Das Wissen über Mechanismen, die einen Einfluss auf Muster der Artenvielfalt und biotische Interaktionen haben, ist grundlegend für den Schutz von Biodiversität. Darüber hinaus kann es von direktem ökonomischem Nutzen sein, zum Beispiel im biologischen Pflanzenschutz oder bei Bestäubungsdienstleistungen. Die Größe eines Organismus kann ein solcher Faktor sein, der die Artenzahl und Interaktionen der assoziierten Organismen beeinflusst, denn große Organismen sind auffälliger als kleine und ihr Angebot an Ressourcen und Nischen für mit ihnen assoziierte Organismen ist oft reicher. Bezogen auf Pflanzen könnte daher die Größe einer Pflanze einen erheblichen Einfluss auf die Artenzahl der mit ihr assoziierten Arthropoden und ihre biotischen Interaktionen wie Herbivorie oder Bestäubung haben. Trotzdem ist der Einfluss der Pflanzengröße auf mutualistische und antagonistische Interaktionspartner der Pflanze und der sich daraus ergebende Einfluss auf die reproduktive Fitness der Pflanze bisher nicht umfassend und unter standardisierten Bedingungen untersucht worden. In der vorliegenden Studie wurden die Auswirkungen der Pflanzengröße auf die Artenzahl von Herbivoren, deren Gegenspielern und Bestäubern untersucht, sowie die Auswirkungen dieser Interaktionspartner auf die Pflanzenfitness. Dabei wurde zusätzlich zwischen endophagen und ektophagen Herbivoren und deren Gegenspielern unterschieden. Außerdem wurden die Herbivoren einzelner Pflanzenkompartimente und deren Gegenspieler separat analysiert. Des Weiteren wurde der Einfluss der Pflanzengröße auf den Herbivorieschaden an den verschiedenen Pflanzenkompartimenten und deren Einfluss auf die reproduktive Fitness der Pflanze, d.h. auf ihre Samenzahl, Tausendkorngewicht und Samengesamtgewicht, untersucht. Zuletzt wurde besonderes Augenmerk auf den Einfluss der Pflanzengröße auf mutualistische und antagonistische Blütenbesucher und deren Einfluss auf die reproduktive Fitness gelegt und untersucht, ob und inwiefern die reproduktive Fitness letztendlich von der Pflanzengröße abhängig ist. Zur Untersuchung dieser Fragen wurde ein „Common Garden“-Experiment angelegt. Um einen interspezifischen Pflanzengrößengradienten zu erzeugen, wurden 21 annuelle Pflanzenarten aus der Familie der Kreuzblütler (Brassicaceae) ausgewählt, deren Größe von 10 bis 130 cm reichte (gemessen als Pflanzenhöhe vom Boden bis zur Spitze). So konnten die Einflüsse des Habitats und der umgebenden Landschaft für alle Pflanzenarten standardisiert und trotzdem ein breiter Gradient realisiert werden. Dadurch hebt sich diese Studie von den bisherigen ab, die den Effekt von meist intraspezifischer Pflanzengröße auf die assoziierten Tiere anhand wild wachsender Pflanzen untersucht haben. Pflanzengröße sowie Zahl, Biomasse und Größe der unterschiedlichen überirdischen Pflanzenkompartimente (Blüten, Schoten, Blätter, Stängel) sowie Blütendeckung und -farbe wurden aufgenommen. Der Herbivorieschaden an diesen Pflanzenkompartimenten und die reproduktive Fitness (Samenzahl, Tausendkorngewicht und Gesamtsamengewicht) wurden gemessen. An und in Blüten, Schoten, Blättern und Stängeln wurden herbivore, räuberische, parasitäre und bestäubende Arthropoden gezählt. Die Pflanzengröße hatte einen positiven Einfluss auf die Artenzahl von Herbivoren, deren Gegenspielern und Bestäubern. Das traf ebenso auf endophage und ektophage sowie auf mit Blättern und Schoten assoziierte Herbivore und deren Gegenspieler zu. Des Weiteren konnte ein Anstieg des Herbivorieschadens an Blüten und Schoten mit zunehmender