A major question in ecology is 'why are tropical forests so species diverse?' One hypothesis to explain tropical species diversity is the Janzen-Connell hypothesis. This model assumes high levels of host-specific seed and seedling predation and / or pathogen attack when seedlings occur at high density near to the parent tree; seedlings are more likey to survive and reach maturity the further they are away from parents / conspecific adults. Theoretically this should lead to a random distribution of each species in the forest, which in turn will lead to high species diversity. Here I test the Janzen-Connell hypothesis for the first time in a submontane dry forest in Nigeria, West Africa. Specifically I tested whether or not a) leaf herbivory decreases and b) seedling survival increases with distance from parent / conspecific adult trees. These two components were tested separately on naturally occurring seedlings and on experimentally planted seedlings. I also tested whether or not conspecific adult trees showed clumped distributions by testing if conspecific nearest neighbours were observed more often than would be expected by chance alone. Naturally occurring seedlings of three species, Pouteria altissima, Newtonia buchananii and Isolona pleurocarpa showed significantly greater survival at distances away from parent / conspecific adult trees. Two out of a total of three species (Entandrophragma angolense, Deinbollia pinnata and Sterculia pinnata) of experimentally planted seedlings showed increased survival at distances away from conspecific adult trees, but this trend was non-significant. Leaf herbivory decreased with distance from parent / conspecific adult trees for four out of a total of six study species, but all relationships of leaf herbivory with distance for these six species were non-significant. Of two individual species, Anthonotha noldeae and Carapa procera, and two species groups tested for clumping, all had a greater number of conspecific nearest neighbours than would be expected to occur by chance alone, and this was significant for the two species groups. The decreased survival of seedlings under parent / conspecific adult trees is likely to maintain tree species diversity in West African submontane forests as predicted by the Janzen-Connell model. The role of host specific seedling herbivores in reducing recruitment under parent / conspecific adult trees requires further investigation. Although conspecific adults showed some degree of clumping no conclusion was reached as to whether this was evidence for or against the Janzen-Connell model.

A major question in ecology is 'why are tropical forests so species diverse?' One hypothesis to explain tropical species diversity is the Janzen-Connell hypothesis. This model assumes high levels of host-specific seed and seedling predation and / or pathogen attack when seedlings occur at high density near to the parent tree; seedlings are more likey to survive and reach maturity the further they are away from parents / conspecific adults. Theoretically this should lead to a random distribution of each species in the forest, which in turn will lead to high species diversity. Here I test the Janzen-Connell hypothesis for the first time in a submontane dry forest in Nigeria, West Africa. Specifically I tested whether or not a) leaf herbivory decreases and b) seedling survival increases with distance from parent / conspecific adult trees. These two components were tested separately on naturally occurring seedlings and on experimentally planted seedlings. I also tested whether or not conspecific adult trees showed clumped distributions by testing if conspecific nearest neighbours were observed more often than would be expected by chance alone. Naturally occurring seedlings of three species, Pouteria altissima, Newtonia buchananii and Isolona pleurocarpa showed significantly greater survival at distances away from parent / conspecific adult trees. Two out of a total of three species (Entandrophragma angolense, Deinbollia pinnata and Sterculia pinnata) of experimentally planted seedlings showed increased survival at distances away from conspecific adult trees, but this trend was non-significant. Leaf herbivory decreased with distance from parent / conspecific adult trees for four out of a total of six study species, but all relationships of leaf herbivory with distance for these six species were non-significant. Of two individual species, Anthonotha noldeae and Carapa procera, and two species groups tested for clumping, all had a greater number of conspecific nearest neighbours than would be expected to occur by chance alone, and this was significant for the two species groups. The decreased survival of seedlings under parent / conspecific adult trees is likely to maintain tree species diversity in West African submontane forests as predicted by the Janzen-Connell model. The role of host specific seedling herbivores in reducing recruitment under parent / conspecific adult trees requires further investigation. Although conspecific adults showed some degree of clumping no conclusion was reached as to whether this was evidence for or against the Janzen-Connell model.

