[Truncated abstract] In order to assess the sustainability of mixed plantings on saline land, this thesis examined the importance of leaf-litter trapping and microsites on recruitment in a salt affected alley-belted (tree/shrub) agricultural system in Western Australia. Located in the low rainfall region (MAR <330 mm) of the wheatbelt, the 60 ha site consists of concentric rows of Eucalyptus sargentii trees with mounded (6 - 11 cm high) 10 -15 m inter-rows of Atriplex spp. Sustainability of this system and fulfilment of productive and ameliorative functions is dependant on successful recruitment (perennials). Although the present study site was conducted on farmland in a Mediterranean-type climate, low annual rainfall and spatial arrangement of perennial shrubs and trees, allow useful comparisons to be made with naturally occurring banded semi-arid systems and vice-versa. Of key interest were leaf-litter redistribution and trapping by tree and shrub rows and whether litter-cover/microsites facilitated/interfered with seedling recruitment (establishment, growth and survival). Litter from the tree row, redistributed by prevailing winds and rain, accumulated adjacent to saltbush seeding mounds, creating a mosaic of bare and littered areas across the site (total litter 10 t/ha over 22 months). Accumulated litter was hypothesized to differentially influence seasonal soil abiotic parameters (depending on litter-cover density) including; salinity, water availability, infiltration rates, water repellency and temperature. These abiotic conditions were also hypothesized to vary between tree and shrub microsites. Biotically, recruitment at this site was also hypothesized to be determined by interactions (positive and negative) between perennial components and understorey annuals/perennial seedlings. Accumulation of litter and resultant heterogeneity was influenced by shrub morphology, microtopography, wind direction and distance from litter source, with increased litter on the leeward sides of hemispherical Atriplex undulata shrubs and shrubs closest to tree rows. ... The importance of tree/shrub microsites varied seasonally, with no influence in winter due to moderate temperatures and increased water availability. In warmer months saltbush mid-row microsites were most favourable for seedling recruitment due to moderate litter-cover; reducing salinity, temperatures and increasing infiltration; and reduced root-competition/shading by the tree row. Tree microsites also directly inhibited seedling recruitment through increased salinities and water repellency. However, trees also indirectly facilitated recruitment in adjacent areas through provision of leaf-litter. As litter-trapping and recruitment patterns at this site mirror those found in semi-arid natural and artificial systems, the results of this study provide useful insights into creating appropriate mimics of low rainfall natural banded woodland and chenopod shrublands. Saltbush seeding mounds, shrub morphology and litter were key components for litter trapping and recruitment heterogeneity at this site. In this tree/shrub alley planting, where litter quantities directly influence vegetation cover densities, future saline plantings need to consider appropriate tree/shrub row spacings and orientation for efficient resource (seeds, litter and water) capture.

The overwhelming majority of plant species introduced into a new range never become invasive. Consequently, identification of factors allowing the small fraction of successful invaders to naturalize, increase in abundance, and displace resident species continues to be a key area of research in invasion biology. Of the considerable number of hypotheses that have been proposed to resolve why some plant species become noxious pests, the enemy release hypothesis (ERH) is one of the most commonly cited. The ERH maintains that invasive plants succeed in a new range because they are no longer regulated by their coevolved natural enemies, and this reduction in enemy pressure imparts a competitive advantage over native species, which continue to be negatively impacted by top-down processes. Alternatively, the ability of invasive plant species to outperform their counterparts, rather than escape from enemies, may be key in conferring invasion success. The importance of preadapted traits and release from natural enemies in successful invasion remains unclear, likely owing to a lack of empirical studies comparing their effects on relative performance and population growth of closely related species that differ in origin and invasiveness. A system of co-occurring native, introduced non-invasive, and invasive Eugenia congeners exists in south Florida, providing an opportunity to address deficiencies in our understanding of plant invasions by investigating the factors leading to invasion success for Eugenia uniflora. This approach is novel because very few studies have simultaneously incorporated both native and introduced non-invasive congeners into tests of these hypotheses, and no others have done so using this system of Eugenia congeners. The first study in this dissertation tested the ERH using an insect herbivore exclusion experiment in the field to compare the effects of natural enemies on the performance and population growth of Eugenia uniflora and its native congeners. The results showed that E. uniflora sustained more herbivore damage than its native counterparts, and that the effects of herbivores were sufficient to have negative impacts on performance and population growth. In sum, these findings contradict the ERH. Surprisingly, the vast majority of damage to E. uniflora was caused by the recently introduced Sri Lankan weevil (Myllocerus undatus), with which it shares no coevolutionary history. The second study compared seedling performance among native, introduced non-invasive, and invasive Eugenia congeners to determine if the success of E. uniflora can be attributed to superior performance traits. Invasive E. uniflora was found to outperform its native and introduced non-invasive counterparts in a number of seedling traits, including emergence, growth, and survival, in spite of sustaining higher levels of herbivore damage in the field. This result was consistent across years and sites, suggesting that superior performance may be an important factor in invasion success by E. uniflora. The final experiment investigated the role of enemy release on performance of native, introduced non-invasive, and introduced invasive Eugenia seedlings using an insect herbivore exclusion experiment in the field. In this study, the invasive E. uniflora was again found to sustain more damage by foliar herbivores compared to its native and introduced non-invasive counterparts. However, in spite of higher levels of herbivore damage, E. uniflora continued to outperform its congeners in terms of stem growth, and its congeners did not outperform E. uniflora in any attribute. Insect herbivores negatively affected survival of all species, but were found to have little effect on growth. In combination, the results of these studies indicate that the ability of E. uniflora to outperform its native and introduced congeners at the seedling stage, and not release from insect herbivores, may contribute to its success as an invader. Additionally, E. uniflora exhibits relatively low resistance to herbivory in the new range, and instead may possess an ability to tolerate moderate levels of damage. The implications of this study are that enemy release may not be important in determining invasion success in some systems, and that the accumulation of new enemies may mitigate the effects of invasive plants over time. The paucity of studies investigating interactions among invasive plants and herbivores that share no coevolutionary history warrants further research. Finally, this system of Eugenia congeners provides valuable opportunities to test additional hypotheses and to further explore factors leading to invasion success.