Pflanzengröße festgestellt werden, wohingegen der Schaden an Blättern und Stängeln von der Biomasse des entsprechenden Kompartiments positiv beeinflusst wurde. Der Schaden an Blüten hatte den stärksten Einfluss auf die reproduktive Fitness und reduzierte neben der Samenzahl auch das Tausendkorngewicht und das Gesamtsamengewicht der Pflanze. Die genaue Analyse der blütenbesuchenden Insekten ergab einen positiven Einfluss der Pflanzengröße auf die Abundanz und Artenzahl von Bestäubern (allerdings nicht bei extrem großem Blütenangebot), wie auch auf die Abundanz der adulten und juvenilen Rapsglanzkäfer und deren Parasitierungsrate. Steigende Rapsglanzkäferzahlen verringerten die Samenzahl sowie das Tausendkorngewicht, während die Bestäuber sich lediglich auf die Samenzahl positiv auswirkten. Insgesamt führte ein Anstieg der Pflanzenhöhe zu einer Abnahme des Tausendkorngewichts, aber nicht zu einer Veränderung der Samenzahl oder des Gesamtsamengewichts, was auf einen Ausgleich der Effekte von zunehmender Antagonistenzahl und zunehmender Mutualistenzahl hindeutet. Großen Pflanzen entstehen also durch ihre Auffälligkeit und Attraktivität für Herbivore hohe Fitnesskosten, wobei insbesondere der Blütenschaden durch Rapsglanzkäfer einen starken negativen Einfluss auf Samenzahl, Tausendkorngewicht und Gesamtsamengewicht hat. Diesen Fitnesskosten großer Pflanzen wirkt der Nutzen durch ihre Auffälligkeit und Attraktivität für Bestäuber entgegen, die die Samenzahl positiv beeinflussen. Hinsichtlich der Samenzahl sollten also große Pflanzen gegenüber kleineren im Vorteil sein, wenn die Insektengemeinschaft des Habitats von Bestäubern dominiert wird. Wird sie aber von herbivoren Blütenbesuchern dominiert, sollten kleine Pflanzen gegenüber großen einen Vorteil haben. Im Gegensatz dazu sollten große Pflanzen immer einen Nachteil bezüglich des Tausendkorngewichts haben, das von Antagonisten, nicht aber von Mutualisten beeinflusst wurde. Der Einfluss der Pflanzengröße auf biotische Interaktionen wurde bisher oft unterschätzt, obwohl er sich auf komplexe Weise über die mutualistischen und antagonistischen Insekten auf die reproduktive Fitness der Pflanze auswirkt.

Os ecossistemas de ilhas, continentais ou oceânicas, são considerados os mais sensíveis à perturbação humana. A maioria das ilhas costeiras do Brasil sofreu fortes alterações na fauna e flora silvestre. A Ilha Anchieta (Ubatuba, SP) é uma Área Protegida com um longo histórico de perturbação, tendo sido amplamente desmatada no passado e sofrido a introdução de animais. Sabe-se que a fauna possui um papel chave na composição e estrutura da comunidade vegetal, favorecendo algumas espécies e prejudicando outras. Sendo assim, a perda ou alteração dos processos de interações entre animais e plantas afetam na estrutura e composição de espécies. Este estudo analisa as interações antagônicas entre planta e animais como possíveis fatores limitantes no processo de regeneração natural em três ambientes com distintos estágios sucessionais presentes na Ilha Anchieta: campo aberto (CA) e florestas ombrófilas rala (FOR) e densa (FOD). Para isso foram testados nos três ambientes os processos pós-dispersão de sementes: a predação de sementes pós-dispersas, a germinação do banco de sementes e a herbivoria. A predação de sementes apresentou forte efeito espacial (FOD>FOR>CA) e sazonal, com maiores proporções de predação de sementes no período de escassez de alimento (estação seca). Quantitativamente o banco de sementes não representa um fator limitante, porém foi composto