The Janzen-Connell mechanism is proposed to maintain plant diversity: predators and diseases of seeds reduce the number of seeds that survive near a parent tree, but allow seeds far from the parent tree to grow into adulthood. In the area where seeds don’t survive, seedlings from other tree species which are not affected by the seed consumer can grow. At large scales, this effect is thought to increase overall plant diversity. Soil-borne pathogens can contribute to seed mortality in this way, but we don’t know how important different parts of their lifecycle are in creating Janzen-Connell patterns. To determine the role of soil-borne pathogens in the development of Janzen-Connell patterns, we constructed a simulation model to examine how tree and pathogen characteristics affect plant spatial patterns. Under specific combinations of tree and pathogen characteristics, we found that pathogens could create Janzen-Connell patterns. The most important parameters were how far trees dispersed their seeds, and how many seeds a tree produced in a year. These characteristics determined how much of an impact the pathogens were able to have on the tree population because they rely on the density of seeds to determine how far they can spread and how many spores they will produce.

Muitas teorias, englobando diferentes fatores e mecanismos, já foram postuladas para explicar a alta coexistência de espécies arbóreas nos trópicos, a qual permanece como uma questão intrigante e subentendida na ecologia vegetal. O estudo da dinâmica de árvores ao nível populacional contribui e embasa, por sua vez, o entendimento desses fatores e mecanismos atuando ao nível da comunidade. O objetivo geral do presente estudo foi investigar alguns aspectos relacionados com a dinâmica de uma espécie arbórea comum Calophyllum brasiliense Camb.(Clusiaceae) em floresta de restinga alta Ilha do Cardoso, Cananéia, SP. Para isso, além de termos testado se a sobrevivência e o desempenho de suas plântulas estavam relacionados com níveis de herbivoria dependentes da densidade, testamos se a ocorrência da espécie apresentava associação com hábitats de solo e caracterizamos sua distribuição espacial na área de estudo. Na Introdução Geral (Capítulo 1) nós enumeramos as principais teorias já propostas para explicar a alta diversidade tropical, as quais incluem diversos mecanismos atuantes na dinâmica de espécies arbóreas. Tradicionalmente, modelos que focam em fatores dependentes da densidade eram freqüentemente contrastados com modelos baseados na segregação de hábitats e nichos para explicar a coexistência de espécies, ainda que atualmente sabe-se que atuem concomitantemente na estruturação das comunidades. Por esse motivo, os descrevemos em linhas gerais. Apresentamos, também, como o estudo da distribuição espacial de uma espécie pode dar indícios de processos subjacentes responsáveis pelos padrões gerados, os quais devem ser inferidos e posteriormente testados. Adicionalmente, resumimos uma teoria em especial, a qual embasa o capítulo seguinte e foi o ponto inicial dessa dissertação: o Modelo Janzen-Connell. Além de termos explorado brevemente seu contexto conceitual, também revisamos os principais resultados de investigações de seus efeitos em outras áreas de estudo e com diferentes metodologias. Como os dois demais capítulos foram desenvolvidos utilizando a mesma área e espécie de estudo, também incluímos neste capítulo suas respectivas descrições. Para testar o modelo Janzen-Connell (Capítulo 2), nós delineamos um experimento no qual avaliamos os danos por herbivoria, a mortalidade e o desempenho das plântulas de C. brasiliense, sob diferentes tratamentos: proteção contra herbivoria, distância e agrupamento de adultos coespecíficos. Encontramos que os efeitos dependentes da distância e da densidade não atuaram como previsto pela teoria para a espécie de estudo, a qual, apesar de ter sofrido altos danos por herbivoria, mostrou-se tolerante e apresentou crescimento compensatório em resposta a estes. Além de termos proposto que a pressão por seus herbívoros especialistas parece encontrar-se amplamente distribuída na área de estudo, e não agregada ao redor de densidades de coespecíficos, também sugerimos que o micro-hábitat, especialmente a umidade do solo parece ser melhor preditora da sobrevivência da espécie do que a herbivoria. No Capítulo 3, portanto, testamos a ocorrência de associação da espécie com o hábitat de solo, como proposto no capítulo 2. Para isso, adotamos uma abordagem conjunta com o estudo da distribuição espacial, possibilitando a inferência de outros mecanismos possivelmente relacionados com a dinâmica da espécie, além do micro-hábitat. Caracterizamos o padrão de distribuição espacial através do uso de metodologias espaciais de segunda-ordem complementares: K-Ripley e O-ring, e testamos a questão da associação com hábitat a partir de torus translation, uma metodologia relativamente nova que incorpora a autocorrelação espacial entre troncos coespecíficos. Além de detectarmos um padrão de distribuição agregado, com escalas críticas de agregação variáveis entre as classes de tamanho investigadas, encontramos que a espécie apresenta uma associação positiva com o tipo de solo alagável (Neossolo), onde sua densidade relativa foi 30% maior em comparação com os outros tipos de hábitat. Os adultos, além de terem se mostrado positivamente associados a esse solo, também apresentaram uma associação negativa com os solos mais arenosos e menos úmidos. Os jovens, encontrados em relativamente baixa densidade, não mostraram associação com nenhum hábitat de solo. Sugerimos que a tolerância ao encharcamento e à condições anóxicas, bem como a ocorrência de hidrocoria como uma de suas formas de dispersão, os principais fatores favoráveis à sua sobrevivência e rápido desenvolvimento ontogenético nessas condições de solo. Utilizamos informações sobre a ecologia de C. brasiliense já disponibilizadas por outros estudos para inferir ou excluir possíveis fatores relacionados com sua distribuição espacial, e a associação com hábitat e o experimento de herbivoria (Capítulo 