Made available in DSpace on 2014-06-11T19:23:03Z (GMT). No. of bitstreams: 0 Previous issue date: 2010-02-26Bitstream added on 2014-06-13T19:28:54Z : No. of bitstreams: 1 potascheff_cm_me_rcla.pdf: 651964 bytes, checksum: 79461ec2253d037c431fc957d54d050d (MD5) / Eschweilera nana é a única Lecythidoideae (Lecythidaceae) com ampla distribuição no Cerrado. A evolução floral dessa subfamília ocorreu principalmente no androceu, provavelmente em resposta aos polinizadores. Foram estudadas a atividade floral e a biologia reprodutiva da espécie e o comportamento dos seus visitantes florais. Não existe relação específica com nenhum táxon polinizador. Seu sistema de polinização envolve uma guilda de polinizadores, formada por abelhas, uma vespa e um besouro, apesar do capuz estaminal restringir o acesso ao néctar. / Eschweilera nana is the only wide spread Lecythidoideae (Lecythidaceae) native to the Brazilian Cerrado. Floral evolution of this subfamily has occurred mostly in the androecium, probably in response to selective pressures by pollinators. Floral biology and the reproductive biology of this species and the behavior of the floral visitors were studied. There is no one-toone relashionship. The pollination system of E. nana involves a guild of pollinators of mostly bees, one wasp and one beetle, despite the difficulty to access the nectar.

Resumo: Eschweilera nana é a única Lecythidoideae (Lecythidaceae) com ampla distribuição no Cerrado. A evolução floral dessa subfamília ocorreu principalmente no androceu, provavelmente em resposta aos polinizadores. Foram estudadas a atividade floral e a biologia reprodutiva da espécie e o comportamento dos seus visitantes florais. Não existe relação específica com nenhum táxon polinizador. Seu sistema de polinização envolve uma guilda de polinizadores, formada por abelhas, uma vespa e um besouro, apesar do capuz estaminal restringir o acesso ao néctar. / Abstract: Eschweilera nana is the only wide spread Lecythidoideae (Lecythidaceae) native to the Brazilian Cerrado. Floral evolution of this subfamily has occurred mostly in the androecium, probably in response to selective pressures by pollinators. Floral biology and the reproductive biology of this species and the behavior of the floral visitors were studied. There is no one-toone relashionship. The pollination system of E. nana involves a guild of pollinators of mostly bees, one wasp and one beetle, despite the difficulty to access the nectar. / Orientador: Julio Antonio Lombardi / Coorientador: Scott Alan Mori / Banca: Kayna Agostini / Banca: Fernando Pedroni / Mestre

Submitted by Automa??o e Estat?stica (sst@bczm.ufrn.br) on 2016-05-31T20:42:20Z No. of bitstreams: 1 AlexandreTorresSilvaDosSantos_TESE.pdf: 4479347 bytes, checksum: 3c5c147ce62a6256530f17e60cf8fb0f (MD5) / Approved for entry into archive by Arlan Eloi Leite Silva (eloihistoriador@yahoo.com.br) on 2016-06-03T00:41:26Z (GMT) No. of bitstreams: 1 AlexandreTorresSilvaDosSantos_TESE.pdf: 4479347 bytes, checksum: 3c5c147ce62a6256530f17e60cf8fb0f (MD5) / Made available in DSpace on 2016-06-03T00:41:26Z (GMT). No. of bitstreams: 1 AlexandreTorresSilvaDosSantos_TESE.pdf: 4479347 bytes, checksum: 3c5c147ce62a6256530f17e60cf8fb0f (MD5) Previous issue date: 2014-12-08 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior (CAPES) / O objetivo deste trabalho foi determinar a variabilidade sazonal-interanual e tend?ncia da velocidade do vento e validar o modelo num?rico de mesoescala para depois acoplar com o modelo num?rico de microescala no intuito de obter recursos e?licos em algumas localidades no Nordeste do Brasil (NEB). Para isso usamos dois conjuntos de dados de velocidade do vento (esta??es meteorol?gicas e torres anemom?tricas) e dois modelos din?micos; um de mesoescala e outro de microescala. Usamos ferramentas estat?sticas para avaliar e validar os dados medidos e simulados. As simula??es do modelo din?mico de mesoescala foram realizadas atrav?s de m?todos de assimila??o de dados (relaxamento newtoniano e filtro de Kalman). Os principais resultados apontam: (i) Cinco grupos homog?neos de velocidade do vento no NEB com valores maiores no inverno e primavera e menores para ver?o e outono; (ii) A variabilidade interanual da velocidade do vento em alguns grupos se destacaram com valores maiores; (iii) A circula??o de grande escala modificada por fen?menos