por uma baixa diversidade de espécies lenhosas. A capacidade de germinação do banco de sementes do solo foi similar entre os ambientes florestais apresentando menor emersão de plântulas no campo aberto, provavelmente associado ao intenso escoamento superficial no local. Adicionalmente, a mortalidade de juvenis transplantados foi de 72,27%, sendo superior nas parcelas abertas aos vertebrados para todas as espécies em todos os ambientes, demonstrando um forte efeito negativo dos herbívoros vertebrados na comunidade vegetal. Sendo assim, foram detectados distintos gargalos atuando simultaneamente na regeneração natural nos três ambientes da Ilha Anchieta, sendo necessário o estabelecimento de práticas de manejo visando minimizar os efeitos abióticos no campo aberto e floresta ombrófila rala, favorecendo as interações mutualísticas e inibindo a atividade de animais antagônicos nas florestas ombrófilas rala e densa, acelerando assim o processo de regeneração natural na Ilha Anchieta. Considerar simultaneamente os obstáculos no processo de regeneração nos auxiliará a traçar práticas de restauração e recuperação de áreas degradadas mais efetivas e viáveis economicamente. / Island ecosystems, either continental or oceanic, are considered the most sensible to anthropogenic influences. Most Brazilian coastal islands have their original fauna and flora composition altered. The Anchieta Island (southeast Brazil) is a Protected Area that suffered deforestation and introduction of alien species. It is known that the fauna plays a key role in composing and structuring the vegetal community, favoring some species and harming others. Thus, the loss or change of animal-plant interactions affects the framework and composition of species. We simultaneously analyzed the antagonistic animal-plant interactions as possible limiting factors in the natural regeneration on tree successional stages of the Anchieta Island: the old fields (OF), the early secondary forest (ESF) and old growth forest (OGF). Therefore, in each habitat we are evaluating post-dispersal seed predation processes: post-dispersal seed predation, soil seed bank, and the herbivory. We found spatial (OGF>ESF>OF) and temporal effects on seed predation, with highest rates on dry season. Quantitatively, soil seed bank did not represent a limiting factor; but qualitatively, it was composed by few woody species. Our data pointed to the absence of a viable soil seed bank in the OF, probably due to an intense runoff. The forested habitats presented similar soil bank. Moreover, the total mortality of saplings was 72.27%, being higher on the unfenced treatment for all species and in all habitats, showing a very strong negative effect of vertebrate herbivores on the vegetal community. Therefore, we are detecting distinct bottlenecks acting simultaneously in the natural regeneration process in all of the successional stages of the Anchieta Island. Our results showed that management actions are required, aiming to both minimize the abiotic effects on the old fields and in the early secondary forest and to favor the mutual interactions and inhibit the activity of antagonic animals in the old growth and early secondary forests, accelerating, this way, the natural regeneration process of the Anchieta Island. Considering simultaneously the obstacles in the regeneration process will help to define restoration and recuperation procedures of degraded areas more effective and affordable.

No description available.

No description available.

No description available.