2) como testes desses possíveis fatores. Nas considerações finais, reunimos todas as informações propostas e testadas sobre os mecanismos que atuam na dinâmica de C. brasiliense e sugerimos que o padrão espacial agregado detectado para a espécie pode ser decorrente da interação entre três fatores principais: (1) associação diferenciada com ambos tipos de solos, alagáveis e arenosos; (2) ocorrência simultânea e complementar de três agentes dispersores da espécie (vento, morcego e água) e (3) alta competição intra-específica dependente da densidade, ao longo do estágio ontogenético. Finalizamos com a construção de um modelo hipotético acompanhado de predições testáveis sobre a distribuição espacial e dinâmica da espécie. Estes resultados atribuem à ação conjunta de processos bióticos e abióticos a possível resposta para complementarmos o entendimento sobre o padrão espacial encontrado, como já sugerido por outros estudos nos trópicos. / Several theories, including different factors and mechanisms, have been postulated to explain the high tree species coexistence in tropics, which remains an unsolved question that continues to pose a challenge to plant ecologists. Population-level tree dynamics studies contribute to a better understanding of the processes acting on community-level. The aim of the present study was to investigate some aspects related to the dynamics of a common tree species, Calophyllum brasiliense Camb. (Clusiaceae), in a Restinga Alta forest in Ilha do Cardoso, Cananéia, SP. We investigated if seedling survival and fitness were related to density-dependent herbivory, tested if the species presented an association with soil habitats and characterized its spatial patterns distribution in the study area. In General Introduction (Chapter 1) we enumerated the main theories developed so far to explain high tropical diversity, which include many processes acting on the tree species dynamics. Traditionally, models focused on density-dependent factors were frequently contrasted with models based on habitat or niche partitioning, but we know nowadays that both are acting simultaneously to determine community structure. For this reason we described in general lines both models. We also discussed how detected spatial patterns of a species may give account for underlying processes responsible for the generated patterns and the need of experimental tests after such inferences. Additionally, we resumed Janzen-Connell model which embases next chapter and was also the starting point of this dissertation. We did a brief contextualization about Janzen-Connell model and reviewed main results of investigations of its effects in others study areas using alternatives methodologies. As the next two chapters were developed in the same study area and with the same species, we also included their descriptions in the general introduction. In order to test the Janzen-Connell model (Chapter 2), we designed an experiment to evaluate C. brasiliense seedlings survival and fitness under three treatments: protection against herbivory, distance from conespecific adults and tree parental density. We found that the distance and density-dependent effects did not act as predicted by the model for our study species. Despite the high herbivory damages it suffered, its seedlings showed tolerance and compensatory growth responses. We proposed that pressure by host-specific herbivores seems to be widespread in the study area instead of aggregated around conespecific densities. Patterns detected also suggest that soil moisture is a better predictor for the species survival than herbivory. On chapter 3, therefore, we tested if density of this species presented any association with soil habitats, as suggested in chapter 2. We used an approach conjunct with the spatial distribution, permitting the inference of other underlying processes possibly related to the species dynamics besides the micro-habitat. We characterized the spatial distribution patterns using two complementary second-order point pattern statistics, K-Ripley and O-ring, and tested the habitat association using the torus translation procedure that incorporates spatial autocorrelation between conespecific stems. Besides detecting clumped distribution patterns, with variable critic scales with the analyzed size classes, we also detected a positive habitat association with temporally flooded soil (Neossolo), where its relative density was 30% greater comparing to others soils types. Adult stage was also positively associated with Neossolo and, in the other hand, was negatively associated with Espodossolo arênico, which is characterized by lower moisture soil levels. Young stage corresponded to only a quarter of all species stems and did not show any association with soils habitats. We suggest that flooding and anoxic conditions tolerance, as well as the occurrence of hidrocory among its dispersion types, the main factors favorable to survival and fast ontogenetic development in these soil conditions. We used information about C. brasiliense ecology from other studies to infer or exclude possible related factors with its spatial distribution and the habitat association and Janzen-Connell tests to complement these supposed factors. On Final Considerations we synthesize all proposed and tested information about underlying processes acting on C. brasiliense dynamics and suggest that the clumped spatial pattern detected may be an interaction result of three main factors: (1) differential association with both temporally flooded and unflooded soils, (2) occurrence of three simultaneous and complementary seed dispersal agents (gravity, bats and water) and (3) high density-dependent intra-specific competition through ontogenetic stages. We finalize proposing a hypothetic scenario with testable predictions about the species spatial pattern detected to C. brasiliense in the study area. These results attribute to grouping acting effects of both abiotic and biotic processes the possible answer to complement our understanding about tree spatial patterns founded, as suggested by other studies in the tropics.