Effect of soil moisture potential (?), temperature (T), genotype, seeding depth (SD) and rate (SR) on seedling emergence (SE), emergence velocity (EV), root yield and grades of cut and peel carrots were studied. SE was reduced at –120 kPa and totally inhibited at -156 kPa. EV was the lowest at – 5 kPa and – 90 kPa. SE was delayed by 33 d at 5°C, reduced at 30°C and totally inhibited at 35 and 40 °C. Heat units 99.75 and 159.60°Cd were the lowest to initiate and complete SE respectively while the optimum was 300 – 350 °Cd. There was no interaction effect between ? and T on SE. Honey snax at 85 seeds/ 30 cm showed the best SE whereas, Triton recorded the highest total yield at 2.54 cm SD and Fancy yield at 85 seeds/ 30 cm implying certain crop ecological and management factors can influence SE, root yield and quality.

Modeling the seedling recruitment microsite involves characterization of the soil environment of the shallow profile from which weed seedlings recruit. Understanding the environment of the seedling recruitment microsite is the prelude to weed emergence studies. Because of spatial and temporal heterogeneity of the recruitment microsite, sufficient measurements are often not feasible. An experiment was established in 2003 and 2004 across topography within an annually cropped field in south-central Manitoba to determine the effect that hillslope aspect and position, and soil residue and depth would have on microsite environment within the shallow seedling recruitment zone. Microclimatic, topographic, soil surface and soil properties were assessed in the context of the weed recruitment microsite. The soil water retention characteristic was measured by pressure plate to determine water availability to germinating seeds at the various topographic positions. The soil water characteristic was evaluated across topography and soil depth. Evaluation of the soil water characteristic by pedotransfer function indicated that a single soil water characteristic is representative of the recruitment zone. Field and laboratory experimental measurements were used as parameterization for the simultaneous heat and water (SHAW) model to generate continuous water and temperature profiles for the recruitment zone. Soil temperature and temperature fluctuation decreased with depth in the recruitment zone. Despite differences of texture, bulk density, and organic matter across topography and soil depth, the soil water characteristic differed only across topography. Soil water potential fluctuated considerably at the soil surface due to numerous precipitation events and direct evaporation. Implications for germinating seeds is that the seedling recruitment zone is influenced by spatial effects of topography and the vertical location of the seed microsite. Physical process based modeling used in this study to predict temperature and water within the seedling recruitment zone enables better understanding of interactions between above-ground microclimate and the recruitment microsite. Such interactions enable linkage between atmospheric models and recruitment models that can enhance our ability to evaluate crop management decisions.

The New Zealand alpine zone has many fleshy-fruited plant species, but now has a relatively depauperate animal fauna. The key question is, therefore, are native alpine plants still being dispersed, if so where to and by what? I first measured fruit removal rates among nine common species using animal-exclusion cages to compare natural fruit removal by all animals, and by lizards only. Over two years, mean percent of fruit removed by early winter ranged from 25–60% among species. Speed of fruit removal also varied depending on species. Secondly, I quantified which animals disperse (or predate) seeds of those fruits, into which habitats they deposit the seeds, and the relative importance of each animal species for dispersal, in two ways. A 2-year study using fixed-area transects to monitor faecal deposition showed that introduced mammals (especially possums, rabbits, hares, sheep, pigs and hedgehogs) were abundant and widespread through alpine habitat. Of the 25,537 faeces collected, a sub-sample of 2,338 was dissected. Most mammals dispersed most (> 90%) seeds intact. However, possums (numerically the important disperser) moved most seeds into mountain beech (Nothofagus solandri) forest, while rabbits, hares, and sheep dispersed seeds mainly into open grassland dominated by thick swards of exotic grasses (e.g. Agrostis capillaris and Anthoxanthum odoratum); all are less suitable microsites. Kea (Nestor notabilis), the largest and most mobile of only three remaining native alpine bird species, are potentially useful as a long-distance seed disperser, even though parrots are typically seed predators. I found that kea are numerically more important than all other birds combined, damage very few seeds, and are probably responsible for most dispersal of seeds between mountain ranges. Finally, I investigated the effects of seed deposition microsite (shady/high-light), pulp-removal (whole/cleaned), competition (soil dug/not-dug) and predation (caged/ not) on germination, growth and survival of eight subalpine plant species. There were strong positive effects of shady microsites for seed germination and seedling survival to 3.5 years for six of the eight species. Effects of other treatments were less important and varied among species and stages. Hence, both native birds and introduced mammals are dispersing alpine seeds, but the mammals often deposit seeds in habitats unsuitable for establishment. Any evaluation of the dispersal effectiveness of frugivores must consider their contribution towards the long-term success for plant recruitment through dispersal quantity and quality.