Frugivorous insects, specialised herbivores that consume fruit and seeds, are considered detrimental to host plant fitness. Their direct link to genetic fitness via consumption of plant reproductive tissue, and their negative socioeconomic association with agriculture exacerbates their harmful status. However, empirical testing of insect frugivore effects on host plants, and ecological research on the contribution of insect frugivores to multitrophic frugivory systems, is lacking. In the current study, direct effects of a non-mutualistic, insect frugivore/host plant system were tested and results showed variable effects. Beneficial, detrimental, but predominantly neutral effects on germination and seed production were observed between the Queensland fruit fly (Bactrocera tryoni) and tomato and capsicum plants. Significant effects on seed production were unexpected because infestation occurs after seed set. It was also found that eggplant, although a recorded host of B. tryoni, is inconsistent in its ability to sustain B. tryoni larvae through to its final instar. These results confirmed a simplification and presumption associated with insect frugivore (specifically fruit fly)/host plant interactions. Larval movement, infestation-induced fruit decay, pulp removal and germination were then investigated. For all hosts (tomato, apple and paw paw), treatments infested by B. tryoni decayed significantly quicker and to a greater extent than uninfested treatments, with obvious but variable changes to the texture and appearance. The movement of B. tryoni larvae, pattern of infestation-induced decay and pulp removal was unique and host dependent for all hosts. Only seeds from infested tomato were shown to germinate during the experiment. This indicated that host fruit characteristics are responsible, in part, for variable direct effects on host plant fitness by insect frugivores. Variable direct effects between insect frugivores and host plants, combined with the more rapid decay of infested fruits is likely to have implications for seed dispersal and seed predation by a third trophic level. The characteristics of fruit that are changed by infestation by an insect frugivore were then tested for their effect on a vertebrate frugivore, to illustrate the importance of recognising multitrophic interactions and indirect effects in frugivory. Specifically, seed predating rodents were incorporated into the study and their response to infested and uninfested fruits were recorded, as well as their reaction to the changes in fruit caused by insect frugivores (i.e. texture, smell, larvae presence and sound). Apple and pear infested with B. tryoni larvae were found to attract rodents, while infested tomato and paw paw had a neutral effect on the native rats. This differed from the predominant finding in the literature, which was a deterrent effect on avian seed dispersers. Vertebrate response to fruit infested with insect frugivores therefore, is variable. Assessing the indirect effect of insect frugivores on host plant fitness by attracting or deterring another trophic level requires knowledge of the direct effect between the introduced trophic level and the host plant. For example, the attraction of a seed predator may be as detrimental to host plant fitness as the deterrence of a seed disperser. This illustrates the complexity associated with assessing insect frugivore effects on host plant fitness. Results also indicated that differences in pulp texture, caused by infestation, have a significant effect on rodent preference for infested or uninfested treatments. Pulp texture is likely to effect rodent foraging efficiency, whereas the presence of B. tryoni larvae was observed to be inconsequential to rodent response to fruits. For rodents, and indeed any trophic level motivated by foraging efficiency, this finding raises the issue that for long lived fruiting plants, outside factors such as food abundance and competition for food, may cause a variable response to fruits infested by insect frugivores. From these investigations it has become apparent that insect frugivores are not consistently harmful to host plant fitness, as suggested by their negative stigma, but are likely to contribute variable effects, directly and indirectly, on multiple components of plant fitness and multitrophic frugivory systems.

After European settlement, Australian rangelands especially in South Australia underwent significant changes because of the main land use of pastoralism. Many studies have revealed that the plant communities are negatively effected by herbivory mainly by sheep. The main aim of this study is to separate the different effects of sheep, rabbits and kangaroos. This was examined by survey supported by experimental and modelling research. A 32,000 km² area previously surveyed by Tiver and Andrew (1997) in eastern South Australia was re-surveyed to monitor populations of perennial plant species at sites of various intensity of grazing by sheep, rabbits and kangaroos (goats populations are low in the study area), the most important vertebrate herbivores. Plant population data were collected in both sheep paddocks and historically ungrazed by sheep (road reserves) by using the Random Walk method and analyzed using Generalized Linear Modelling (GLM) to separate the effects of sheep and rabbits on plant regeneration and their regeneration in response to grazing. These data were also compared to similar data collected by Tiver and Andrew in 1992 (1997) to ascertain if the reduction in rabbit numbers through introduction of RCV had allowed increased regeneration. Regeneration of many species inside paddocks were negatively affected and species in roadside reserves neither did not significantly increase from 1992 to 2004. However, some species showed increase of populations in spite of sheep grazing, with some species being less susceptible than others. This research also indicates kangaroo grazing impact on some plant species. Reduction in rabbit numbers following the 1995 release of calicivirus has not been effective in restoring regeneration. Another experiment was conducted at Middleback Field Station near Whyalla to identify herbivore grazing pressure on the arid zone plant species Acacia aneura using unfenced, sheep fenced and rabbit fenced grazing exclosures. This experiment was set up with seedlings in exclosures, ten replicates of each treatment, at plots four different distances from the watering point to identify the survivorship of seedlings. Data were collected by recording canopy volumes of seedling over an 18 month period and analyzed by Residual Maximal Likelihood (REML). Seedlings both near and far from the watering point were severely effected by large herbivores, either sheep, kangaroos or both, and in a separate experiment kangaroo grazing effects on the seedling were also identified. Seedlings browsed by the rabbits were recovered better than the seedlings grazed by the large herbivores. Decreasing kangaroo activities has been noticed when the rabbit movements increased. Computer modelling was conducted to predict the future plant population structure over 500 years using a matrix population model developed by Tiver et al. (2006) and using data collected in the survey as a starting point. Extinction probabilities of populations of Acacia aneura near watering points, far from watering points and under pulse grazing scenarios were compared. Sheep grazing was found to cause eventual extinction of populations in all parts of sheep paddocks. Together, the results indicate that sheep are the major herbivore suppressing regeneration of perennial plant species. Kangaroo and rabbits have an identifiable but lesser effect. The results have implications for conservation and pastoral management. To achieve ecological sustainability of arid lands a land-use system including a network of reserves ungrazed by sheep and with control of both rabbit and kangaroo numbers will be required.