El Ni?o e La Ni?a alterou a intensidade da velocidade do vento para os grupos com maiores valores; (iv) A an?lise de tend?ncia mostrou valores negativos mais significativos para os grupos G3, G4 e G5 em todas as esta??es do ano e na m?dia anual; (v) O desempenho do modelo din?mico de mesoescala apresentou menores erros nas localidades Paracuru e S?o Jo?o do Cariri e maiores erros foram observados em Triunfo; (vi) Aplica??o do filtro de Kalman diminuir significativamente os erros sistem?ticos apresentados nas simula??es do modelo din?mico de mesoescala; (vii) Os recursos e?licos apontam que as ?reas Paracuru e Triunfo s?o favor?veis para gera??o de energia, sendo que a t?cnica de acoplamento ap?s valida??o apresentou melhor resultado para Paracuru. Conclu?mos que o objetivo do estudo foi alcan?ado, sendo poss?vel agora identificar tend?ncias em grupos homog?neos do comportamento do vento, al?m de avaliar ? qualidade das simula??es tanto com o modelo din?mico de mesoescala como de microescala para responder quest?es necess?rias antes de planejar projetos de pesquisas na ?rea e?lica. / The objective of this study was to determine the seasonal and interannual variability and calculate the trends of wind speed in NEB and then validate the mesoscale numerical model for after engage with the microscale numerical model in order to get the wind resource at some locations in the NEB. For this we use two data sets of wind speed (weather stations and anemometric towers) and two dynamic models; one of mesoscale and another of microscale. We use statistical tools to evaluate and validate the data obtained. The simulations of the dynamic mesoscale model were made using data assimilation methods (Newtonian Relaxation and Kalman filter). The main results show: (i) Five homogeneous groups of wind speed in the NEB with higher values in winter and spring and with lower in summer and fall; (ii) The interannual variability of the wind speed in some groups stood out with higher values; (iii) The large-scale circulation modified by the El Ni?o and La Ni?a intensified wind speed for the groups with higher values; (iv) The trend analysis showed more significant negative values for G3, G4 and G5 in all seasons and in the annual average; (v) The performance of dynamic mesoscale model showed smaller errors in the locations Paracuru and S?o Jo?o and major errors were observed in Triunfo; (vi) Application of the Kalman filter significantly reduce the systematic errors shown in the simulations of the dynamic mesoscale model; (vii) The wind resource indicate that Paracuru and Triunfo are favorable areas for the generation of energy, and the coupling technique after validation showed better results for Paracuru. We conclude that the objective was achieved, making it possible to identify trends in homogeneous groups of wind behavior, and to evaluate the quality of both simulations with the dynamic model of mesoscale and microscale to answer questions as necessary before planning research projects in Wind-Energy area in the NEB

O presente trabalho apresenta a modelagem do escoamento de ar sobre um terreno complexo empregando, e comparando os resultados, três programas comerciais de previsão de potencial eólico, WaSP, Meteodyn WT e WindSim. Por se tratar de um modelo simplificado, o programa WaSP nem sempre é adequado para previsão de energia em terrenos muito complexos pois não é capaz de prever turbulência, separação de escoamento e efeitos de estabilidade existentes neste tipo de terreno. Para que se consiga modelar a turbulência existente de maneira mais coerente, utilizam-se programas de mecânica dos fluidos computacional como o Meteodyn WT e o WindSim que modelam o escoamento através das equações médias de Reynolds. A finalidade do estudo é identificar a capacidade de previsão das velocidades de vento por meio de cada um dos programas em um parque eólico localizado na Chapada da Diamantina, tendo em vista que a correta previsão da produção de energia depende de um entendimento correto e detalhado dos recursos disponíveis. Avaliando apenas condição atmosférica neutra e comparando os resultados obtidos, conclui-se que o programa WindSim obteve o menor erro na estimativa da velocidade do vento, -3,7% quando comparado aos outros programas WaSP e Meteodyn, -5,1% e -6,5% respectivamente. O maior erro de aproximação foi obtido pelo programa