The aim of this thesis was to study patterns of reproduction of Betula pendula and B. pubescens coll. along an altitudinal, coastal-inland, gradient in northern Sweden. The altitudinal variation was related to the distribution of the birch taxa along the gradient. Six years field data showed a steep decrease of seed germinability of B. pendula towards its altitudinal range limit in the Scandes every year. In contrast, B. pubescens ssp. pubescens showed significant positive correlations between seed germinability and altitude in three of the six years. Furthermore, there was a highly significant positive correlation between seed weight and altitude for B. pubescens coll., but not for B. pendula. Production of viable seeds fluctuated strongly between years in most populations, except in marginal B. pendula populations in the mountain area where it was constantly very low. On average 15 - 41 % of the seeds produced in B. pendula populations above 400 m altitude were attacked by gall midges (Semudobia ssp.). Corresponding values for B. pendula populations below 400 m altitude were 4 - 7 %. In B. pubescens populations, the seeds attacked by Semudobia ssp. never exceeded 3 %. The high frequency of Semudobia attackes in high altitude marginal B. pendula populations was suggested to be due to limited resources for defense against the seed predator. A 3-yr study documented large variations in pollination and seed quality between taxa, high and low altitude populations, and between years. Empty seeds (without embryos) dominated among the sampled seeds in most cases. This proportion was decreased by pollen addition, in both B. pendula and B. pubescens, in mountain populations, but not in coastal populations. The high percentage of empty seeds was therefore suggested to be partly caused by pollen-limitation, but failure of pollen tube penetration - fertilisation, or maternal resource supply could also have had an influence. A laboratory experiment showed increased pollen germination and length of the longest pollen tube per style after increased pollination. The correlations between number and length of pollen tubes per style were however mostly low in natural populations, suggesting low probability of pollen competition in the natural situation. Sowing experiments indicated that differences in initial seedling density between B. pendula and B. pubescens was more affected by interspecific differences in seed quality than by interspecific differences in survival of seedlings after establishment A study of the age structure of a B. pendula stand, planted about 250 m above its natural altitudinal limit, indicated that this birch may reproduce above its recent range limit during temporally warmer periods. It was concluded that the level of accumulated resources in B. pendula in marginal sites in the Scandes would mostly be too low for completion of the reproductive cycle. In contrast, B. pubescens ssp. pubescens is able to accumulate enough rescources for reproduction also at relative high altitudes. / <p>Diss. (sammanfattning) Umeå : Umeå universitet, 1994, härtill 5 uppsatser.</p> / digitalisering@umu

Modeling the seedling recruitment microsite involves characterization of the soil environment of the shallow profile from which weed seedlings recruit. Understanding the environment of the seedling recruitment microsite is the prelude to weed emergence studies. Because of spatial and temporal heterogeneity of the recruitment microsite, sufficient measurements are often not feasible. An experiment was established in 2003 and 2004 across topography within an annually cropped field in south-central Manitoba to determine the effect that hillslope aspect and position, and soil residue and depth would have on microsite environment within the shallow seedling recruitment zone. Microclimatic, topographic, soil surface and soil properties were assessed in the context of the weed recruitment microsite. The soil water retention characteristic was measured by pressure plate to determine water availability to germinating seeds at the various topographic positions. The soil water characteristic was evaluated across topography and soil depth. Evaluation of the soil water characteristic by pedotransfer function indicated that a single soil water characteristic is representative of the recruitment zone. Field and laboratory experimental measurements were used as parameterization for the simultaneous heat and water (SHAW) model to generate continuous water and temperature profiles for the recruitment zone. Soil temperature and temperature fluctuation decreased with depth in the recruitment zone. Despite differences of texture, bulk density, and organic matter across topography and soil depth, the soil water characteristic differed only across topography. Soil water potential fluctuated considerably at the soil surface due to numerous precipitation events and direct evaporation. Implications for germinating seeds is that the seedling recruitment zone is influenced by spatial effects of topography and the vertical location of the seed microsite. Physical process based modeling used in this study to predict temperature and water within the seedling recruitment zone enables better understanding of interactions between above-ground microclimate and the recruitment microsite. Such interactions enable linkage between atmospheric models and recruitment models that can enhance our ability to evaluate crop management decisions.

Submitted by Alexandre Soares (alexandredesoares@yahoo.com.br) on 2016-07-15T14:46:30Z No. of bitstreams: 1 karin_frichis_do_nascimento.pdf: 1481669 bytes, checksum: c542b6f2d89d5f67694bf36d33eecd30 (MD5) / Rejected by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br), reason: on 2016-07-15T20:04:35Z (GMT) / Submitted by Alexandre Soares (alexandredesoares@yahoo.com.br) on 2016-07-15T20:19:47Z No. of bitstreams: 1 karin_frichis_do_nascimento.pdf: 1481669 bytes, checksum: c542b6f2d89d5f67694bf36d33eecd30 (MD5) / Rejected by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br), reason: rever: Refer?ncia ABNT Tipo: opem Acess Nota de disserta??o/Tese: colocar ano Palavra Chave: cada uma em um campo, sem pontua??o Ag?ncia Financiadora: Sempre por extenso e com a abrevia??o entre par?nteses no final on 2016-07-18T14:41:49Z (GMT) / Submitted by Alexandre Soares (alexandredesoares@yahoo.com.br) on 2016-07-19T12:58:18Z No. of bitstreams: 1 karin_frichis_do_nascimento.pdf: 1481669 bytes, checksum: c542b6f2d89d5f67694bf36d33eecd30 (MD5) / Rejected by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br), reason: Rever: Refer?ncia ABNT Tipo Nota de disserta??o/Tese Resumos Palavra Chave Ag?ncia Financiadora Seguir modelo: http://acervo.ufvjm.edu.br/jspui/handle/1/360 on 2016-07-21T15:22:35Z (GMT) / Submitted by Alexandre Soares (alexandredesoares@yahoo.com.br) on 2016-08-25T11:45:56Z No. of bitstreams: 1 