Les mécanismes expliquant le maintien de la diversité végétale dans les forêts tropicales sont mal connus. Une hypothèse particulièrement étudiée est l’hypothèse Janzen-Connell qui postule que ces mécanismes sont essentiellement causés par les interactions entre les plantes et leurs ennemis naturels, en particulier les organismes pathogènes. Dans cette thèse, nous nous sommes donc intéressés aux agents pathogènes présents dans les sols d’une forêt guyanaise et à leur lien de spécificité avec les plantes. Dans le cas où l’hypothèse Janzen-Connell serait vérifiée, on peut s’attendre à ce que les plantes structurent les communautés de micro-organismes pathogènes. Nos travaux se sont focalisés sur les Oomycètes et en particulier les Phytophthora, pathogènes des arbres très importants, mais nous nous sommes aussi intéressés aux champignons pathogènes. Ainsi, nous avons développé et comparé des jeux d’amorces PCR spécifiques des Phytophthora et des Péronosporomycètes afin d’étudier ces organismes par metabarcoding. Ces amorces ont ensuite servi à étudier la diversité des communautés de Phytophthora dans des échantillons de sols de deux sites forestiers de Guyane Française prélevés au pied d’arbres appartenant à 10 familles végétales. Une faible diversité a été retrouvée, avec seulement 8 taxons en tout, et la très large dominance d’un complexe d’espèces Phytophthora heveae. La structuration par la plante-hôte de ces communautés est plutôt faible. Dans une étude complémentaire, nous avons analysé la diversité des Oomycètes et des champignons pathogènes dans les sols et les litières de six plantations monospécifiques et au sein d’une forêt naturelle de Guyane. La structuration par l’hôte s’est avérée nulle pour les Oomycètes et faible pour les champignons pathogènes. Enfin, nous n’avons pas réussi à déclencher expérimentalement des mortalités ou dépérissements par des Oomycètes sur le wacapou, une espèce d’arbre guyanaise, via des inoculations de sols de forêt. Au final, les résultats de cette thèse ne supportent pas l’hypothèse selon laquelle les Oomycètes seraient d’importants acteurs du maintien de la diversité végétale dans les forêts tropicales. Par ailleurs, ils nous interrogent sur la faible diversité de ce groupe de microorganismes dans les sols et litières dans un hotspot de diversité végétale. / The mecanisms implied in the maintenance of plant diversity in tropical forests are still poorly known. One particularly studied hypothesis is the Janzen-Connell hypothesis, which posits that these mecanisms are essentially caused by the interactions between plant and their natural enemies, including pathogenic organisms. In this thesis, we looked at the pathogenic organisms present in the soils of a Guyanese forest, and the specificity of their interactions withplants. In the case where the Janzen-Connell hypothesis would be verified, we could expect that pathogenic micro-organisms communities would be structured by plants. Our works focused on Oomycetes and especially the Phytophthora, which are very important pathogens of trees, but we also took an interest on pathogenic Fungi. Thus, we developed PCR primer sets specific of the Phytophthora and Peronosporomycete groups, in order to study these organismsthroughmetabarcoding. These primers were then used to investigate the community of Phytophthora in soils sampled from two French Guiana sites, near trees belonging to 10 families. A low diversity was described, with a total of only 8 taxas, and the overwhelming dominance of the species complex P. heveae. A weak host effect was detected. In a complementary study, we looked at the diversity of Oomycetes and Fungi in soils and litters of six monospecific tree plantations and a Guianese natural forest. Structuration by host appeared to be null for Oomycetes and weak for pathogenic Fungi. Finally, we did not success in trying to experimentally provoke, through forest soil inoculations, Janzen-Connell mortalities due to Oomycetes on the Wacapou, a Guianese tree species. In the end, the results of this thesis do not support the hypothesis that Oomycetes may be important agents of the maintenance of tree diversity in tropical forests. Moreover, they bring some questions about the low diversity of this group of micro-organisms in a tree diversity hotspot.