WaSP, +11% quando comparado aos outros programas Meteodyn e WindSim, +10,6% e +7,9% respectivamente. Para o programa Meteodyn foram avaliados casos com diferentes condições de estabilidade, o erro máximo e mínimo foi de +10,6% e -6,5%, respectivamente em atmosfera estável, mantendo-se elevado. O programa WindSim permite a alteração de variáveis que são fixas no Meteodyn e também a simulação dos efeitos de estabilidade através da inclusão da equação de energia inicializada através do comprimento de Monin-Obukhov o que, implicou em uma melhora significativa nos resultados obtidos e em um maior tempo computacional. O erro, menor e maior, apresentado pelo programa WindSim foi reduzido a, -1,8% e +4,4% respectivamente. Destaca-se que as conclusões obtidas se aplicam ao caso estudado. / This work studies the air flow modeling over a complex terrain using and comparing the results of three commercial programs that estimates the wind field in a site: WaSP, Meteodyn WT and WindSim. Being WAsP a simplified model, it is not always suitable to estimate the flow in complex terrain because it is not able to predict the turbulence, flow separations and stability effects present in these kind of terrains. To model the existing turbulence two computational fluid mechanics programs i.e. Meteodyn WT and WindSim, which models the flow through the Reynolds Averaged Navier-Stokes equations have been used. This study aims to identify the wind speed predictive capacity of each program in a windfarm located on Chapada da Diamantina, as a correct production prediction depends on a correct estimation of the wind field. Evaluating only neutral atmospheric setup and comparing it with the measures, WindSim had the lowest wind speed estimation error, -3.7% compared with WaSP and Meteodyn programs, that provided -5.1% and -6.5% respectively. The greatest error was obtained by WaSP program, +11%, followed by Meteodyn, +10.6% and WindSim, +7.9%. WindSim enables changes in some variables that are fixed in Meteodyn and enables also the simulation of stability effects through the introduction of the energy equation with a Monin-Obukhov length initialization. For Meteodyn program different stabilities were evaluated, but the results obtained were not satisfactory. The maximum and minimum error were reduced to +10.6% and -6.5%, respectively, in stable atmosphere. Due to the specific site studied conditions the simulation of stability effects resulted in a significant improvement in the results even though a larger computational time. Enabling the energy equations setting up stable atmospheric stability, both WindSim errors are reduced to -1.8% and +4.4% respectively. It is noteworthy that the conclusions apply to the case studied.

O presente trabalho apresenta a modelagem do escoamento de ar sobre um terreno complexo empregando, e comparando os resultados, três programas comerciais de previsão de potencial eólico, WaSP, Meteodyn WT e WindSim. Por se tratar de um modelo simplificado, o programa WaSP nem sempre é adequado para previsão de energia em terrenos muito complexos pois não é capaz de prever turbulência, separação de escoamento e efeitos de estabilidade existentes neste tipo de terreno. Para que se consiga modelar a turbulência existente de maneira mais coerente, utilizam-se programas de mecânica dos fluidos computacional como o Meteodyn WT e o WindSim que modelam o escoamento através das equações médias de Reynolds. A finalidade do estudo é identificar a capacidade de previsão das velocidades de vento por meio de cada um dos programas em um parque eólico localizado na Chapada da Diamantina, tendo em vista que a correta previsão da produção de energia depende de um entendimento correto e detalhado dos recursos disponíveis. Avaliando apenas condição atmosférica neutra e comparando os resultados obtidos, conclui-se que o programa WindSim obteve o menor erro na estimativa da velocidade do vento, -3,7% quando comparado aos outros programas WaSP e Meteodyn, -5,1% e -6,5% respectivamente. O maior erro de aproximação foi obtido pelo programa WaSP, +11% quando comparado aos outros programas Meteodyn e WindSim, +10,6% e +7,9% respectivamente. Para o programa Meteodyn foram avaliados casos com diferentes condições de estabilidade, o erro máximo e mínimo foi de +10,6% e -6,5%, respectivamente em atmosfera estável, mantendo-se elevado. O