Woody plant encroachment is one of the major problems worldwide because it affects negatively the herbaceous layer, which provide forage for livestock production. However, the role of ruminants particularly browsers in the dispersal of woody plant seeds still remains a concern for farmers interested in grass production. Seedpods of various woody plant species constitute a crucial part of the diet of herbivores during the dry season because of their high nutritional quality compared to herbaceous material. The interaction of associated diet quality, seed characteristics and animal species among other factors play a pivotal part in the success of livestock faecal seeds dispersion. Furthermore, dispersed seeds that successfully grow into mature woody plants become an important source of protein for herbivores. The use of woody plants as a source forage is known to be limited by plant secondary metabolites (PSMs) such as condensed tannins. The objectives of this study were to determine 1) the effects of condensed tannins and crude protein of Vachellia tortilis and Dichrostachys cinerea pods in seed recovery and germination fed to goats, and 2) the effects of diet mixing on the feed intake of plant species by goats. In the first experiment, a total of 12 female indigenous goats and 12 female Pedi sheep were utilised in this study, with the average body weights of 29.50 kg ± 1.60 (S.E) and 28.70 kg ± 1.60, respectively. Twelve goats were grouped into two groups of six goats per group, one group was fed D. cinerea pods and the other group was fed V. tortilis pods. The group of 12 sheep were divided similarly, the one group was fed D. cinerea pods and the other group was fed V. tortilis pods. Each animal was given V. tortilis and D. cinerea pods at 2.50% of their body weight. All animals were allowed to consume D. cinerea or V. tortilis pods within 24 h, after which the remaining pods were collected and weighed. Faecal collection commenced immediately after the 24 h pods feeding and was carried on until no seeds were discovered in faeces. All faeces extracted from sheep and goats were collected daily in the morning from the faecal bags. In the second experiment, a total of 24 indigenous goats with average body weight of 26.6 kg ± 0.51 were utilised. Goats were arbitrarily selected and grouped into four groups of six goats per group (goats were placed individually in 2 m2 pens). Each group was fed one of the following diets: diet one - Searsia lancea, diet two - S. pyroides, diet three - Euclea crispa and diet four - was a combination of the three plant species (Searsia lancea, S. pyroides and Euclea crispa). Searsia lancea, S. pyroides and E. crispa branches were collected every morning prior to feeding, and were weighed before offering the animals. Refusals were gathered and weighed, and intake was calculated as distinction between weight in and refusals. Plant species foliage were analysed for crude protein, condensed tannin, acid detergent lignin, acid detergent fibre and neutral detergent fibre. During the first experiment, the cumulative percentage seed recovery of V. tortilis from goats (46.00 % ± 1.90) and sheep (52.00 % ± 2.93) was significantly higher than D. cinerea from goats (13 % ± 1.47) and sheep (24.00 % ± 1.16). Germination percentage of D. cinerea seeds that passed through the gastro-intestinal tract of goats (33.12 % ± 2.94) and sheep (36.00 % ± 2.68) was significantly higher than V. tortilis seeds that passed through the gastro-intestinal tract of goats (28.98 % ± 2.68) and sheep (23.04 % ± 2.81). Average D. cinerea (34.56 % ± 1.99) and V. tortilis (26.02 %± 2.10) seeds that went through the gastro-intestinal of goats and sheep had a significantly higher germination rate than the control (i.e. no passage through the gut; D. cinerea = 2.31 % ± 1.55, V. tortilis = 5.07 % ± 2.68). The high mean cumulative percentage seed recovery of V. tortilis (18.80 %) may