and weather conditions that coincided with flatwoods salamander movements into and out of breeding wetlands. I then projected potential movement opportunities based on multiple future climate scenarios for each breeding season from 2030–2099. My results showed that few years are likely to be ideal for flatwoods salamander reproduction, which is similar to trends observed in recent years. In Chapter 3, I built a population model for flatwoods salamanders. I then combined the model with predictions made in Chapter 2 to estimate the probability that populations would go extinct by the end of the century. The results indicated that the two flatwoods salamander populations examined had a high probability (>0.5) of extinction in about 50% of the climate scenarios. In Chapter 4, I constructed a population model for indigo snakes using a variety of available data. I used this model to examine the effects of road density, initial population size, and removal of individuals to support a captive colony on indigo snake populations. The results suggested that populations experiencing high road densities or high collection rates were likely to decline over time. These projects highlight some of the difficulties in conserving often-overlooked reptiles and amphibians in the southeastern United States and provide important tools for ongoing conservation projects working with these two imperiled species.

Globally, amphibian species have been in decline and a wide range of factors have been purported to be driving the decline. The Gulf Coastal Plain of Florida has a high degree of endemism and rarity and the biodiversity in the region includes a diverse suite of amphibian species. Degradation of habitat has been considered by many to be a major part of amphibian declines, however amphibian declines are complex and in many cases multiple factors are occurring in concert. My dissertation research examined aspects of habitat ecology and occupancy for two rare amphibians, Florida Bog Frog (Rana okaloosae) (Chapter 1, 2, and 3) and Reticulated Flatwoods Salamander (Ambystoma bishopi) (Chapter 5), that are both restricted to the Northern Gulf Coastal Plain. Further, for R. okaloosae I examined the influence of a sympatric congener, Bronze Frog (R. clamitans clamitans), on microhabitat selection (Chapter 1) and growth of tadpoles (Chapter 4). My overall goal was to be able to elucidate factors that limit the geographic range of R. okaloosae and A. bishopi and to identify habitat characteristics that managers could maintain or create to conserve or increase populations of these species. My first chapter examined the microhabitat relationships between R. okaloosae and R. c. clamitans. Rana okaloosae is endemic to northwestern Florida and is sympatric with R. c. clamitans, a more common and widely distributed congener. Further, the two species appeared to be syntopic, have overlapping breeding seasons, and are known to hybridize. The objectives of this chapter were to assess the microhabitat selection of both species and to assess differences in microhabitat use of males of both species during the breeding season. My modeling of habitat selection and comparison of variables used by each species suggests that males of these species select different resources when calling. Therefore, these sympatric ranids select for different resources at a fine scale, however there does appear to be some overlap among some selected habitat characteristics. In Chapter 2, I assessed the habitat use of R. okaloosae at multiple spatial scales. I surveyed for R. okaloosae and evaluated habitat characteristics at used sites and sites where I had no detections to develop among- and within-stream habitat models for R. okaloosae. Rana okaloosae used habitats with high amounts of emergent vegetation at both the among-stream scale and the within-stream scale. Emergent vegetation appears frequently in models of anuran habitat selection, particularly those that occur in fire-dominated landscapes. Further understanding the