This dissertation begins with proposing a novel street Connectivity Indicator (C.I.) to predict transit performance by identifying the role that street network connectivity plays in influencing the service quality of demand responsive feeder transit services. This new C.I. definition is dependent upon the expected shortest path between any two nodes in the network, includes spatial features with transit demand distribution information and is easy to calculate for any given service area. Subsequently, a methodology to identify and locate critical links within a grid street system for operating feeder transit services is also developed. A 'critical' street link causes the largest change in transit performance due to the link's removal or addition to an existing network. The most important contribution of this section on link criticality is to present a simple closed-form analytical formula in locating the critical link(s) for a grid street network system of 'any' size. Easily computable formulas have been provided and validated by simulation analyses. Another related model is proposed to compute the optimal grid street spacing that would enhance performance of a demand responsive feeder transit system. The model is tested using simulation. Lastly, an analytical model is also developed for estimating optimal service cycle length or headway of a demand responsive feeder transit service designed to serve passengers, especially during peak periods of demand. Simulation analyses over a range of networks have been conducted to validate the new C.I. definition. Results show a desirable monotonic relationship between transit performance and the proposed C.I., whose values are directly proportional and therefore good predictors of the transit performance, outperforming other available indicators, typically used by planners. Further, useful insights indicate a monotonic decrease in link criticality as we depart from the centrally located links to those located at boundaries. Using a real case example from Denver of the Call-n-Ride system operating similar to a demand responsive feeder transit, optimal cycle lengths differed very modestly from those computed using the model. Extensive simulations performed for different sets of feeder service areas and demand densities, further validated the optimal cycle length model.

Wildlife corridors have become a widely adopted management strategy for the conservation of species in fragmented habitats. Fragmentation reduces the size of habitat patches and increases the isolation of the populations within them, potentially resulting in extinction due to stochastic processes. The provision of a corridor between habitat patches is believed to increase the level of connectivity through the integration of populations into a single demographic unit, thus increasing the probability of survival. This assumption remains largely untested due to both a lack of investigation, and design limitations in some of the few studies performed. Connectivity is often assumed to occur simply from the presence of individuals within the corridor. Design criteria essential for the rigorous assessment of connectivity were identified and a landscape meeting these criteria selected. The vegetation within the corridor was found to be comparable in both structure and species composition to that of the patches that it connected. Two target species (Melomys cervinipes and Uromys caudimaculatus) were shown to occur along the corridor but not within the surrounding matrix. The combination of these factors indicated that the corridor was suitable for use as a model system and ensured that any subsequent results truly reflected the capacity of the corridor to function in the desired manner. The population structure was similar within the corridor and the connected patches for both species. Weights of individuals, sex ratios and the percentage of juveniles were consistent between the two system components, suggesting that the corridor contained breeding populations. Connectivity was therefore possible via generational gene flow for both species, while long distance movement events for U. caudimaculatus also indicated that direct movement between habitat patches may be possible for this larger species. Despite all ecological parameters indicating that connectivity was likely, genetic markers (mtDNA and nDNA) revealed significant population differentiation between the connected patches for both species. Populations linked by the corridor and those in isolated habitats were found to show the same level of genetic differentiation. Sampling at a finer spatial scale within connected patches and a continuous control habitat showed that population differentiation was common for M. cervinipes. Given the continuity of suitable vegetation, and the presence of individuals of breeding age along the corridor system, this was attributed to social structuring. U. caudimaculatus populations also showed evidence of genetic differentiation within a connected patch and along the corridor, despite panmixia within the continuous habitat. Having investigated a model system, the data from this study has implications for other wildlife corridor studies and for landscape managers. Firstly, the advantages of using an integrated ecological and genetic approach have been demonstrated. While genetic data determined the level of connectivity, the ecological data provided an understanding of the processes operating within the system. Secondly, the level of scale at which wildlife corridor studies are conducted may need addressing. Most studies currently treat a fragmented landscape in a binary manner and consider the connected patch to be the finest "grain". However, the processes responsible for the lack of connectivity were found to operate at the much finer within-patch scale. Finally, this study clearly indicated that not all wildlife corridors will provide connectivity between the connected populations and that connectivity cannot be inferred from the presence of individuals within the corridor. Given that social behaviour such as territorial defence and philopatry are common in many species, especially small mammals, a lack of connectivity via a wildlife corridor may be more common than currently assumed. The successful use of wildlife corridors as a management strategy, and the accurate assessment of their effectiveness therefore requires careful consideration of not only structural attributes of the corridor, but also behavioural, demographic and genetic parameters of the target species.