be attributed to the relatively higher crude protein than D. cinerea (12.20 %). This may encourage animal seed dispersal and germination of woody plant species with relatively high crude protein content. In the second experiment, Searsia lancea contained 8.50 % CP, 21.46 % acid detergent fibre (ADF), 12.50 % ADL and 39.37 % NDF. Searsia pyroides had 9.03 % CP, 27.07 % ADF, 10.89 % ADL and 40.30 % NDF. Euclea crispa had 6.19 % CP, 26.20 % ADF, 16.63 % ADL and 30.02 % NDF. Mixed diet (combination of the three plant species) had 8.96 % CP, 23.72 % ADF, 11.13 % ADL and 38.28 % NDF. Searsia lancea had 2.70 % of CTs while S. pyroides had 5.20 % CT, E. crispa had 6.44 % CT and mixed diet had 7.20 % CT. The mean dry matter intake varied significantly among dietary groups (P < 0.001). Similarly, goats offered a mixed diet consumed more CTs (P < 0.01) than those offered individual forage species. The high mean cumulative percentage seed recovery of V. tortilis may be attributed to the higher crude protein of V. tortilis (18.80 %) than D. cinerea (12.20 %). Higher passage rate may encourage animal seed dispersal and germination of plant species. The results from experiment two support the postulation that animals foraging in mixed diet systems consume more PSMs and achieve higher dry matter intake than animals confined to monocultures or single species feeding systems. Given that woody plant encroachment is already reducing farm-grazing capacities in African savannas and this problem is predicted to double by 2050, strategies that improve herbivore ability to consume woody plants will increase forage availability and inform bush control programmes and policies. Moreover, the concomitant increase in CTs by goats exposed to diets with diverse species also has positive implications for animal / Agriculture, Animal Health and Human Ecology / M. Sc. (Agriculture)

Made available in DSpace on 2016-06-02T19:31:49Z (GMT). No. of bitstreams: 1 2941.pdf: 567635 bytes, checksum: 18b3f5d53030b198ba538da511d6b2e3 (MD5) Previous issue date: 2010-03-02 / Universidade Federal de Minas Gerais / Together with fire and climate changes, soil is considered a major determinant in the Braziliancerrado, the richest savanna in the world. Soil can influence plants by filtering species capableof acquiring resources and compete for them, but can also influence plant patterns ofallocation in defense against herbivory. Although many studies focused on plant soilrelationship in cerrado, few focused on the influence of soil at fine scale. We expectedcommunity descriptors, such as floristic composition, richness, evenness, diversity, and totalabundance to be related to soil features at fine scale within a physiognomy. We also expectedplants on nutrient-poor soils to present higher anti-herbivory defenses. In a cerrado site, weplaced 100 contiguous 25 m2 plots, in which we identified all woody individuals, measuredsoil variables and the following leaf traits: specific leaf area, C:N ratio, water content,toughness, trichomes, latex, and presence of tannins, alkaloids, and terpenoids. We did apartial redundancy analysis to test for relationship between soil features and floristiccomposition, controlled for spatial dependence. We also did multiple regression or spatialautoregressive models to test for relationships between soil features and: (1) the abundance ofthe five commonest species, (2) total abundance, (3) richness, (4) evenness, and (5) diversityand to predict defense traits based on soil features. We found no relationship between soil andfloristic composition, probably due to functional redundancy or limited dispersal. Organicmatter was positively related to Myrsine umbellata, the most abundant species, and totalabundance, and negatively related to evenness, what suggests positive feedbacks to cause thedominance by Myrsine umbellata. We also found a positive relationship between sum of basisand species