programa WindSim permite a alteração de variáveis que são fixas no Meteodyn e também a simulação dos efeitos de estabilidade através da inclusão da equação de energia inicializada através do comprimento de Monin-Obukhov o que, implicou em uma melhora significativa nos resultados obtidos e em um maior tempo computacional. O erro, menor e maior, apresentado pelo programa WindSim foi reduzido a, -1,8% e +4,4% respectivamente. Destaca-se que as conclusões obtidas se aplicam ao caso estudado. / This work studies the air flow modeling over a complex terrain using and comparing the results of three commercial programs that estimates the wind field in a site: WaSP, Meteodyn WT and WindSim. Being WAsP a simplified model, it is not always suitable to estimate the flow in complex terrain because it is not able to predict the turbulence, flow separations and stability effects present in these kind of terrains. To model the existing turbulence two computational fluid mechanics programs i.e. Meteodyn WT and WindSim, which models the flow through the Reynolds Averaged Navier-Stokes equations have been used. This study aims to identify the wind speed predictive capacity of each program in a windfarm located on Chapada da Diamantina, as a correct production prediction depends on a correct estimation of the wind field. Evaluating only neutral atmospheric setup and comparing it with the measures, WindSim had the lowest wind speed estimation error, -3.7% compared with WaSP and Meteodyn programs, that provided -5.1% and -6.5% respectively. The greatest error was obtained by WaSP program, +11%, followed by Meteodyn, +10.6% and WindSim, +7.9%. WindSim enables changes in some variables that are fixed in Meteodyn and enables also the simulation of stability effects through the introduction of the energy equation with a Monin-Obukhov length initialization. For Meteodyn program different stabilities were evaluated, but the results obtained were not satisfactory. The maximum and minimum error were reduced to +10.6% and -6.5%, respectively, in stable atmosphere. Due to the specific site studied conditions the simulation of stability effects resulted in a significant improvement in the results even though a larger computational time. Enabling the energy equations setting up stable atmospheric stability, both WindSim errors are reduced to -1.8% and +4.4% respectively. It is noteworthy that the conclusions apply to the case studied.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2016-05-20T15:18:24Z No. of bitstreams: 1 carlosalbertodossantossouza.pdf: 1765667 bytes, checksum: 50df1c267e4657d7cc09cb359cc8f9ab (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2016-07-02T11:24:49Z (GMT) No. of bitstreams: 1 carlosalbertodossantossouza.pdf: 1765667 bytes, checksum: 50df1c267e4657d7cc09cb359cc8f9ab (MD5) / Made available in DSpace on 2016-07-02T11:24:49Z (GMT). No. of bitstreams: 1 carlosalbertodossantossouza.pdf: 1765667 bytes, checksum: 50df1c267e4657d7cc09cb359cc8f9ab (MD5) Previous issue date: 2012-02-08 / A vespa escavadora Sphex ingens apresenta distribuição conhecida na Bahia e no litoral Atlântico do Sudeste brasileiro representando uma espécie pouco compreendida em termos comportamentais, ecológicos e biológicos. O estudo visou descrever o comportamento sexual in natura da espécie e avaliar as estratégias comportamentais e padrões morfológicos sujeitos a seleção sexual, sendo conduzido nas praias dos Meros e do Aventureiro entre a 2ª quinzena de 2010 e a 1ª e 2ª quinzena de 2011 através da tabulação e filmagem dos comportamentos de acasalamento de indivíduos em ambas fisionomias. O etograma foi elaborado pelos métodos ad libitum e sequence samples. Diagramas de fluxos foram construídos a partir do estado atual e das probabilidades estocásticas dos comportamentos. O acasalamento foi constituído por 4 etapas seqüenciais através de um sistema promíscuo. Um total de 10 comportamentos foram exclusivos para as fêmeas e 9 para os machos. Sete comportamentos foram comuns entre os sexos não havendo diferença significativa (x2=16,622; Gl=n-1; p<0,01). Comportamentos de comunicação constituíram maior freqüência de estado entre os sexos no acasalamento, entretanto, as predições registradas pelo Diagrama de Cadeia de Markov indicaram que estes comportamentos pouco influenciam no acasalamento, sendo mais relevante a permanência da monta