La vision classique et hiérarchique de l’organisation du système moteur est remise en question par la découverte d’autres structures que le cortex moteur primaire capables d’influer en parallèle sur la sortie motrice, dans le cadre d’un modèle hodotopique. L’objectif de ce travail a été d’étudier les réseaux de contrôle du mouvement à l’aide de stimulations électriques directes cérébrales lors de chirurgies en condition éveillée pratiquées pour l’exérèse de tumeurs cérébrales. Cette méthode permet la préservation de la fonction motrice tout en apportant de nombreuses informations sur l’organisation du système moteur en général. Le contrôle moteur a été étudié au travers du phénomène moteur négatif, qui consiste en un arrêt du mouvement sans perte de tonus ou de conscience lors d’électrostimulations. Décrit initialement au niveau cortical, nos travaux ont montré qu’il était possible d’identifier ce phénomène dans la substance blanche de façon bilatérale et que les fibres supportant ces réponses motrices négatives sont organisées d’une manière somatotopique, à l’instar de la voie pyramidale. Nous avons pu mettre en évidence une voie modulatrice motrice bilatérale capable d’inhiber les mouvements des deux membres supérieurs lors d’une stimulation sous-corticale unilatérale. Nous avons montré qu’une lésion des sites générant des réponses motrices négatives conduisait à un syndrome de l’aire motrice supplémentaire et à des troubles définitifs des mouvements fins et de la coordination bimanuelle. Ces résultats expliquent les troubles observés lors de chirurgies des régions prémotrices réalisées sans cartographie motrice active, c’est-à-dire cherchant uniquement à identifier les structures motrices primaires. Ils valident la nécessité de pratiquer une cartographie motrice en condition éveillée et ce quelle que soit la dominance hémisphérique. Ces résultats à l’échelle sous-corticale nous ont conduit à définir le concept de réseau moteur négatif et son implication dans les réseaux de contrôle moteur. Le preuve de l’existence de tels réseaux nous a permis d’explorer le niveau cortical et de mettre en évidence une organisation systématique des aires motrices négatives, proposant ainsi une autre vision que celles d’une répartition aléatoire ou somatotopique. Cette organisation en plusieurs aires effecteurs-dépendantes et redondantes a permis de confirmer par les stimulations électriques directes la ségrégation du gyrus précentral en plusieurs gradient rostro-caudal et dorso-ventral. L’ensemble de ces éléments nous a permis d’émettre plusieurs hypothèses concernant le rôle de ces réseaux. Nous supposons qu’il s’agit de plusieurs réseaux interconnectés fonctionnant à l’aide de mécanismes inhibiteurs internes, dont le rôle va de la modulation du flux moteur dans le cadre de processus décisionnels compétitifs intégrés aux aires négatives à l’inhibition vraie d’un comportement moteur dans le cadre de circuits entre le cortex et les noyaux gris centraux. La carte probabiliste réalisée permettra de planifier les chirurgies cérébrales mais aussi de servir de zone d’intérêt pour les nouvelles thérapies par stimulations et la recherche en neurosciences. / The classical and hierarchical view of the motor system has been challenged since the discovery of other structures able to modulate the motor output in the framework of a hodotopic model. The aim of this work was to study the motor control network thanks to direct electrostimulations performed during awake surgery for brain tumors. This method has shown its effectiveness to preserve motor functions while giving new highlights about the organization of the motor system. In our work, motor control has been studied through the negative motor phenomenon, which consists in a complete arrest of movement without loss of tonus or consciousness during electrostimulations. Initially described at a cortical level, our work demonstrated the possibility to elicit negative motor phenomenon in both hemispheres at a subcortical level. Moreover, we identified a bilateral modulatory motor pathway able to inhibit both upper limbs during unilateral subcortical stimulations. We also shown that fibers driving negative motor responses are organized in a somatotopic manner, like the pyramidal pathway. Resection of these fibers lead to a supplementary motor area syndrome with permanent deficit in fine motor skills and bimanual coordination. These results explain the neurological deficits which might occur after surgery in premotor areas when no active brain mapping is performed, that is, when only primary motor structures are sought. They emphasize the necessity to perform a motor mapping during awake surgery whatever the side and hemispheric dominance. These subcortical results led us to define the concept of negative motor networks and their involvement in motor control networks. Evidences of this network allowed us to explore the cortical level and to report a well-defined organization of the negative motor area, different from the random or somatotopic distributions previously described. This effector-dependent and redundant organization in several areas defined by direct electrostimulations has been helpful to confirm the rostro-caudal and dorso-ventral segregation of the precentral gyrus. Consequently, it was possible to propose several hypothesis about the role of these networks. We presume that they are constituted by several large-scale interconnected networks, based on internal inhibitor mechanisms, whose role goes from modulation of the motor output in a competitive model of decision-making integrated in the negative motor area to real inhibition of motor behaviors thanks to cortico-basal ganglia circuitry. The probabilistic map created with these works will be helpful to plan surgery but could also provide regions of interest for brain stimulations therapies as well as neuroscientific research.