richness, probably reflecting a fertility gradient. Contrary to our expectations, wefound no relationship between total defenses and total soil fertility or soil variables, whatcould result from low variability in soil fertility at fine scale or of high phenotypic variability.Presence of tannins was positively related to organic matter, possibly reflecting a strategytowards lower tolerance due to low reserve allocation or interactions with other resources.However, since tannins decrease leaf decomposition rates, organic matter could beaccumulating in soil. Overall, we suggest that soil is an important factor structuring cerradocommunity even at fine scales and that the dominance of cerrado species could be related topositive plant-soil feedbacks. / O solo, juntamente com o fogo e as variações climáticas, é considerado um dos principaisdeterminantes do cerrado brasileiro, a savana mais rica do mundo. O solo pode influenciar asespécies de planta selecionando aquelas adaptadas a explorar e competir por recursos, mastambém pode influenciar os padrões de alocações em defesas contra herbivoria. Emboramuitos estudos tenham se voltado a entender as relações entre solo e vegetação no cerrado,poucos se focaram efeitos em escala local. De forma geral, nossa expectativa é de quedescritores da comunidade, como composição florística, riqueza, equabilidade, diversidade eabundância total estejam relacionados com o solo, mesmo em escala local, dentro de umadeterminada fisionomia, e que plantas em solos pobres em nutrientes invistam mais emdefesas contra herbivoria devido ao alto custo em repor as folhas perdidas. Em uma área decerrado, lançamos 100 parcelas contíguas de 25 m2 cada, identificamos todos os indivíduosem nível de espécies, coletamos amostras de solo e medimos os seguintes traços foliares dedefesa contra herbivoria: área foliar específica, razão C:N, quantidade de água, dureza,densidade de tricomas, quantidade de látex, e presença de alcaloides, terpenoides e taninos.Para testar a relação entre a composição florística e as variáveis do solo, usamos uma análisede redundância parcial, controlando a autocorrelação espacial. Para testar a relação entre asvariáveis do solo e (1) a abundância das cinco espécies mais abundantes, (2) a abundânciatotal, (3) a riqueza de espécies, (4) a equabilidade e (5) a diversidade de espécies; para prevera distribuição dos traços de defesa contra herbivoria por meio das variáveis do solo,utilizamos regressões múltiplas ou modelos autorregressivos, na presença de autocorrelaçãoespacial. Encontramos uma baixa relação entre o solo e a composição florística,provavelmente devido à presença de espécies funcionalmente redundantes e espécies comdispersão limitada. O conteúdo de matéria orgânica esteve positivamente relacionado àabundância de Myrsine umbellata, a espécie mais abundante na área, e à abundância total, enegativamente relacionada à equabilidade, o que sugere que um mecanismo deretroalimentação positiva pode ser a causa da dominância de Myrsine umbellata. Tambémencontramos uma relação positiva entre soma de bases e a riqueza de espécies provavelmenterefletindo um gradiente de fertilidade. Contrariamente às nossas expectativas, nãoencontramos relação entre o investimento total em defesas e a fertilidade do solo,provavelmente refletindo uma baixa variação nas variáveis do solo em escala local ouvariações fenotípicas entre indivíduos da mesma espécie ou ambas. Entretanto, a presença detaninos esteve relacionada positivamente com o conteúdo de matéria orgânica, o que poderefletir menor tolerância à herbivoria em solos mais pobres ou uma alta acumulação dematéria orgânica no solo devido à lenta taxa de decomposição de folhas com tanino. De formageral, sugerimos que o solo é um importante fator estruturando a comunidade, mesmo emescala local, e que a dominância de espécies de cerrado pode estar relacionada a mecanismosde retroalimentação positiva.