Muitas teorias, englobando diferentes fatores e mecanismos, já foram postuladas para explicar a alta coexistência de espécies arbóreas nos trópicos, a qual permanece como uma questão intrigante e subentendida na ecologia vegetal. O estudo da dinâmica de árvores ao nível populacional contribui e embasa, por sua vez, o entendimento desses fatores e mecanismos atuando ao nível da comunidade. O objetivo geral do presente estudo foi investigar alguns aspectos relacionados com a dinâmica de uma espécie arbórea comum Calophyllum brasiliense Camb.(Clusiaceae) em floresta de restinga alta Ilha do Cardoso, Cananéia, SP. Para isso, além de termos testado se a sobrevivência e o desempenho de suas plântulas estavam relacionados com níveis de herbivoria dependentes da densidade, testamos se a ocorrência da espécie apresentava associação com hábitats de solo e caracterizamos sua distribuição espacial na área de estudo. Na Introdução Geral (Capítulo 1) nós enumeramos as principais teorias já propostas para explicar a alta diversidade tropical, as quais incluem diversos mecanismos atuantes na dinâmica de espécies arbóreas. Tradicionalmente, modelos que focam em fatores dependentes da densidade eram freqüentemente contrastados com modelos baseados na segregação de hábitats e nichos para explicar a coexistência de espécies, ainda que atualmente sabe-se que atuem concomitantemente na estruturação das comunidades. Por esse motivo, os descrevemos em linhas gerais. Apresentamos, também, como o estudo da distribuição espacial de uma espécie pode dar indícios de processos subjacentes responsáveis pelos padrões gerados, os quais devem ser inferidos e posteriormente testados. Adicionalmente, resumimos uma teoria em especial, a qual embasa o capítulo seguinte e foi o ponto inicial dessa dissertação: o Modelo Janzen-Connell. Além de termos explorado brevemente seu contexto conceitual, também revisamos os principais resultados de investigações de seus efeitos em outras áreas de estudo e com diferentes metodologias. Como os dois demais capítulos foram desenvolvidos utilizando a mesma área e espécie de estudo, também incluímos neste capítulo suas respectivas descrições. Para testar o modelo Janzen-Connell (Capítulo 2), nós delineamos um experimento no qual avaliamos os danos por herbivoria, a mortalidade e o desempenho das plântulas de C. brasiliense, sob diferentes tratamentos: proteção contra herbivoria, distância e agrupamento de adultos coespecíficos. Encontramos que os efeitos dependentes da distância e da densidade não atuaram como previsto pela teoria para a espécie de estudo, a qual, apesar de ter sofrido altos danos por herbivoria, mostrou-se tolerante e apresentou crescimento compensatório em resposta a estes. Além de termos proposto que a pressão por seus herbívoros especialistas parece encontrar-se amplamente distribuída na área de estudo, e não agregada ao redor de densidades de coespecíficos, também sugerimos que o micro-hábitat, especialmente a umidade do solo parece ser melhor preditora da sobrevivência da espécie do que a herbivoria. No Capítulo 3, portanto, testamos a ocorrência de associação da espécie com o hábitat de solo, como proposto no capítulo 2. Para isso, adotamos uma abordagem conjunta com o estudo da distribuição espacial, possibilitando a inferência de outros mecanismos possivelmente relacionados com a dinâmica da espécie, além do micro-hábitat. Caracterizamos o padrão de distribuição espacial através do uso de metodologias espaciais de segunda-ordem complementares: K-Ripley e O-ring, e testamos a questão da associação com hábitat a partir de torus translation, uma metodologia relativamente nova que incorpora a autocorrelação espacial entre troncos coespecíficos. Além de detectarmos um padrão de distribuição agregado, com escalas críticas de agregação variáveis entre as classes de tamanho investigadas, encontramos que a espécie apresenta uma associação positiva com o tipo de solo alagável (Neossolo), onde sua densidade relativa foi 30% maior em comparação com os outros tipos de hábitat. Os adultos, além de terem se mostrado positivamente associados a esse solo, também apresentaram uma associação negativa com os solos mais arenosos e menos úmidos. Os jovens, encontrados em relativamente baixa densidade, não mostraram associação com nenhum hábitat de solo. Sugerimos que a tolerância ao encharcamento e à condições anóxicas, bem como a ocorrência de hidrocoria como uma de suas formas de dispersão, os principais fatores favoráveis à sua sobrevivência e rápido desenvolvimento ontogenético nessas condições de solo. Utilizamos informações sobre a ecologia de C. brasiliense já disponibilizadas por outros estudos para inferir ou excluir possíveis fatores relacionados com sua distribuição espacial, e a associação com hábitat e o experimento de herbivoria (Capítulo 2) como testes desses possíveis fatores. Nas considerações finais, reunimos todas as informações propostas e testadas sobre os mecanismos que atuam na dinâmica de C. brasiliense e sugerimos que o padrão espacial agregado detectado para a espécie pode ser decorrente da interação entre três fatores principais: (1) associação diferenciada com ambos tipos de solos, alagáveis e arenosos; (2) ocorrência simultânea e complementar de três agentes dispersores da espécie (vento, morcego e água) e (3) alta competição intra-específica dependente da densidade, ao longo do estágio ontogenético. Finalizamos com a construção de um modelo hipotético acompanhado