Le balbuzard pêcheur Pandion haliaetus est une espèce de rapace cosmopolite et migratrice mais menacée. C'est un modèle intéressant pour explorer comment les adaptations comportementales au sein de populations géographiquement distantes, ayant évoluées dans conditions écologiques différentes, pourraient être une cause proximale de leur distribution géographique et de différences génétiques et comportementales (stratégies migratrices et comportement de recherche alimentaire). Ainsi, j'ai mené une étude intégrative à différentes échelles afin de répondre à plusieurs questions fondamentales et appliquées.Grâce à une approche phylogéographique (phylogénie moléculaire) j'ai mis en évidence l'existence de quatre lignées différentes à l'échelle mondiale qui peuvent être considérées comme des Unités Evolutives Significatives (ESUs) et qui méritent des mesures de gestion spécifiques. En combinant des méthodes d'écologie moléculaire et d'observations d'oiseaux marqués, j'ai pu estimer le taux de connexion populationnelle dans le Paléarctique occidental. En combinant des outils d'écologie trophique en utilisant les isotopes stables, et d'écologie spatiale en utilisant de nouveaux outils en bio télémétrie, j'ai étudié les comportements migratoires et de l'écologie hivernale des balbuzards pêcheurs méditerranéens, que j'ai pu comparer avec ceux des populations nord européennes. Finalement, j'ai étudié la nécessité d'adopter des mesures de gestions adéquates pour trois population du bassin méditerranéen (Corse, Maroc, Italie), en comparant la dynamique des populations, en lien avec leurs ressources alimentaires (recensement des espèces piscicoles) et les perturbations d'origine anthropique. Dans l'ensemble, ce travail a conduit à des avancées significatives par rapport à la conservation des balbuzards pêcheurs à différentes échelles biogéographiques. Ces connaissances sont particulièrement utiles à l'égard d'une gestion efficace de cette espèce emblématique. / Being a cosmopolitan and migratory species, the osprey Pandion haliaetus provides a good opportunity to explore how behavioural adaptations in different populations, that evolved under different ecological conditions and are widely spaced, can be the proximate causes of geographical distribution, genetic divergence, population connectivity, migratory strategies and foraging ecology. According to this, a multi-scale integrated approach has been adopted for the osprey: through an interdisciplinary framework made by molecular ecology, trophic ecology as assessed via stable isotopic analyses, spatial ecology through the use of novel biotelemetry tools, as well as population dynamics, fish censuses and assessments of levels of human disturbance, we: a) evidenced the existence of four different lineages at global scale that should be treated as Evolutionary Significant Units (ESUs) and deserve specific management; b) revealed population connectivity in the Western Palearctic; c) disentangled the migratory behaviour and winter ecology of Mediterranean ospreys and compared it with northern European populations, in an evolutionary context; d) stressed the need of adequate management measures to be adopted for three local populations of the Mediterranean basin (Corsica, Morocco and Italy). Overall, this work has led to some important advances with respect to the conservation biogeography of ospreys at different scales. Those insights are particularly valuable with respect to the effective management of this emblematic species.