karin_frichis_do_nascimento.pdf: 1481669 bytes, checksum: c542b6f2d89d5f67694bf36d33eecd30 (MD5) / Approved for entry into archive by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br) on 2016-08-25T12:22:54Z (GMT) No. of bitstreams: 1 karin_frichis_do_nascimento.pdf: 1481669 bytes, checksum: c542b6f2d89d5f67694bf36d33eecd30 (MD5) / Made available in DSpace on 2016-08-25T12:22:54Z (GMT). No. of bitstreams: 1 karin_frichis_do_nascimento.pdf: 1481669 bytes, checksum: c542b6f2d89d5f67694bf36d33eecd30 (MD5) Previous issue date: 2015 / O trabalho teve como objetivo desenvolver procedimentos de micropropaga??o para a esp?cie pau-terra-liso (Qualea multiflora Mart.) a partir de sementes germinadas in vitro e avaliar a emerg?ncia, o crescimento inicial e a sobreviv?ncia de mudas em fun??o de diferentes substratos e ambientes, em condi??es de viveiro. No primeiro cap?tulo, as sementes de Qualea multiflora foram submetidas ? desinfesta??o com hipoclorito de s?dio em diferentes concentra??es e tempos de imers?o para a sua introdu??o in vitro. Foi avaliado o percentual de germina??o e contamina??o. Utilizando o melhor tratamento do experimento de desinfesta??o, foi instalado outro experimento para comparar as composi??es distintas de meio de cultura MS (Murashige & Skoog, 1962) e WPM (Lloyd & McCown, 1981) na germina??o de pau-terra-liso. Avaliou-se o percentual de germina??o e de pl?ntulas normais. Na fase de multiplica??o foram utilizados dois tipos de explantes (segmento nodal e segmento cotiledonar) retirados das pl?ntulas germinadas in vitro que foram inoculados em meio de cultura WPM. Este foi suplementado com BAP em concentra??es diferenciadas e ANA. A fase foi constitu?da pelo cultivo inicial e dois subcultivos. Avaliaram-se os n?meros de brota??es por explantes e a altura da maior brota??o. Constatou-se que a concentra??o de 5% de hipoclorito de s?dio durante 20 minutos de imers?o proporcionou os melhores resultados de desinfesta??o e germina??o in vitro. Observou-se que o tipo de meio de cultura e a concentra??o influenciam na germina??o e na qualidade das pl?ntulas de Qualea multiflora, logo, recomenda-se o meio WPM com 100% de sais e vitaminas para essa esp?cie. Os melhores resultados de multiplica??o foram alcan?ados utilizando o explante cotiledonar e a concentra??o de 0,6 mg L-? de BAP. No cap?tulo 2, os experimentos foram instalados em ambiente de casa de sombra e de casa de vegeta??o utilizando quatro tipos de substratos, sendo: 1) 100% substrato comercial Bioplant?, 2) 70% de vermiculita de granulometria m?dia + 30% de fibra de coco, 3) 70% de vermiculita + 30% Bioplant?, e 4) 40% de vermiculita + 30% de fibra de coco + 30% de Bioplant?. Realizaram-se avalia??es aos 60, 90, 120 e 150 dias para verificar a emerg?ncia, o crescimento em altura e di?metro e a sobreviv?ncia das mudas. No final do experimento, foram obtidos o peso da mat?ria seca da xii parte a?rea, o peso de mat?ria seca de ra?zes, o peso de mat?ria seca total e a rela??o peso de mat?ria seca da parte a?rea e peso de mat?ria seca das ra?zes. Em casa de vegeta??o a emerg?ncia de Qualea multiflora obteve os maiores percentuais com o uso do substrato VB (70% de vermiculita + 30% de Bioplant ?). N?o ocorreu diferen?a no crescimento em altura entre as mudas que estavam em casa de vegeta??o e em casa de sombra. Para o ambiente casa de sombra, n?o houve diferen?as significativas entre as caracter?sticas de mat?ria seca analisadas, em fun??o dos substratos. Com os dados de sobreviv?ncia nos dois ambientes, conclui-se que a Qualea multiflora ? de dif?cil propaga??o em condi??es de viveiro, sendo necess?rios mais estudos para a produ??o de mudas da esp?cie. / Disserta??o (Mestrado) ? Programa de P?s-Gradua??o em Ci?ncia Florestal, Universidade Federal dos Vales do Jequitinhonha e Mucuri, 2015. / The study aimed to develop micropropagation procedures for the ?pau-terra-liso? species (Qualea multiflora Mart.) from seeds germinated in vitro and evaluate the emergency, the initial growth and survival of seedlings for different substrates and environments in arboretum conditions. In the first chapter, the Qualea multiflora seeds were subjected to disinfection with sodium hypochlorite in different concentrations and immersion times for its introduction in vitro. The percentage of germination and contamination was evaluated. Using the best treatment of disinfestation experiment, it was installed another experiment to compare the different compositions of MS (Murashige and Skoog, 1962) and WPM (Lloyd & McCown, 1981) on the ?pau-terra-liso? germination. It was evaluated the percentage of germination and normal seedlings. In the multiplication phase it was used two types of explants (nodal segments and cotyledon segment) taken from seedlings germinated in vitro which were inoculated in WPM culture. This one was supplemented with ?BAP? and ?ANA? in different concentrations. The stage was set for the initial culture and two subcultures. It was evaluated the shoot numbers per explant and the height of the larger shoot. It was found that the concentration of 5% sodium hypochlorite for 20 minutes immersion gave the best results of disinfestation and in vitro germination. It was observed that the type of culture environment and the concentration influence the germination and quality of seedlings of Qualea multiflora, so it is recommended a WPM environment with 100% salts and vitamins for this species. The best multiplication results were achieved using cotyledon explants and concentration of 0.6 mg L-? of BAP. In Chapter 2, the experiments were conducted in shade house and greenhouse environment using four types of substrates, as follows: 1) 100% commercial substrate Bioplant?, 2) 70% of average grain size of vermiculite + 30% coconut fiber, 3) 70% of vermiculite + 30% Bioplant?, and 4) 40% of vermiculite and 30% coconut fiber + 30% Bioplant?. Evaluations were performed at 60, 90, 120 and 150 days to verify the emergence, growth in height and diameter and survival of seedlings. At the end of the experiment, it was obtained the dry matter weight of the aerial part, the dry matter weight of the roots, the entire dry matter weight and the relation between the weight of dry matter of aerial part and weight xiv of dry matter of the roots. In a greenhouse, the emergency Qualea multiflora obtained the highest percentages with the use of VB substrate (70% vermiculite and 30% of Bioplant ?). There was no difference in height growth among the seedlings that were in the greenhouse and in the shade house. To the environment shade house, there were no significant differences among the characteristics of the analyzed dry matter, according to the substrates. With the survival data in both environments, it is concluded that the Qualea multiflora is difficult to spread in arboretum conditions; further research is needed for the production of seedlings of the species.