de predições testáveis sobre a distribuição espacial e dinâmica da espécie. Estes resultados atribuem à ação conjunta de processos bióticos e abióticos a possível resposta para complementarmos o entendimento sobre o padrão espacial encontrado, como já sugerido por outros estudos nos trópicos. / Several theories, including different factors and mechanisms, have been postulated to explain the high tree species coexistence in tropics, which remains an unsolved question that continues to pose a challenge to plant ecologists. Population-level tree dynamics studies contribute to a better understanding of the processes acting on community-level. The aim of the present study was to investigate some aspects related to the dynamics of a common tree species, Calophyllum brasiliense Camb. (Clusiaceae), in a Restinga Alta forest in Ilha do Cardoso, Cananéia, SP. We investigated if seedling survival and fitness were related to density-dependent herbivory, tested if the species presented an association with soil habitats and characterized its spatial patterns distribution in the study area. In General Introduction (Chapter 1) we enumerated the main theories developed so far to explain high tropical diversity, which include many processes acting on the tree species dynamics. Traditionally, models focused on density-dependent factors were frequently contrasted with models based on habitat or niche partitioning, but we know nowadays that both are acting simultaneously to determine community structure. For this reason we described in general lines both models. We also discussed how detected spatial patterns of a species may give account for underlying processes responsible for the generated patterns and the need of experimental tests after such inferences. Additionally, we resumed Janzen-Connell model which embases next chapter and was also the starting point of this dissertation. We did a brief contextualization about Janzen-Connell model and reviewed main results of investigations of its effects in others study areas using alternatives methodologies. As the next two chapters were developed in the same study area and with the same species, we also included their descriptions in the general introduction. In order to test the Janzen-Connell model (Chapter 2), we designed an experiment to evaluate C. brasiliense seedlings survival and fitness under three treatments: protection against herbivory, distance from conespecific adults and tree parental density. We found that the distance and density-dependent effects did not act as predicted by the model for our study species. Despite the high herbivory damages it suffered, its seedlings showed tolerance and compensatory growth responses. We proposed that pressure by host-specific herbivores seems to be widespread in the study area instead of aggregated around conespecific densities. Patterns detected also suggest that soil moisture is a better predictor for the species survival than herbivory. On chapter 3, therefore, we tested if density of this species presented any association with soil habitats, as suggested in chapter 2. We used an approach conjunct with the spatial distribution, permitting the inference of other underlying processes possibly related to the species dynamics besides the micro-habitat. We characterized the spatial distribution patterns using two complementary second-order point pattern statistics, K-Ripley and O-ring, and tested the habitat association using the torus translation procedure that incorporates spatial autocorrelation between conespecific stems. Besides detecting clumped distribution patterns, with variable critic scales with the analyzed size classes, we also detected a positive habitat association with temporally flooded soil (Neossolo), where its relative density was 30% greater comparing to others soils types. Adult stage was also positively associated with Neossolo and, in the other hand, was negatively associated with Espodossolo arênico, which is characterized by lower moisture soil levels. Young stage corresponded to only a quarter of all species stems and did not show any association with soils habitats. We suggest that flooding and anoxic conditions tolerance, as well as the occurrence of hidrocory among its dispersion types, the main factors favorable to survival and fast ontogenetic development in these soil conditions. We used information about C. brasiliense ecology from other studies to infer or exclude possible related factors with its spatial distribution and the habitat association and Janzen-Connell tests to complement these supposed factors. On Final Considerations we synthesize all proposed and tested information about underlying processes acting on C. brasiliense dynamics and suggest that the clumped spatial pattern detected may be an interaction result of three main factors: (1) differential association with both temporally flooded and unflooded soils, (2) occurrence of three simultaneous and complementary seed dispersal agents (gravity, bats and water) and (3) high density-dependent intra-specific competition through ontogenetic stages. We finalize proposing a hypothetic scenario with testable predictions about the species spatial pattern detected to C. brasiliense in the study area. These results attribute to grouping acting effects of both abiotic and biotic processes the possible answer to complement our understanding about tree spatial patterns founded, as suggested by other studies in the tropics.