In this study, we used resting-state functional magnetic resonance imaging (fMRI) scans from subjects with early mild cognitive impairment (EMCI) and control subjects to study functional network connectivity. The scans were acquired by the Alzheimer's Disease Neuroscience Initiative (ADNI). We used genetic data from the ADNI database to further subdivide the EMCI and control groups into genotype groups with or without the ApoE4 allele. ROI-to-ROI resting-state functional connectivity was measured using Freesurfer and the Functional Connectivity Toolbox for Matlab (CONN). In our analysis, we compared whole-brain ROI connectivity strength and ROI-to-ROI functional network connectivity strength between EMCI, control and genotype subject groups. We found that the ROI network properties were disrupted in EMCI and ApoE4-containing groups. Notably, we show that (1) EMCI disrupts functional connectivity strength in many areas; (2) the ApoE4 allele disrupts functional connectivity strength in similar areas to EMCI; and (3) the differences in functional connectivity between groups shows a multifactor contribution to functional network dysfunction along the trajectory leading to dementia.

No description available.

Esse estudo teve como objetivo comparar duas formas de manejo de cultivos de banana em relação ao seu potencial de abrigar espécies florestais. Nós assumimos que o registro de espécies florestais utilizando um elemento da matriz mostra que a espécie não evita a matriz totalmente e pode ser capaz de cruzá-la, indicando a importância dessas estruturas para a manutenção de conectividade funcional entre as áreas de hábitat. Foram amostrados 30 cultivos incluindo áreas sob manejo convencional e agroflorestal. Em cada unidade amostral foram realizadas transecções para acessar a proporção de aves florestais, generalistas e de áreas abertas. Variáveis descritoras da estrutura interna e da estrutura da paisagem do entorno dos cultivos foram consideradas para verificar se a avifauna responde à características intrínsecas dos cultivos ou ao contexto de paisagem em que os cultivos estão inseridos. A proporção de espécies florestais foi maior em cultivos agroflorestais (F28;0,001=19,29; p<0,0004), enquanto a proporção de generalistas não apresentou diferença significativa (F28;0,001=0,22; p<0,22) e a de espécies de áreas abertas foi maior em cultivos convencionais (F28;0,001=8,92; p<0,006). Os dois conjuntos de variáveis considerados, isto é, a estrutura interna do cultivo e do entorno deste foram importantes para predizer a proporção de espécies florestais. Embora tenhamos observado o efeito da paisagem na proporção de espécies florestais, salientamos aqui que sistemas convencionais, mesmo em contextos de paisagem similares à sistemas agroflorestais, não apresentaram a mesma proporção de espécies florestais. Esse resultado indica um diferencial de cultivos agroflorestais de banana para a ocorrência de espécies florestais. Em nível de paisagem, a proporção de florestas e consequentemente a diminuição do isolamento entre manchas é o responsável pela presença potencial de espécies em fragmentos. No entanto, em nível de mancha, mesmo esta apresentando um contexto de paisagem favorável para que seja ocupada por espécies de áreas fonte, o mecanismo que influencia a presença das espécies é a elevada estrutura interna dos bananais. / This study aimed to compare two management of banana crops in relation to its potential to host forest bird species. We assume that the record of forest bird species using a matrix element shows that the species does not entirely avoid this area and may be able to cross it, indicating the importance of these structures for the maintenance of functional connectivity between areas of habitat. We sampled 30 areas including crops under conventional and agroforestry management. In each sampling unit we used transects to access the composition and abundance of bird species. Descriptive variables of the internal structure and the structure of the surrounding landscape of crops were considered to determine whether the bird responds to the intrinsic characteristics of the crops or the landscape context in which crops are inserted. The proportion of forest species was higher in agroforestry crops (F28, 0,001 = 19.29, p <0.0004), while the proportion of generalists showed no significant difference (F28, 0,001 = 0.22, p <0.22) and species of open areas was greater in conventional crops (F28, 0,001 = 8.92, p <0.006). The two sets of variables considered, i.e. the internal structure of the crop and of surroundings, were important to predict the proportion of forest birds species. Although we observed the effect of landscape in the proportion of forest species, we note here that conventional systems, even in contexts similar to the landscape agroforestry systems, did not show the same proportion of forest species. This result indicates a upper value of agroforestry crops for the occurrence of forest bird species. At landscape level, the proportion of forests and consequently the lower isolation between patches is responsible for the potential presence of species in fragments. However, at patch level, even it is presenting a landscape context favorable to be occupied by a species of source areas, the mechanism that influences the presence of the species is the high internal structure of the banana plantations.