A região dos morros graníticos de Porto Alegre (Estado do Rio Grande do Sul), inserida no sul do Brasil, apresenta uma cobertura vegetal em forma de mosaico natural de florestas e áreas abertas (savanas ou campos). Neste locais, segundo o clima atual, as formações florestais tendem a avançar sobre a matriz herbácea, principalmente de forma agrupada, formando ilhas de nucleação florestal. Este padrão é mantido por forças seletivas como a precipitação, o fogo e a herbivoria (vertebrados e invertebrados), que incidem de maneira mais intensa sobre as fases iniciais das populações de plantas pioneiras da floresta. As forças atuam de maneira distinta nos diversos hábitats disponíveis para o estabelecimento de plântulas, criando um arranjo espacial em forma de ilhas. Esta tese aborda, no primeiro capítulo, os padrões florísticos e espaciais de ilhas de nucleação florestal em ecótono natural de floresta-savana, segundo as diferentes exposições solares (norte, topo e sul) dominantes do morro Santana, comparando-as, floristicamente, com as bordas de mata. No segundo capítulo, é avaliada, experimentalmente, a sobrevivência de plântulas de guamirim (Myrcia palustris DC.; Myrtaceae) sob a influência de uma seca severa em diferentes hábitats (borda de mata, ilhas de nucleação florestal e matriz herbácea) presentes em três exposições solares (norte, topo e sul). Em ambos os casos, os estudos foram conduzidos no morro Santana (30°03’ S, 51°07’ W), cuja altitude máxima é de 311m acima do nível do mar. Para avaliar os padrões florísticos e espaciais, foram selecionadas vinte oito ilhas de nucleação florestal das várias área de cobertura (entre 5,5 a 904m2) em um mosaico natural de floresta-campo e quatro parcelas na borda da mata de dois tamanhos distintos (121,5 e 243m2). As espécies vegetais foram identificadas, classificadas em síndromes de dispersão e a abundância de indivíduos foi registrada. Os padrões florísticos foram observados através de análise multivariada. As relações entre a área de cobertura e diversidade/riqueza e entre o número de indivíduos e a riqueza foram analisadas por regressão linear. Para avaliar a sobrevivência de M. palustris, foram analisadas as taxas de sobrevivência e de mortalidade, bem como as causas desta mortalidade para esta espécie arbórea florestal generalista. As sobrevivências e causas de morte foram avaliadas mensalmente, durante os primeiros meses de vida desta espécie, em um período de seca severa em três hábitats e dois controles: borda de mata (B), ilhas de nucleação florestal (I) e matriz herbácea (H), além dos controle-interno (CI) e controleexterno (CE) repetidas em três ambientes geomorfológicos (norte, topo e sul). Em cada hábitat foram transplantadas plântulas isoladas e agrupadas (2-4), com o objetivo de avaliar a mortalidade densidade-dependente. Foram registrados 4214 indivíduos (2828 nas ilhas de nucleação e 1386 nas bordas de mata) distribuídos em 38 famílias e 111 espécies. A análise de agrupamento formou três grupos: g1, com ilhas de nucleação florestal mais iniciais dos três ambientes geomorfológicos (predomínio das ilhas do topo); g2, também por ilhas dos três ambientes geomorfológicos (predomínio das ilhas do norte); e g3, com as ilhas mais desenvolvidas dos três ambientes geomorfológicos (predomínio das ilhas do sul) unidas às bordas de mata. A dispersão zoocórica é dominante nas ilhas dos três grupos, ao passo que a dispersão autocórica apresentou um tendência de aumento de g1 para g3 e a dispersão anemocórica, o inverso. Nas mais desenvolvidas, houve maior riqueza e abundância de espécies lenhosas florestais tardias do que em ilhas mais iniciais. Os testes de aleatorização entre os três grupos revelaram que as áreas de cobertura de g3 (344,5m2) e g2 (156m2) são significativamente maiores que g1 (40,9m2) (p = 0,001), o que não ocorreu entre g3 e g2 (p = 0,063). Os valores de diversidade foram diferentes (p = 0,001) entre g1 (1,56) e g2 (2,16) ou g3 (2,97) e, levemente diferentes (p = 0,04) entre g2 e g3. Retirando as bordas de mata, as ilhas das três diferentes exposições solares diferiram (p = 0,02) em relação à riqueza e a abundância do norte e do topo, do norte e do sul, mas não entre o sul e o topo (p = 0,06). A análise de congruência obteve valor máximo (0,72) unindo riqueza e abundância; se acrescidos a eles a área de cobertura, a diversidade e a densidade das ilhas, este valor ainda permanece alto (0,7). A área de cobertura foi o único parâmetro especial que se mostrou correlacionado com riqueza (R2 = 0,676; p < 0,001) e com a diversidade (R2 = 0,49; p < 0,001). Quanto maior o número de indivíduos, maior a riqueza em g1 (R2 g1 = 0,63; p = 0,002), em g2 (R2 g2 = 0,66; p = 0,004), mas não em e g3 (R2 g3 = 0,6; p = 0,07). A falta de significância na correlação em g3 é resultado do baixo número de unidades amostrais neste grupo (6). com a Os resultados com M. palustris revelam que as taxas de sobrevivências médias mensais aumentaram inversamente com a quantidade de precipitação mensal para todos os ambientes (R2 = 0,46; p = 0,006). A curva de sobrevivência decaiu mais abruptamente no topo do que nos demais ambientes geomorfológicos (p < 0,01), sem diferenças entre norte e sul (p = 0,263). Esta mesma curva foi diferente entre todos os hábitats (p < 0,01), resultando numa maior chance de sobrevivência em B, seguida por I e nula em H. As forças seletivas que causam a morte das plântulas são distintas entre os hábitats (p < 0,01): a herbivoria por invertebrado é maior em H; por vertebrado, é maior em I e H; a seca/doença é maior em B; e o fogo, maior em I. A partir destes resultados, são apresentadas as seguintes conclusões: Os padrões florísticos e espaciais do avanço florestal em ilhas de nucleação refletem os processos de sucessão de um estádio mais inicial e simples para um estádio mais “tardio” e complexo, similar a uma borda de mata. Este processo é limitado pela distribuição de micro-hábitats adequados ao estabelecimento das espécies lenhosas florestais de dispersão zoocórica. A sucessão nas ilhas é retroalimentada positivamente, sendo que a chegada e o estabelecimento de novos indivíduos de espécies florestais atrai mais dispersores de sementes, acelerando a sucessão. A composição florística é dependente, também, da localização da ilha em relação às diferentes exposições solares. Para a sobrevivência de plântulas de M. palutris, a precipitação mensal é o parâmetro determinante primário. A distribuição dos hábitats afeta a dinâmica do avanço florestal na medida em que varia, entre eles, a taxa de sobrevivência de plântulas. Mesmo em anos de seca severa, ocorre o avanço da floresta em hábitats mais sombreados presentes no mosaico. Por fim, pode-se concluir que o avanço da floresta sobre a matriz herbácea é um fenômeno vinculado à natureza florística das matas adjacentes. Na matriz da paisagem, há diversos hábitats potencialmente disponíveis que modificam as chances de sobrevivência de plântulas e, portanto, contribuem, de formas diferentes, para a dinâmica da vegetação de comunidades de populações de espécies lenhosas. Entre os hábitats, a matriz herbácea é o hábitat mais inóspito para plântulas destas espécies e a borda