pelos machos e os comportamentos agonísticos e de insociabilidade das fêmeas. O número de tentativas de cópulas entre machos patrulheiros e residentes foram significativamente diferentes na 2ª quinzena de 2010 e na 1ª quinzena de 2011, respectivamente, t=3,767; Gl=n-1; p<0,01 e t=12,780; Gl=n-1; p<0,01. Embora, não tenha existido diferença significativa para o número de cópulas entre machos patrulheiros e residentes na 2ª quinzena de 2011 (t=0,285; Gl=n-1; p<0,01), os resultados sugerem que o sucesso copulatório entre estes padrões se mostra variável ao longo da estação reprodutiva. O número de toques de antenas e o tempo de investimento entre cópula bem sucedidas e mal sucedidas foram significativamente correlacionados (rs=0,838; t=14,127; p<0,001 e rs=0,847; t=14,083; p<0,001), porém, o número de toques de antenas entre cópulas bem sucedidas e mal-sucedidas não houve diferiram significativamente (t=0,607; Gl=n-1; p<0,01), ao contrário do tempo de investimento entre as respectivas cópulas (t=4,236; Gl=n-1; p<0,01). Análises de variância demonstraram diferenças significativas nos caracteres morfológicos mensurados, enquanto as análises multivariadas predisseram a existência de assimetria entre fêmeas, machos residentes e patrulheiros. A assimetria entre machos residente e patrulheiro indica maior adaptação ao territorialismo do que a seleção das fêmeas para este padrão, enquanto a promiscuidade e a agressividade das fêmeas interferem no esforço de corte dos machos e fomenta o aparecimento de coercividade copulatória. / Wasps of the genus Sphex ingens (commonly known as digger wasps) are widely distributed in Bahia and in Brazil’s southeast Atlantic coast, representing a poorly understood species in terms of behavior, ecology and biology. The present study aims to describe the sexual behaviour in natura of the species, as well as to evaluate behavioral strategies and morphological patterns subject to sexual selection. The study was conducted on Meros and Aventureiro beaches, between the 2nd fortnight of 2010 and the 1st and 2nd fortnights of 2011, using tabulation and footage of the mating behaviors of individuals in both faces. The ethogram was developed using ad libitum and sequence samples methods. Flow diagrams were drawn based on the current state and on the stochastic probabilities of the behaviors. The mating was composed of four sequential steps of a promiscuous mating system. There were 10 types of behavior which were exclusive to females and 9 other types which were exclusive of males. Seven behaviors were common between the sexes, with no significant difference (x2=16,622; Gl=n-1; p<0,01). Communication behaviors were more frequent between the sexes during the mating process, however, the recorded predictions in the Markov chain diagram indicated that these behaviors have little influence on mating, being the males’ permanency and the females’ agonistic & intractability behaviors much more relevant. The number of attempted copulations between patrol and resident males were significantly different in the 2nd fortnight of 2010 and 1st fortnight of 2011, respectively, t=3,767; n=Gl-1; p<0,01 and t=12,780; Gl=n-1, p<0,01. Although no significant differences existed for the quantities of mating between patrol x residents males in the 2nd fortnight of 2011 (t=0,285, Gl=n-1, p<0,01), the results suggest that the copulatory success between these patterns is variable along the reproductive season. The number of touches on each other antennas and the investment time between successful x unsuccessful mating were significantly correlated (rs=0,838; t=14,127, p<0,001 and rs=0,847; t=14,083; p<0,001), however, the number of antennas touches occurring between successful x unsuccessful copulations did not differ significantly (t=0,607; Gl=n-1; p<0,01), unlike the investment period between the matings (t=4,236; Gl=n-1; p<0,01). A variance analysis demonstrated significant differences in measured morphological traits, while a multivariate analysis predicted an asymmetry between females, patrol males and resident males. The asymmetry between resident and trooper males indicates a greater adaptation to the territorialism than the females selection for this pattern. Meanwhile, the females aggression and promiscuity may interfere on the males' courtship behaviour and fosters the development of copulatory coercivity.