The decline in range and density of frugivorous birds worldwide could have consequences for the functioning of ecosystem processes such as seed dispersal. I endeavoured to determine the effects of bird declines on seed dispersal and assess the current status of dispersal in New Zealand. My first aim was to determine whether decreased bird density can reduce competitive interactions between birds, leading to birds concentrating on higher-reward fruit species. I measured fruit removal rates and fruit preferences of birds in Canterbury forest remnants with high (Hay and Prices) and low (Lords and Kaituna) bird densities. Removal rates of Melicytus ramiflorus and Coprosma spp. were lower at low-bird sites, and the size of this effect was greater for low-reward plant species. Coprosma areolata (a low-sugar fruit) had limited dispersal at Kaituna (59% fruit removed by end of 2012 season), compared with 92% removal at Hay, whereas 99-100% of M. ramiflorus and C. robusta (higher-sugar) fruits were removed at both sites. My second aim was to determine whether seed dispersal by introduced possums (Trichosurus vulpecula) can compensate for bird declines. I analysed seeds present in possum and bird faecal samples collected at Kowhai Bush, Kaikoura. Possums dispersed <3% of the total seeds, much less than bellbirds (21%), silvereyes (16%), song thrushes (33%) and blackbirds (28%). Possums also destroyed approximately 15% of seeds found in faeces, reduced the germination of gut-passed C. robusta seed to half of that from bird faeces (30% vs. 60-70%), and did not swallow fruits any larger than those moved by the much smaller birds (c. 7 mm diameter). My third aim was to determine the relationship between percentage of fruits dispersed and distance from parent tree for three large-seeded trees; Beilschmiedia tawa, Elaeocarpus dentatus, and Prumnopitys ferruginea. I fitted dispersal kernels to the observed dispersal distances out to 50 m for both undispersed whole fruits and seeds consumed by a bird, and found that dispersal quantity below the parent tree strongly underestimates total dispersal quantity. The average percentage dispersed overall was 81% for B. tawa, 75% for E. dentatus and 91% for P. ferruginea, and for all species finding only 11-18% clean seeds under the parent tree would correspond to an overall percent dispersed of at least 50% of the whole seed crop. My final aim was to determine the consequences of dispersal failure for recruitment in three plant species; Elaeocarpus hookerianus, Ripogonum scandens, and C. robusta. I compared the fate of dispersed and undispersed seeds in a manipulative field experiment. Exclusion of mammals plus removal of fruit pulp had the greatest effect on survival, while the combined effects of dispersal failure (under parent, high density, whole fruit) and inclusion of mammals decreased the number of live seedlings present at the final count by 75-92%. Overall, most native plants were receiving adequate dispersal and there was little evidence for strong risks to plant regeneration from dispersal failure, despite the reduced bird densities and ongoing negative effects of introduced mammals in New Zealand.