da mata, o mais adequado. As ilhas apresentam uma condição intermediária entre os dois primeiros hábitats, sendo as mais desenvolvidas bastante semelhantes às bordas. Os mecanismos que retardam esta dinâmica são, em ordem de importância, a disponibilidade de água (em todos os hábitats), o herbivoria por vertebrado (nas ilhas e matriz herbácea), o fogo (na matriz herbácea) e, por último, a herbivoria por invertebrado (na matriz herbácea). / The region of granitic hills of Porto Alegre (State of Rio Grande do Sul) in southern brazilian’s region is contituted by a mosiac of forest and open areas (savannas or grassland). In these sites, according to the actual climate, the conditions are suitable for forest, which should mean that this vegetation type is in expansion, invading the surrounding herbaceous matrix. This expansion shows, mainly, an aggregation pattern by woody thickets of forest nucleation. The maintenance of the pattern is made by selective forces like precipitation, fire and herbivory (vertebrade and invertebrade) which are stronger in a juvenil phases of tree popualations. The selective forces act differently within all habitats available for seedling stablishment, creating a spatial arrangment of woody island. On the first chapter of this study, we analysed the floristic and spatial patterns of thickets island of forest expansion in a natural ecotone of forest-savanna, according to the slope (north, top and south) on Santan hill. It compares the floristic ans spatial patterns of thickets and forest edges in this three slopes. On the second chapter, we evaluated monthly, the seedling suvivorship of guamirim (Myrcia palustris DC.; Myrtaceae) under the influence of a severious drought (2004/2005) within differents habitats (forest edge, thickets and herbaceous matrix) of three slopes (north, top and south), during 15 months, by an experimental approach. In both cases, the study was carried out in Santana hill (30°03’ S, 51°07’ W), with maximum altitude 311m above sea level. In order to evaluate the floristic and spatial patterns, we selected 28 thickets of many cover areas (between 5,5 to 904m2) and four forest edges of two sizes (121,5 e 243m2). All woody individuals (h ³ 1m) were registered. The species, their abundance and dispersal modes were registered too. The floristic patterns were analysed by multivariate analyses. The relationship between cover area and diversity/richness or abundance and richness were analysed by linear regression. For survivorship of M. palutris seedlings, we analysed the survival/mortality rates and the mortality causes for this generalist tree. The survival and mortality causes were tested within three habitats and two controls: forest edge (B), thickets (I), herbaceous matrix (H), innercontrol (CI) and external-control (CE). The habitats were placed in three differents slopes (north, top and south). In all traits, we tranplanted isolated and group seedlings (1-4), to investigate the density-dependent mortality. We registered 4214 individuals (2828 in the thickets and 1386 in edge) amongst 38 botanical families and 111 species. The cluster analyses showed three groups: g1 with initial thickets of three slopes (predominance of top thickets), g2 also with thickets of three slopes (predominance of north thickets) and g3 with more development thickets of three slopes (predominance of south thickets) together with four forest edges. The dispersal by animals was dominant in all three groups. The autochorie dispersal showed an increase trend from g1 to g3, while anemochorie dispersal was in opposite way. The randomization tests indicates that g3 cover area (344,4m2) and g2 (156m2) are bigger (p = 0,001) than g1 (40,9m2) and g3 and g2 are no different (p = 0,063). The diversity (H’) was almost the same, with g3 (2,97) or g2 (2,16) higher (p = 0,001) than g1 (1,56), but g3 and g2 are slightly differents (p = 0,04). Without the forest edges samples, the thickets had differents richness and abundance (p = 0,02) from the north to the top or to the south, but no differences (p = 0,06) form the south to the top. The congruence analyses had maximum value (0,72) joining richness and abundance, however with more area cover, diversity and densitiy together, the value keeps high (0,7). Just one single parameter, cover area, could explain the increase of richness (R2 = 0,676; p < 0,001) and diversity (R2 = 0,49; p < 0,001). With increase number of individuals, increase the richness in g1 (R2 g1 = 0,63; p = 0,002), in g2 (R2 g2 = 0,66; p = 0,004), but not in g3 (R2 g3 = 0,6; p = 0,07). The lack of significance in g3 is related to the few sample units of this group (6). The results with M. palustris seedlings showed a inverse relation (R2 = 0,46; p = 0,006) between the mean mensal mortality (mm) and the the mensal precipitation for all habitats in all slopes. The suvivorship curve decreased more abruptly in the top than north or south (p < 0,01), but not form the south to north (p = 0,263). This curve were also different (p < 0,01) amongst all habitats, which makes the chance of survival be higher in B than I and nule in H. The selective forces that cause seedlings death are distinct amongst the habitats (p < 0,01): by invertebrade herbivory higher in H, by vertebrade herbivory higher in I and, drought/disease higher in B and by fire higher in H. Based on this results, we found this conclusions: The floristic and spatial patterns of forest expansion by woody thickets reflect the successional process that ranges from an initial and simple stage (more pioneer) to an latter and complex stage (less pioneer) similar with the forest edge. This process is limited by micro-habitats distribution which are suitable for forest tree seedlings with dispersal by animals. The successional process within the thickets have a positive feedback, where the arrival and stablishment of new individuals from forest species atract more seed dispersors, accelerating the process. The floristic composition depends also from the thicket slope location. For the survivorship of M. palutris seedling, the mensal precipitation was main determinant parameter. The habitat distribution affects the forest expansion dynamics due the differents suvival rates amongst them. Even with a severious drought, there is forest expansion in the mosaic within more shade habitats. We concluded that the forest expansion over the herbaceous matriz is a phenomena dependent from the surrounding forest edges. Within the mosaic landscape, there are several potencial habitats available wich modified the chances of seedling survival and, therefore, had distinct contributions to communities and popualtion dynmics of woody species. Amongst habitats, the herbaceous matriz is the less suitable for seedlings of this specie, while the forest edge is the most one. The thickets has the intermediate conditions, where the most development are very similar to the edges. The mechanisms wich delay this dynamic, on the decreased importance order, is the water availability (all habitats), the vertebrate herbivore (thickets and herbaceous matrix), the fire (herbaceous matrix) and the invertebrade herbivory (herbaceous matrix).