We present new 3.6 and 4.5 mu m Spitzer phase curves for the highly irradiated hot Jupiter WASP-33b and the unusually dense Saturn-mass planet HD 149026b. As part of this analysis, we develop a new variant of pixel-level decorrelation that is effective at removing intrapixel sensitivity variations for long observations (>10 hr) where the position of the star can vary by a significant fraction of a pixel. Using this algorithm, we measure eclipse depths, phase amplitudes, and phase offsets for both planets at 3.6 and 4.5 mu m. We use a simple toy model to show that WASP-33b's phase offset, albedo, and heat recirculation efficiency are largely similar to those of other hot Jupiters despite its very high irradiation. On the other hand, our fits for HD 149026b prefer a very high albedo. We also compare our results to predictions from general circulation models, and we find that while neither planet matches the models well, the discrepancies for HD 149026b are especially large. We speculate that this may be related to its high bulk metallicity, which could lead to enhanced atmospheric opacities and the formation of reflective cloud layers in localized regions of the atmosphere. We then place these two planets in a broader context by exploring relationships between the temperatures, albedos, heat transport efficiencies, and phase offsets of all planets with published thermal phase curves. We find a striking relationship between phase offset and irradiation temperature: the former drops with increasing temperature until around 3400 K and rises thereafter. Although some aspects of this trend are mirrored in the circulation models, there are notable differences that provide important clues for future modeling efforts.

Plant–animal interaction systems are complex food webs where the members—plants, pollinators, herbivores, parasites and predators of the pollinators/herbivores—interact with each other in ways which maximize their own fitness. Based on the net outcome, such interactions could be mutually beneficial to the interacting members (mutualism) or beneficial to only one of the interacting members at the cost of the other interacting members (herbivory, predation, parasitism). It is possible that such outcomes are actually a continuum and could swing in either direction from beneficial to detrimental and vice versa. Such transitions happen not only over long time scales, but could also happen within shorter time scales based on conditionalities. Conditional outcomes are those in which the outcome of an interaction between two partners is conditional on the involvement of a third partner. Thus, studying such outcomes necessitates taking into account systems beyond the classical two-partner interactions. In such complex multitrophic plant–animal interaction systems in which there are direct and indirect interactions between species, comprehending the dynamics of these multiple partners is very important for an understanding of how the system is structured. In Chapter 2 we investigate Ficus racemosa and its community of obligatory mutualistic and parasitic fig wasps that develop within the fig inflorescence or syconium, as well as their interaction with opportunistic ants. We focus on temporal resource partitioning among members of the fig wasp community over the development cycle of the fig syconia during which wasp oviposition and development occur and we study the activity rhythm of the ants associated with this community. We found that the members of the wasp community partitioned their oviposition across fig syconium development phenology and showed interspecific variation in activity across the diel cycle. The wasps presented a distinct sequence in their arrival at fig syconia for oviposition. We documented night oviposition in several fig wasp species for the first time. Ant activity on the fig syconia was correlated with wasp activity and was dependent on whether the ants were predatory or trophobiont-tending species; only numbers of predatory ants increased during peak arrivals of the wasps. In Chapter 3, we found that predatory ants (Oecophylla smaragdina) patrolling F. racemosa trees were attracted to the odour from fig syconia at different developmental phases, as well as to the odours of fig wasps, whereas other predatory ants (Technomyrmex albipes) responded only to odours of syconia from which fig wasps were dispersing and to fig wasp odour. However, trophobiont-tending ants (Myrmicaria brunnea) patrolling the same trees and exposed to the same volatiles were unresponsive to fig or fig wasp odours. The predatory ants demonstrated a concentration-dependent response towards volatiles from figs receptive to pollinators and those from which wasps were dispersing while the trophobiont-tending ants were unresponsive to such odours at all concentrations. Naıve predatory ants failed to respond to the volatiles to which the experienced predatory ants responded, indicating that the response to fig-related odours is learned. In Chapter 4 we characterise the dynamics of the volatile bouquet of the fig syconium from the initiation through pre-receptive, receptive, and late