Natural enemies are ubiquitous in nature. In many communities, natural enemies have a major effect on the diversity of their prey. Their effects are very diverse: they can promote or undermine the ability of their prey to coexist through a variety of mechanisms. As such, an important step in understanding how diversity is maintained will be to understand how different forms of predator behavior affect prey coexistence. In this dissertation, I study how two major types of predators affect plant coexistence in two different communities. First, I study natural enemies in tropical forests, using both theory and empirical work. In tropical forests, most natural enemies are thought have a narrow host range, and be distance-responsive (i.e., mainly harm seeds and seedlings that are near adults of their main host). Previous theoretical work has shown that specialized natural enemies can maintain diversity of their prey, whether or not they are distance-responsive. However, it is unknown whether specialist natural enemies are more or less able to promote prey coexistence if they are distance-responsive. Using theoretical models, I show that distance-responsive predators are less able to maintain diversity. Additionally, I show that habitat partitioning does not interfere with the ability of distance-responsive predators to maintain diversity, even if it causes seedling survival to be highest near conspecific adults. From an empirical aspect, I studied the host range of seed-associated fungi. Soilborne microbes, such as fungi, are thought to play an important role in maintaining diversity in tropical forests. However, the microbial community itself is often treated as a black box, and little is known about which microbes are causing major effects, or how 8 specialized seed-microbe associations are. Here I use experimental inoculations to examine the host range and effect of a guild of seed-associated fungi that are thought to be mainly pathogens. I show that fungal species are differentially able to colonize different seed species, and have species-specific effects on seed germination. I show that in many cases, plant phylogeny, and to a lesser extent fungus phylogeny, are good predictors of colonization. Finally, I study how an optimally foraging granivore can promote (or undermine) coexistence amongst annual plants, using theory. Optimal foraging theory is one of the major theories for how predators behave; despite this, little is known about whether an optimally foraging predator could promote coexistence amongst a diverse community of prey. Previous models have shown than two species can coexist due to optimal foraging, but did not test whether multiple prey can coexist, nor if the effect is altered by environmental variation. Here, I show that if the predators specialize on different prey at different times, the predators can allow multiple prey species to coexist. In this case, environmental variation has little effect on the ability of predators to maintain diversity. If the predators are generalists, they cannot maintain diversity. Additionally, I show that generalist predators will create a negative storage effect, undermining coexistence.

Submitted by Amanda Silva (amanda.osilva2@ufpe.br) on 2015-03-11T15:00:49Z No. of bitstreams: 2 Tese Edgar A E S Silva .pdf: 1480606 bytes, checksum: 5f83b405866d21135bd0eb3bac6eae6e (MD5) license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) / Made available in DSpace on 2015-03-11T15:00:49Z (GMT). No. of bitstreams: 2 Tese Edgar A E S Silva .pdf: 1480606 bytes, checksum: 5f83b405866d21135bd0eb3bac6eae6e (MD5) license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Previous issue date: 2014-02-26 / FACEPE; CNPq; CAPES / Dentre as hipóteses criadas para explicar a manutenção da diversidade de árvores em florestas tropicais, uma das mais aceitas é a hipótese de Janzen-Connell. Essa hipótese propõe que inimigos naturais diminuem a probabilidade de sobrevivência de sementes e plântulas quando estas estão adensadas ou próximas ao adulto parental. Entretanto, é provável que esse mecanismo proposto por Janzen-Connell seja menos intenso em florestas fragmentadas. Esta tese teve como objetivo avaliar se a alta densidade de determinadas espécies arbóreas em paisagens fragmentadas é resultado de um relaxamento na mortalidade dependente da densidade e distância. Foram realizados três estudos em uma paisagem fragmentada de Floresta Atlântica, comparando pequenos fragmentos florestais com áreas de floresta madura no interior de um grande remanescente. No primeiro, foram analisados os padrões de mudança nas densidades de populações arbóreas no estágio de plântula e adulto para investigar quais espécies tendem a predominar em pequenos fragmentos florestais. No segundo, foi analisada a distribuição espacial de árvores adultas e jovens para inferir se a distância entre indivíduos coespecíficos é menor nos pequenos fragmentos. No terceiro, foi realizado um experimento de predação de sementes com a espécie Tapirira guianensis para avaliar se existe uma redução na predação de sementes em pequenos fragmentos florestais. Em síntese, os resultados demonstram que: (1) apenas um pequeno grupo de espécies pioneiras apresenta um aumento na densidade populacional em pequenos fragmentos florestais, enquanto muitas espécies tendem a se extinguir localmente ou apresentar abundâncias extremamente baixas nesses fragmentos; (2) não existe uma redução na distância entre jovens e adultos coespecíficos nos pequenos fragmentos ao analisar todas as espécies, apenas quando analisadas somente as espécies pioneiras; (3) exceto pela pequena redução na distância média (2 m) entre adultos coespecíficos, não foi observada uma agregação nas populações nos pequenos fragmentos florestais; (4) não existe um relaxamento na predação de sementes dependente da distância em Tapirira guianensis. Os resultados obtidos não indicam que as mudanças observadas na estrutura das assembleias arbóreas nos pequenos fragmentos resultam de relaxamentos na mortalidade dependente da densidade e distância. Entretanto, mesmo após 200 anos, esses fragmentos ainda apresentam uma redução na diversidade de árvores.

No description available.