Master of Agribusiness / Department of Agricultural Economics / Terry Griffin / Large acreage farms and even moderate sized farms employing custom applicators and harvesters have multiple machines in the same field at the same time conducting the same field operation. As a method to control input costs and minimize application overlap, these machines have been equipped with automatic section control (ASC). For nearly all these multiple-vehicle operations, over application is a concern especially for more irregularly shaped fields; however modern technology including automated guidance combined with automatic section control allow reduced doubling of input application including seeds, fertilizer, and spray. Automatic section control depends on coverage maps stored locally on each vehicle to determine whether or not to apply input products and up to now, there has not been a clear method to share these maps between vehicles in the same field. Without sharing coverage maps, an individual ASC planting unit only has location data where it has applied individually and no location data for where other planting units have applied seed in that same field. Automatic section control relies upon shared coverage maps to be continually updated between each planting unit and utilizes existing machine telematics infrastructure for map data sharing. Telematics utilizes a cloud computing platform and cellular connectivity which in rural areas is known to have limited service levels. Planting operations were simulated for two 16-row planters, each using two John Deere GreenStar3 2630 monitors, simulated GPS location data stream, electronic rate control units, and individual row unit clutches to have control at the finest granularity. Each simulated planting unit is equipped with automatic section control and telematics gateways to share coverage map data from the first planting unit to JDLink cloud infrastructure then out to the second. This study evaluates the impact that field size and shape have on using multiple ASC planters and coverage map sharing, and estimates seed cost savings from reducing over application because coverage maps are shared between planting units. The impact of sharing coverage maps with both planting units using field boundaries with automatic section control and without using field boundaries were evaluated. Guidance line headings were determined using AgLeader SMS’s mission planning feature to minimize the number of passes across each field based on the field boundary and implement width. Each field was run twice using parallel tracking, once each with and without coverage map sharing to observe the extent of over application. The field level data were then taken to examine a fictious 3,000 acre farming operation where the field level data was used as a partial composition of the farm operation. An embedded Microsoft Excel macro was used to create 8,008 different composition scenarios to determine farm level savings. The average farm savings was $58,909 per year. Additionally, using the 8,008 scenarios, time value of money was examined to determine the the minimum area required annually for five years for this technology to pay back. The average was 133 acres each year for five years. Equipment manufacturers and farmers have interest in these results. In general, equipment manufacturers desire to create a service-based product to be sold such that continual revenue path provides value added services after the precision agriculture hardware is sold. In this study, the existing telematics product offerings are tied to shared coverage maps to provide a value-add to an existing service. Farmers want to ensure this is a sound equipment investment with payback in a relatively short time period. As farm input costs continue to rise especially relative to crop prices, reducing over application will be critical to limit waste.