habitat requirements of R. okaloosae will allow land managers to use appropriate management activities (e.g., prescribed fire) that will increase emergent vegetation and potentially restore habitat that may help increase populations of R. okaloosae. In Chapter 3, I conducted aural surveys for R. okaloosae at two different spatial scales: range-wide and stream-level scales to understand how occupancy and colonization of R. okaloosae may be influenced by scale. My results suggest that at both spatial scales occupancy of R. okaloosae was best described by the presence of mixed forest wetlands at survey sites. At the range-wide scale, colonization and detection were constant across years, however, at the stream-level scale, colonization was predicted by the number of years since last fire and detection was best predicted by the additive combination of relative humidity and temperature. Occupancy of R. okaloosae was patchy at the range-wide and at the stream-level scales and colonization was low at both scales, while derived estimates of local extinction were moderately high. While R. okaloosae still occur in 3 watersheds where they were initially observed in the 1980's, one of the three watersheds appears to be very isolated and detections there are becoming very infrequent. In Chapter 4, I experimentally evaluated the effects of R. c. clamitans tadpoles on R. okaloosae tadpoles. My results suggest that there was limited influence of R. c. clamitans on R. okaloosae. Conversely, it appeared that Rana c. clamitans was more susceptible to intraspecific competition than interspecific competition. The lack of a strong competitive effect of Rana c. clamitans on Rana okaloosae suggests that competitive interactions among tadpoles may have a limited effect at the densities I examined. In Chapter 5, our objectives were to evaluate a suite of within-pool factors (i.e., vegetation structure, water level, and an index to presence of fish) that could influence occupancy of breeding wetlands by larval flatwoods salamanders on Eglin Air Force Base in Florida, USA. Site occupancy over a 4 year period was best described by a model that incorporated high herbaceous vegetation cover and open canopy cover. Detection probability was assessed, but it varied among years and was not included in the model. Our study suggests that managing the breeding habitat of flatwoods salamander for open canopies and dense herbaceous vegetation may contribute to this species' recovery. In conclusion, Chapters 1-3 of my dissertation contribute to a growing understanding about the habitat ecology of R. okaloosae. I have evaluated habitat use of R. okaloosae at multiple spatial scales. At the finest spatial scale R. okaloosae selected for sites that had an abundance of cover probably decreasing their risk of predation (Chapter 1). Similarly, in Chapter 2 at two spatial scales, among and within-streams, R. okaloosae selected for emergent vegetation. Finally, at the broadest spatial scale, range-wide, R. okaloosae were found to be associated with mixed forest wetlands (Chapter 3). I did not find strong support for competition between R. okaloosae and R. c. clamitans tadpoles, although there was some evidence of asymmetric competition (Chapter 4). Further, adult males of each species did not select the same habitat characteristics for calling sites, so there appeared to be some resource partitioning (Chapter 1). Finally, the presence of A. bishopi larvae was found to be associated with herbaceous vegetation and moderate amounts of canopy cover (Chapter 5). Results from Chapter 2 and 5 suggest that both R. okaloosae and A. bishopi are associated with habitat conditions that are likely a result of fire penetrating wetland areas. / Ph. D.