A região dos morros graníticos de Porto Alegre (Estado do Rio Grande do Sul), inserida no sul do Brasil, apresenta uma cobertura vegetal em forma de mosaico natural de florestas e áreas abertas (savanas ou campos). Neste locais, segundo o clima atual, as formações florestais tendem a avançar sobre a matriz herbácea, principalmente de forma agrupada, formando ilhas de nucleação florestal. Este padrão é mantido por forças seletivas como a precipitação, o fogo e a herbivoria (vertebrados e invertebrados), que incidem de maneira mais intensa sobre as fases iniciais das populações de plantas pioneiras da floresta. As forças atuam de maneira distinta nos diversos hábitats disponíveis para o estabelecimento de plântulas, criando um arranjo espacial em forma de ilhas. Esta tese aborda, no primeiro capítulo, os padrões florísticos e espaciais de ilhas de nucleação florestal em ecótono natural de floresta-savana, segundo as diferentes exposições solares (norte, topo e sul) dominantes do morro Santana, comparando-as, floristicamente, com as bordas de mata. No segundo capítulo, é avaliada, experimentalmente, a sobrevivência de plântulas de guamirim (Myrcia palustris DC.; Myrtaceae) sob a influência de uma seca severa em diferentes hábitats (borda de mata, ilhas de nucleação florestal e matriz herbácea) presentes em três exposições solares (norte, topo e sul). Em ambos os casos, os estudos foram conduzidos no morro Santana (30°03’ S, 51°07’ W), cuja altitude máxima é de 311m acima do nível do mar. Para avaliar os padrões florísticos e espaciais, foram selecionadas vinte oito ilhas de nucleação florestal das várias área de cobertura (entre 5,5 a 904m2) em um mosaico natural de floresta-campo e quatro parcelas na borda da mata de dois tamanhos distintos (121,5 e 243m2). As espécies vegetais foram identificadas, classificadas em síndromes de dispersão e a abundância de indivíduos foi registrada. Os padrões florísticos foram observados através de análise multivariada. As relações entre a área de cobertura e diversidade/riqueza e entre o número de indivíduos e a riqueza foram analisadas por regressão linear. Para avaliar a sobrevivência de M. palustris, foram analisadas as taxas de sobrevivência e de mortalidade, bem como as causas desta mortalidade para esta espécie arbórea florestal generalista. As sobrevivências e causas de morte foram avaliadas mensalmente, durante os primeiros meses de vida desta espécie, em um período de seca severa em três hábitats e dois controles: borda de mata (B), ilhas de nucleação florestal (I) e matriz herbácea (H), além dos controle-interno (CI) e controleexterno (CE) repetidas em três ambientes geomorfológicos (norte, topo e sul). Em cada hábitat foram transplantadas plântulas isoladas e agrupadas (2-4), com o objetivo de avaliar a mortalidade densidade-dependente. Foram registrados 4214 indivíduos (2828 nas ilhas de nucleação e 1386 nas bordas de mata) distribuídos em 38 famílias e 111 espécies. A análise de agrupamento formou três grupos: g1, com ilhas de nucleação florestal mais iniciais dos três ambientes geomorfológicos (predomínio das ilhas do topo); g2, também por ilhas dos três ambientes geomorfológicos (predomínio das ilhas do norte); e g3, com as ilhas mais desenvolvidas dos três ambientes geomorfológicos (predomínio das ilhas do sul) unidas às bordas de mata. A dispersão zoocórica é dominante nas ilhas dos três grupos, ao passo que a dispersão autocórica apresentou um tendência de aumento de g1 para g3 e a dispersão anemocórica, o inverso. Nas mais desenvolvidas, houve maior riqueza e abundância de espécies lenhosas florestais tardias do que em ilhas mais iniciais. Os testes de aleatorização entre os três grupos revelaram que as áreas de cobertura de g3 (344,5m2) e g2 (156m2) são significativamente maiores que g1 (40,9m2) (p = 0,001), o que não ocorreu entre g3 e g2 (p = 0,063). Os valores de diversidade foram diferentes (p = 0,001) entre g1 (1,56) e g2 (2,16) ou g3 (2,97) e, levemente diferentes (p = 0,04) entre g2 e g3. Retirando as bordas de mata, as ilhas das três diferentes exposições solares diferiram (p = 0,02) em relação à riqueza e a abundância do norte e do topo, do norte e do sul, mas não entre o sul e o topo (p = 0,06). A análise de congruência obteve valor máximo (0,72) unindo riqueza e abundância; se acrescidos a eles a área de cobertura, a diversidade e a densidade das ilhas, este valor ainda permanece alto (0,7). A área de cobertura foi o único parâmetro especial que se mostrou correlacionado com riqueza (R2 = 0,676; p < 0,001) e com a diversidade (R2 = 0,49; p < 0,001). Quanto maior o número de indivíduos, maior a riqueza em g1 (R2 g1 = 0,63; p = 0,002), em g2 (R2 g2 = 0,66; p = 0,004), mas não em e g3 (R2 g3 = 0,6; p = 0,07). A falta de significância na correlação em g3 é resultado do baixo número de unidades amostrais neste grupo (6). com a Os resultados com M. palustris revelam que as taxas de sobrevivências médias mensais aumentaram inversamente com a quantidade de precipitação mensal para todos os ambientes (R2 = 0,46; p = 0,006). A curva de sobrevivência decaiu mais abruptamente no topo do que nos demais ambientes geomorfológicos (p < 0,01), sem diferenças entre norte e sul (p = 0,263). Esta mesma curva foi diferente entre todos os hábitats (p < 0,01), resultando numa maior chance de sobrevivência em B, seguida por I e nula em H. As forças seletivas que causam a morte das plântulas são distintas entre os hábitats (p < 0,01): a herbivoria por invertebrado é maior em H; por vertebrado, é maior em I e H; a seca/doença é maior em B; e o fogo, maior em I. A partir destes resultados, são apresentadas as seguintes conclusões: Os padrões florísticos e espaciais do