Made available in DSpace on 2016-06-02T19:29:24Z (GMT). No. of bitstreams: 1 3095.pdf: 1545712 bytes, checksum: fb785bf4285325c762ff7f39a652816f (MD5) Previous issue date: 2010-05-24 / The present study was aimed to assess the role of the flood pulse related to the composition of the ichthyofauna of the Caiçara Basin System (Upper Baía Caiçara/UBC e Lower Baía Caiçara/LBC) in a pantanal environment, in Cáceres, Mato Grosso. 90,041 specimens were collected from September 2005 to September 2007, distributed in two classes, 8 orders, 32 families and 171 species, with 145 of them occurring in both environments. Characiforms, Siluriforms and Gymnotiforms were the most representative orders accounting for more than 90% of all the species obtained in the sampled environments. The UBC, with 155 collected species, considering the two sampled cycles, obtained 68 and 81 constant species, accounting for 50 and 56% of the total species in this bay. In the LBC, with 161 sampled species, the constant species were most significant, with respectively 43 and 56% per cycle and 54 species for the entire sampling period. Odontostilbe pequira, Serrapinnus calliurus and Moenkhausia dichroura were abundant in all the periods. M. dichroura was the most abundant species in practically all the periods and environments. Cyphocharax gillii and Hypoptopoma inexspectatum (detritivores) were abundant in the ebb tide and dry periods. Acestrorhynchus pantaneiro and Roeboides prognathus were the most abundant piscivores in the ebb tide period. Hemiodus orthonops was abundant in the flood and high tide periods. Among the invertivores/insectivores, M. dichroura and O. pequira were abundant in all the seasonal periods. Concomitantly to the fish collection, data from the limnological variables in the two sampled environments were also obtained, which showed significant differences between the seasonal periods. In the dry periods, the UBC remained totally disconnected from the Paraguay River, and the BCI obtained a smaller water amount from the main channel. In the flood period the connectivity between the UBC and the Paraguay River was reestablished. The concentration of dissolved oxygen varied from 0.67 to 5.94 mg/l in BCS and 0.64 and 5.87 mg/l in the LBC. The highest water temperatures were observed during the flood period (31.35 °C) and the lowest temperature during the dry period (22.33°C). Concerning electrical conductivity, the highest values were observed in the flood period (84.19 &#956;S.cm-1). The lowest values were obtained in the ebb tide period (23.30 &#956;S.cm-1). It obtained the lowest pH values (5.66) in the high tide period, and the highest pH values (7.50) in the dry period. Characiforms and Siluriforms were the orders that showed the best adaptation to the variations caused by the flood pulse of this system and the limnological attributes (dissolved oxygen, pH, electrical conductivity in water, pH, water transparency and depth) are decisive in the arrangement of the ichthyofauna, but their action is not isolated. On the contrary, they are interdependent together with the river-basin connectivity. / Este trabalho teve por objetivo estudar o papel do pulso de inundação relacionado à composição da ictiofauna do sistema de baías Caiçara (Baía Caiçara Superior BCS e Baía Caiçara Inferior BCI), no pantanal de Cáceres MT. Foram coletados 90.041 espécimes, de setembro de 2005 a setembro de 2007, compreendendo dois ciclos, distribuídos em duas classes, 8 ordens, 32 famílias e 171 espécies, representando 63% de toda ictiofauna descrita para todo o ambiente pantaneiro. As Ordens Characiformes e Syluriformes foram as mais representativas, com 81,9% de todas as espécies obtidas para o sistema estudado. BCS, com 155 espécies coletadas, obteve 68 (Ciclo I) e 81 (Ciclo II) espécies constantes, representando 50 e 56% do total para essa baía. Em BCI, com 161 espécies amostradas, as espécies constantes foram mais significativas, com respectivamente 43 e 56% por ciclo e 54 espécies para todo o período amostral. Odontostilbe pequira, Serrapinnus calliurus e Moenkhausia dichroura foram abundantes em todos os períodos. M. dichroura foi a espécie mais abundante em praticamente todos os períodos e ambientes. Cyphocharax gillii e Hypoptopoma inexspectatum (detritívoros) foram abundantes no período de vazante e estiagem. Acestrorhynchus pantaneiro e Roeboides prognathus se destacaram entre os piscívoros na vazante. Hemiodus orthonops foi abundante nos períodos de enchente e cheia. Dentre os invertívoros/insetívoros, M. dichroura e O. pequira foram abundantes em todos os períodos sazonais. Concomitante as coletas de peixes também foram obtidos os dados das variáveis físicas e químicas, apresentando diferenças significativas entre os períodos sazonais. Nos períodos de estiagem BCS permaneceu totalmente desconectada do rio Paraguai, e BCI obteve um menor aporte hídrico do canal principal. No período de enchente foi restabelecida a conexão entre BCS e o rio Paraguai. A concentração de oxigênio dissolvido variou de 0,67 a 5,94 mg/l em BCS e 0,64 e 5,87 mg/l em BCI. As maiores temperaturas da água foram observadas no período de enchente (31,35 °C) e a menor no período de estiagem (22,33°C). Para a condutividade elétrica observam-se maiores valores na enchente (84,19 &#956;S.cm-1). Os menores valores foram obtidos na vazante (23,30 &#956;S.cm-1). No período de cheia apresentou os menores valores de pH (5,66), e na estiagem os maiores (7,50). Characiformes e Siluriformes são as ordens mais adaptadas às variações decorrentes do pulso de inundação neste sistema e os atributos limnológicos (oxigênio dissolvido, pH, condutividade elétrica da água, pH, transparência da água e profundidade) são decisivos na estruturação da ictiofauna, mas não atuam de maneira isolada, sendo interdependentes, juntamente com a conectividade rio-baía.