Fundação de Amparo a Pesquisa do Estado de Minas Gerais / Ants that collect nectar from plants bearing extrafloral nectaries (EFN) can defend them against the attacks of herbivores. Many cases of this food-for-protection mutualism have been described and the next step is to understand the mechanisms that structure these interactions. In this study, this issue was addressed through field observations and experiments on ant behaviour in studies compiled into five chapters. Specifically, we evaluated the plant\'s investment in attracting ants (nectar quality) and the competitive relationship between ants (resource dominance hierarchies) as structuring mechanisms of ant-plant interactions. The entire study took place in areas of the ecosystem known as Rupestrian Fields, located in Ouro Preto (Brazil), between June 2011 and March 2014. In Chapter 1, we observe that ant-plant interaction network in Rupestrian Fields is nested and generalized as expected for mutualisms, but differs from ant-hemipteran network, its ecological equivalent, despite both share ant species interactions. In Chapter 2, we delve into the study of the effect of resources in interaction frequency, and we showed that the quality more than quantity of nectar determines the pattern of interaction between plants (10 species) and ants (32 species), as well as the benefits of this interaction (conditional outcomes). In the Chapter 3, we focus on the ants to demonstrate, using paired encounters, that ants species are organized in dominance hierarchies based on aggressiveness. This hierarchies affects the numerical dominance based on territoriality and, more importantly, in the dominance of interactions and nectar. Keeping this line of thought, we demonstrated in Chapter 4 that not all ants that collect nectar really protect the plant against herbivores but only the ecological dominant and highly interacted species. Finally, we close the thesis demonstrating in Chapter 5 that the ant species interacted with plants during the day are not the same at night, but this diurnal-nocturnal turnover does not affect the structure of the interactions by only changing the dominant species. This study brings behavioural and empirical evidences on the importance of the identity of the ant species and its ecological dominance, as well the effect of the nectar quality, for the structure of ant-plant interactions and effectiveness of mutualisms. / As formigas que recolhem o néctar das plantas com nectários extraflorais (EFN) podem defendê-las contra ataques de herbívoros. Muitos casos desse tipo de mutualismo foram descritos mas ainda é preciso entender os mecanismos que estruturam essas interações. Neste estudo, esse problema foi abordado através de observações de campo e experimentos sobre o comportamento das formigas compilados em cinco capítulos. Especificamente, foi avaliado o investimento da planta em atrair formigas (qualidade do néctar) e a relação competitiva entre formigas pelo néctar (hierarquias de dominância) como mecanismos estruturadores das interações formiga-planta (partição das interações). Todo o estudo ocorreu em áreas do ecossistema Campo Rupestre, localizadas em Ouro Preto (Brasil), de junho de 2011 a março de 2014. No Capítulo 1, observamos que redes de interação formiga-planta em Campo rupestre é aninhada e generalizada como esperado para mutualismos, mas difere da rede formiga-hemíptero, seu equivalente ecológico, apesar de compartilharem as mesmas espécies de formigas interagentes. No Capítulo 2, nós aprofundamos no efeito dos recursos estruturando interações, e mostramos que a qualidade mais do que a quantidade de néctar determina o padrão de interação entre formigas (32 espécies) e plantas (10 espécies), bem como os benefícios desta interação (benefícios condicionais). No Capítulo 3, nós demonstramos através de encontros interespecíficos pareados que as formigas são organizadas em hierarquias de dominância comportamental baseadas em agressividade. Essa hierarquia afeta a dominância numérica por territorialidade e, mais importante, o domínio das interações e do néctar. Mantendo esta linha de pensamento, demonstramos no Capítulo 4 que nem todas as formigas que coletam néctar realmente protegem a planta contra herbívoros mas somente as espécies dominantes e altamente interagidas. Finalmente, fechamos a tese demonstrando no Capítulo 5 que as espécies de formiga que interagem com as plantas durante o dia não são o mesmo à noite, mas esta rotatividade diurno-noturno não afeta a estrutura das interações devido a troca específica da formiga dominante. Este estudo traz evidências comportamentais e empíricas sobre a importância da identidade de espécies de formiga e sua dominância ecológica, assim como da qualidade do recurso, para a estrutura de rede da formiga-planta e eficácia do mutualismo. / Doutor em Ecologia e Conservação de Recursos Naturais