As diferentes demandas ecológicas impostas pela vida epígea e hipógea podem favorecer a melhoria das funções cognitivas espaciais. Comparamos a aprendizagem e a memória espacial, em labirinto complexo do Trinomys iheringi denigratus (terrestre de Mata Atlântica), T.yonenagae (semi-fossorial das Dunas na Caatinga) e Clyomys bishopi (fossorial do Cerrado). O aprendizado e memória espacial foram avaliados 10 T.i.denigratus (250±34g), 18 T.yonenagae (125±7,8g) e seis Clyomys bishopi (338±34g), a partir do: tempo até a saída do labirinto (TFL) e o número de erros cometidos (NER), em indivíduos colocados em labirinto de seis caminhos cegos e uma saída (0,20x1,10x1,50m), durante cinco testes consecutivos (30min) por dia, durante três dias consecutivos (fase de aprendizagem). Os ratos foram novamente testados em única sessão realizada após 48, 120 e 504h (fase de memória). O desempenho foi testado ao longo dos testes utilizando uma ANOVA um-fator (p<0,05). As taxas de aprendizagem (TA) do TFL das duas espécies foram mensuradas a partir do b da equação da curva de aprendizagem, e comparadas usando-se ANOVA um-fator (p<0,05). As variáveis do último teste de cada dia com os do primeiro teste do dia seguinte foram usadas como indicativas de memória-de-médio-prazo (MMP) e comparadas por espécie (teste T, dados-pareados, p<0,05). As taxas de retenção da aprendizagem (TR) com relação aos intervalos dos testes de memória, foram usadas como índice de memória-de-longo-prazo (MLP) e foram comparadas pela ANOVA um-fator (p<0,05). As espécies não apresentaram diferença significativa entre si mas não entre os testes tanto para TFL (F=0,85; p = 0,705; F=8,86; P < 0,000) quanto para NER (F=0,56; p=0,979; F = 3,65; p = 0,000). A TA foi marginalmente significativa (F = 2,784; p = 0,077) entre as espécies, sendo que T. yonenagae e C. bishopi não apresentaram diferença entre eles e ambos diferiram do T. i. denigratus. Com relação à memória-de-médio-prazo, T. Yonenagae e T. i. denigratus apresentaram diferença significativa para TFL entre os testes 10-11 (t = -3,406; p = 0,003) (t=- 2,300; p=0,050), mas não entre os testes 5-6 (t=-0,779; p=0,447) (t = -1,061; p=0,320). Já o C.bishopi sofreu redução entre os testes 5-6 e os testes 10-11, mas sem diferença significativa. Para a variável NER, não foi detectada diferença significativa, tanto para T. yonenagae (entre os 5 testes 5 e 6: t = 314,000, P = 0,558; testes 10 e 11: t=325,000, P=0,812) quanto para T.i.denigratus (entre os testes 5-6, t=0,590, P=0,562; testes 10-11: t=-1,855 P=0,080), por outro lado, C. bishopi apresentou a tendência de redução dos valores de NER entre os testes 5-6 (t = 0,442; p = 0,676) e 10-11 (t = 3,558; p = 0,016) sendo esta significativa. Quanto à MLP, dentre cada espécie não houve diferença significativa. Assim o semi-fossorial T.yonenagae e o fossorial C.bishopi se destacam quanto a aprendizagem do que a terrestre T.idenigratus, entretanto apenas C.bishopi é mais eficiente com relação a memoria-de-médio-prazo. Caso o efeito da filogenia seja controlada, sugere-se que o hábito fossorial seja componente do regime seletivo para evolução da cognição espacial / The different ecological contests imposed by epigeal and hypogeal lifestyles can facilitate the improvement of cognition spatial. We compared the spatial learning and memory in complex maze of Trinomys iheringi denigratus (terrestrial Atlantic rainforest), T.yonenagae (semifossorial Dunes of Caatinga) and Clyomys bishopi (fossorial of Cerrado). The spatial learning and memory were assessed 10 T.i.denigratus (250 ± 34g), 18 T.yonenagae (125 ± 7.8 g) and six Clyomys bishopi (338 ± 34g), from: time to exit the maze (TFL) and the number of errors (NER) in individuals placed on the complex maze of paths six blind and an outlet (0,20 x1, 10x1, 50m) for five consecutive tests (30min) per day for three consecutive days (learning phase) . The rats were tested in one session held after 48, 120 and 504Hours (memory phase). The performance was analyzed over the tests using a one-way ANOVA (p <0.05). Learning rates (TA) of the TFL of both species were measured from the b of the equation of the learning curve, and compared using one-way ANOVA (p <.05). The variables for each of the last test day with the first test the following day were used as indicators