Esse estudo teve como objetivo comparar duas formas de manejo de cultivos de banana em relação ao seu potencial de abrigar espécies florestais. Nós assumimos que o registro de espécies florestais utilizando um elemento da matriz mostra que a espécie não evita a matriz totalmente e pode ser capaz de cruzá-la, indicando a importância dessas estruturas para a manutenção de conectividade funcional entre as áreas de hábitat. Foram amostrados 30 cultivos incluindo áreas sob manejo convencional e agroflorestal. Em cada unidade amostral foram realizadas transecções para acessar a proporção de aves florestais, generalistas e de áreas abertas. Variáveis descritoras da estrutura interna e da estrutura da paisagem do entorno dos cultivos foram consideradas para verificar se a avifauna responde à características intrínsecas dos cultivos ou ao contexto de paisagem em que os cultivos estão inseridos. A proporção de espécies florestais foi maior em cultivos agroflorestais (F28;0,001=19,29; p<0,0004), enquanto a proporção de generalistas não apresentou diferença significativa (F28;0,001=0,22; p<0,22) e a de espécies de áreas abertas foi maior em cultivos convencionais (F28;0,001=8,92; p<0,006). Os dois conjuntos de variáveis considerados, isto é, a estrutura interna do cultivo e do entorno deste foram importantes para predizer a proporção de espécies florestais. Embora tenhamos observado o efeito da paisagem na proporção de espécies florestais, salientamos aqui que sistemas convencionais, mesmo em contextos de paisagem similares à sistemas agroflorestais, não apresentaram a mesma proporção de espécies florestais. Esse resultado indica um diferencial de cultivos agroflorestais de banana para a ocorrência de espécies florestais. Em nível de paisagem, a proporção de florestas e consequentemente a diminuição do isolamento entre manchas é o responsável pela presença potencial de espécies em fragmentos. No entanto, em nível de mancha, mesmo esta apresentando um contexto de paisagem favorável para que seja ocupada por espécies de áreas fonte, o mecanismo que influencia a presença das espécies é a elevada estrutura interna dos bananais. / This study aimed to compare two management of banana crops in relation to its potential to host forest bird species. We assume that the record of forest bird species using a matrix element shows that the species does not entirely avoid this area and may be able to cross it, indicating the importance of these structures for the maintenance of functional connectivity between areas of habitat. We sampled 30 areas including crops under conventional and agroforestry management. In each sampling unit we used transects to access the composition and abundance of bird species. Descriptive variables of the internal structure and the structure of the surrounding landscape of crops were considered to determine whether the bird responds to the intrinsic characteristics of the crops or the landscape context in which crops are inserted. The proportion of forest species was higher in agroforestry crops (F28, 0,001 = 19.29, p <0.0004), while the proportion of generalists showed no significant difference (F28, 0,001 = 0.22, p <0.22) and species of open areas was greater in conventional crops (F28, 0,001 = 8.92, p <0.006). The two sets of variables considered, i.e. the internal structure of the crop and of surroundings, were important to predict the proportion of forest birds species. Although we observed the effect of landscape in the proportion of forest species, we note here that conventional systems, even in contexts similar to the landscape agroforestry systems, did not show the same proportion of forest species. This result indicates a upper value of agroforestry crops for the occurrence of forest bird species. At landscape level, the proportion of forests and consequently the lower isolation between patches is responsible for the potential presence of species in fragments. However, at patch level, even it is presenting a landscape context favorable to be occupied by a species of source areas, the mechanism that influences the presence of the species is the high internal structure of the banana plantations.