inter-floral stages which act as signals/ cues for different fig wasp species. We were also interested in diel patterns of volatile emission as some fig wasp species were strictly diurnal (the pollinator, Ceratosolen fusciceps) whereas other fig wasps such as Apocryptophagus fusca were observed ovipositing even during the nocturnal hours. We identified volatiles that were specific to syconium development phase as well as to the time of day in this bouquet. α-muurolene was identified as the sesquiterpene specific to receptive-phase as well as being present only during the day thus coinciding with the diurnal pollinator arrival pattern. Volatiles such as (E)-β-ocimene were present in increasing levels across the developmental stages of the fig and thus could act as background volatiles providing suitable information to fig wasps about host plants and their phases. Chapter 5 examines the responses of predatory and trophobiont-tending ant species to the cuticular hydrocarbon (CHC) extracts of four galler and two parasitoid fig wasp species associated with F. racemosa. Interestingly, the antennation response of both experienced and na¨ıve ants to these wasp extracts was identical indicating that prior exposure to such compounds is not necessary for eliciting such response. We also characterised these cuticular hydrocarbon extracts to find potential compounds which could as short-range cues for predatory ants. Ants were more responsive to the cuticular extracts of parasitoids rather than to those of galler wasps, implying that the CHC profile of carnivorous prey may contain more elicitors of aggressive behaviour in ants compared to herbivorous prey whose profiles may be more similar to those of their plant resources. We also find congruency between the cuticular profiles of parasitoids and their hosts suggesting that parasitoids could sequester compounds from their diet. Important findings and conclusions of the thesis are presented in Chapter 6. The first two parts of the appendices section discuss work carried out on alternative ways of analysing multivariate data sets such as plant volatiles and insect cuticular hydrocarbons. Appendix A details the use of Random Forests, an algorithm-based method of analysing complex data sets where there are more variables than samples, a situation akin to microarray data sets. This work illustrates the use of such techniques in chemical ecology, highlighting the potential pitfalls of classical multivariate tests and the advantages of newer more robust methods. Appendix B, an invited article following the publication of the earlier work, compares different data transformation procedures currently employed in such multivariate analysis. Appendix C details sex-specific differences in cuticular hydrocarbons of fig wasps, using the pollinator C. fusciceps as a case study.

Robots are increasingly expected to go beyond controlled environments in laboratories and factories, to enter real-world public spaces and homes. However, robot behavior is still usually engineered for narrowly defined scenarios. To manually encode robot behavior that works within complex real world environments, such as busy work places or cluttered homes, can be a daunting task. In addition, such robots may require a high degree of autonomy to be practical, which imposes stringent requirements on safety and robustness. \setlength{\parindent}{2em}\setlength{\parskip}{0em}The aim of this thesis is to examine methods for automatically learning safe robot behavior, lowering the costs of synthesizing behavior for complex real-world situations. To avoid task-specific assumptions, we approach this from a data-driven machine learning perspective. The strength of machine learning is its generality, given sufficient data it can learn to approximate any task. However, being embodied agents in the real-world, robots pose a number of difficulties for machine learning. These include real-time requirements with limited computational resources, the cost and effort of operating and collecting data with real robots, as well as safety issues for both the robot and human bystanders.While machine learning is general by nature, overcoming the difficulties with real-world robots outlined above remains a challenge. In this thesis we look for a middle ground on robot learning, leveraging the strengths of both data-driven machine learning, as well as engineering techniques from robotics and control. This includes combing data-driven world models with fast techniques for planning motions under safety constraints, using machine learning to generalize such techniques to problems with high uncertainty, as well as using machine learning to find computationally efficient approximations for use on small embedded systems.We demonstrate such behavior synthesis techniques with real robots, solving a class of difficult dynamic collision avoidance problems under uncertainty, such as induced by the presence of humans without prior coordination. Initially using online planning offloaded to a desktop CPU, and ultimately as a deep neural network policy embedded on board a 7 quadcopter.

No description available.