of memory to medium-term (MMP) and compared by species (t-test, paired-data, p <0.05) . Retention rates of learning (TR) with respect to the intervals of memory tests, were used as an index of memory-to-long-term (MLP) and were compared by one-way ANOVA (p <0.05). The species did not differ significantly from each other but not between tests for both TFL (F = 0.85, p = 0.705 F = 8.86, P <0.000) and for NER (F = 0.56, p = 0.979 , F = 3.65, p = 0.000). The TA was marginally significant (F = 2.784, p = 0.077) between species, and T. yonenagae and C. bishopi showed no difference between them and both were different from T. i. denigratus. With respect to memory-to-medium term, T. yonenagae and T. i. denigratus showed significant difference between the tests 11 and 10 for TFL (t = -3.406, p = 0.003) (t = -2.300, p = 0.050), but not between tests 5 and 6 (t = -0.779, p = 0.447 ) (t = -1.061, p = 0.320). Already C.bishopi showed decreased between tests 5-6 and tests 10-11, but without significant difference. For variable NER, no significant differences were detected for both T. yonenagae (between tests 5 and 6: t = 314.000, P = 0.558, Tests 10 and 11: t = 325,000 P = 0.812) and for T.i.denigratus (between tests 5.6, t = 0.590, P = 0.562; Tests 10-11: 7 t = -1.855 p = 0.080), on the other hand C. bishopi showed a trend to decreased NER between tests 5-6 (t = 0.442, p = 0.676) and 10-11 (t = 3.558, p = 0.016) which is significant. As for MLP, among each species there was no significant difference. Thus the semi-fossorial T.yonenagae and fossorial C.bishopi stand out as the learning of the terrestrial T.i.denigratus, however only C.bishopi is more efficient with respect to memory-to-medium term, if the effect of phylogeny is controlled, we suggested that the fossoriality could comprising the part of selective regime for evolution of the spatial cognition

As diferentes demandas ecológicas impostas pela vida epígea e hipógea podem favorecer a melhoria das funções cognitivas espaciais. Comparamos a aprendizagem e a memória espacial, em labirinto complexo do Trinomys iheringi denigratus (terrestre de Mata Atlântica), T.yonenagae (semi-fossorial das Dunas na Caatinga) e Clyomys bishopi (fossorial do Cerrado). O aprendizado e memória espacial foram avaliados 10 T.i.denigratus (250±34g), 18 T.yonenagae (125±7,8g) e seis Clyomys bishopi (338±34g), a partir do: tempo até a saída do labirinto (TFL) e o número de erros cometidos (NER), em indivíduos colocados em labirinto de seis caminhos cegos e uma saída (0,20x1,10x1,50m), durante cinco testes consecutivos (30min) por dia, durante três dias consecutivos (fase de aprendizagem). Os ratos foram novamente testados em única sessão realizada após 48, 120 e 504h (fase de memória). O desempenho foi testado ao longo dos testes utilizando uma ANOVA um-fator (p<0,05). As taxas de aprendizagem (TA) do TFL das duas espécies foram mensuradas a partir do b da equação da curva de aprendizagem, e comparadas usando-se ANOVA um-fator (p<0,05). As variáveis do último teste de cada dia com os do primeiro teste do dia seguinte foram usadas como indicativas de memória-de-médio-prazo (MMP) e comparadas por espécie (teste T, dados-pareados, p<0,05). As taxas de retenção da aprendizagem (TR) com relação aos intervalos dos testes de memória, foram usadas como índice de memória-de-longo-prazo (MLP) e foram comparadas pela ANOVA um-fator (p<0,05). As espécies não apresentaram diferença significativa entre si mas não entre os testes tanto para TFL (F=0,85; p = 0,705; F=8,86; P < 0,000) quanto para NER (F=0,56; p=0,979; F = 3,65; p = 0,000). A TA foi marginalmente significativa (F = 2,784; p = 0,077) entre as espécies, sendo que T. yonenagae e C. bishopi não apresentaram diferença entre eles e ambos diferiram do T. i. denigratus. Com relação à memória-de-médio-prazo, T. Yonenagae e T. i. denigratus apresentaram diferença significativa para TFL entre os testes 10-11 (t = -3,406; p = 0,003) (t=- 2,300; p=0,050), mas não entre os testes 5-6 (t=-0,779; p=0,447) (t = -1,061; p=0,320). Já o C.bishopi sofreu redução entre os testes 5-6 e os testes 10-11, mas sem diferença significativa. Para a variável NER, não foi detectada diferença significativa, tanto para T. yonenagae (entre os 5 testes 5 e 6: t = 314,000, P = 0,558; testes 10 e 11: t=325,000, P=0,812) quanto para T.i.denigratus (entre os testes 5-6, t=0,590, P=0,562; testes 10-11: t=-1,855 P=0,080), por outro lado, C. bishopi apresentou a tendência de redução dos valores de NER entre os testes 5-6 (t = 0,442; p = 0,676) e 10-11 (t = 3,558; p = 0,016) sendo esta significativa. Quanto à MLP, dentre cada espécie não houve diferença significativa. Assim o semi-fossorial T.yonenagae e o fossorial C.bishopi se destacam quanto a aprendizagem do que a terrestre T.idenigratus, entretanto apenas C.bishopi é mais eficiente com relação a memoria-de-médio-prazo. Caso o efeito da filogenia seja controlada, sugere-se que o hábito fossorial seja componente do regime seletivo para evolução da cognição espacial / The different ecological contests imposed by epigeal and hypogeal lifestyles can facilitate the improvement of cognition spatial. We compared the spatial learning and memory in complex maze of Trinomys iheringi denigratus (terrestrial Atlantic rainforest), T.yonenagae (semifossorial Dunes of Caatinga) and Clyomys bishopi (fossorial of Cerrado). The spatial learning and memory were assessed 10 T.i.denigratus (250 ± 34g), 18 T.yonenagae (125 ± 7.8 g) and six Clyomys bishopi (338 ± 34g), from: time to exit the maze (TFL) and the number of errors (NER) in individuals placed on the complex maze of paths six blind and an outlet (0,20 x1, 10x1, 50m) for five consecutive tests (30min) per day for three consecutive days (learning phase) . The rats were tested in one session held after 48, 120 and 504Hours (memory phase). The performance was analyzed over the tests using a one-way ANOVA (p <0.05). Learning rates (TA) of the TFL of both species were measured from the b of the equation of the learning curve, and compared using one-way ANOVA (p <.05). The variables for each of the last test day with the first test the following day were used as indicators of memory to medium-term (MMP) and compared by species (t-test, paired-data, p <0.05) . Retention rates of learning (TR) with respect to the intervals of memory tests, were used as an index of memory-to-long-term (MLP) and were compared by one-way ANOVA (p <0.05). The species did not differ significantly from each other but not between tests for both TFL (F = 0.85, p = 0.705 F = 8.86, P <0.000) and for NER (F = 0.56, p = 0.979 , F = 3.65, p = 0.000). The TA was marginally significant (F = 2.784, p = 0.077) between species, and T. yonenagae and C. bishopi showed no difference between them and both were different from T. i. denigratus. With respect to memory-to-medium term, T. yonenagae and T. i. denigratus showed significant difference between the tests 11 and 10 for TFL (t = -3.406, p = 0.003) (t = -2.300, p = 0.050), but not between tests 5 and 6 (t = -0.779, p = 0.447 ) (t = -1.061, p = 0.320). Already C.bishopi showed decreased between tests 5-6 and tests 10-11, but without significant difference. For variable NER, no significant differences were detected for both T. yonenagae (between tests 5 and 6: t = 314.000, P = 0.558, Tests 10 and 11: t = 325,000 P = 0.812) and for T.i.denigratus (between tests 5.6, t = 0.590, P = 0.562; Tests 10-11: 7 t = -1.855 p = 0.080), on the other hand C. bishopi showed a trend to decreased NER between tests 5-6 (t = 0.442, p = 0.676) and 10-11 (t = 3.558, p = 0.016) which is significant. As for MLP, among each species there was no significant difference. Thus the semi-fossorial T.yonenagae and fossorial C.bishopi stand out as the learning of the terrestrial T.i.denigratus, however only C.bishopi is more efficient with respect to memory-to-medium term, if the effect of phylogeny is controlled, we suggested that the fossoriality could comprising the part of selective regime for evolution of the spatial cognition