avanço florestal em ilhas de nucleação refletem os processos de sucessão de um estádio mais inicial e simples para um estádio mais “tardio” e complexo, similar a uma borda de mata. Este processo é limitado pela distribuição de micro-hábitats adequados ao estabelecimento das espécies lenhosas florestais de dispersão zoocórica. A sucessão nas ilhas é retroalimentada positivamente, sendo que a chegada e o estabelecimento de novos indivíduos de espécies florestais atrai mais dispersores de sementes, acelerando a sucessão. A composição florística é dependente, também, da localização da ilha em relação às diferentes exposições solares. Para a sobrevivência de plântulas de M. palutris, a precipitação mensal é o parâmetro determinante primário. A distribuição dos hábitats afeta a dinâmica do avanço florestal na medida em que varia, entre eles, a taxa de sobrevivência de plântulas. Mesmo em anos de seca severa, ocorre o avanço da floresta em hábitats mais sombreados presentes no mosaico. Por fim, pode-se concluir que o avanço da floresta sobre a matriz herbácea é um fenômeno vinculado à natureza florística das matas adjacentes. Na matriz da paisagem, há diversos hábitats potencialmente disponíveis que modificam as chances de sobrevivência de plântulas e, portanto, contribuem, de formas diferentes, para a dinâmica da vegetação de comunidades de populações de espécies lenhosas. Entre os hábitats, a matriz herbácea é o hábitat mais inóspito para plântulas destas espécies e a borda da mata, o mais adequado. As ilhas apresentam uma condição intermediária entre os dois primeiros hábitats, sendo as mais desenvolvidas bastante semelhantes às bordas. Os mecanismos que retardam esta dinâmica são, em ordem de importância, a disponibilidade de água (em todos os hábitats), o herbivoria por vertebrado (nas ilhas e matriz herbácea), o fogo (na matriz herbácea) e, por último, a herbivoria por invertebrado (na matriz herbácea). / The region of granitic hills of Porto Alegre (State of Rio Grande do Sul) in southern brazilian’s region is contituted by a mosiac of forest and open areas (savannas or grassland). In these sites, according to the actual climate, the conditions are suitable for forest, which should mean that this vegetation type is in expansion, invading the surrounding herbaceous matrix. This expansion shows, mainly, an aggregation pattern by woody thickets of forest nucleation. The maintenance of the pattern is made by selective forces like precipitation, fire and herbivory (vertebrade and invertebrade) which are stronger in a juvenil phases of tree popualations. The selective forces act differently within all habitats available for seedling stablishment, creating a spatial arrangment of woody island. On the first chapter of this study, we analysed the floristic and spatial patterns of thickets island of forest expansion in a natural ecotone of forest-savanna, according to the slope (north, top and south) on Santan hill. It compares the floristic ans spatial patterns of thickets and forest edges in this three slopes. On the second chapter, we evaluated monthly, the seedling suvivorship of guamirim (Myrcia palustris DC.; Myrtaceae) under the influence of a severious drought (2004/2005) within differents habitats (forest edge, thickets and herbaceous matrix) of three slopes (north, top and south), during 15 months, by an experimental approach. In both cases, the study was carried out in Santana hill (30°03’ S, 51°07’ W), with maximum altitude 311m above sea level. In order to evaluate the floristic and spatial patterns, we selected 28 thickets of many cover areas (between 5,5 to 904m2) and four forest edges of two sizes (121,5 e 243m2). All woody individuals (h ³ 1m) were registered. The species, their abundance and dispersal modes were registered too. The floristic patterns were analysed by multivariate analyses. The relationship between cover area and diversity/richness or abundance and richness were analysed by linear regression. For survivorship of M. palutris seedlings, we analysed the survival/mortality rates and the mortality causes for this generalist tree. The survival and mortality causes were tested within three habitats and two controls: forest edge (B), thickets (I), herbaceous matrix (H), innercontrol (CI) and external-control (CE). The habitats were placed in three differents slopes (north, top and south). In all traits, we tranplanted isolated and group seedlings (1-4), to investigate the density-dependent mortality. We registered 4214 individuals (2828 in the thickets and 1386 in edge) amongst 38 botanical families and 111 species. The cluster analyses showed three groups: g1 with initial thickets of three slopes (predominance of top thickets), g2 also with thickets of three slopes (predominance of north thickets) and g3 with more development thickets of three slopes (predominance of south thickets) together with four forest edges. The dispersal by animals was dominant in all three groups. The autochorie dispersal showed an increase trend from g1 to g3, while anemochorie dispersal was in opposite way. The randomization tests indicates that g3 cover area (344,4m2) and g2 (156m2) are bigger (p = 0,001) than g1 (40,9m2) and g3 and g2 are no different (p = 0,063). The diversity (H’) was almost the same, with g3 (2,97) or g2 (2,16) higher (p = 0,001) than g1 (1,56), but g3 and g2 are slightly differents (p = 0,04). Without the forest edges samples, the thickets had differents richness and abundance (p = 0,02) from the north to the top or to the south, but no differences (p = 0,06) form the south to the top. The congruence analyses had maximum value (0,72) joining richness and abundance, however with more area cover, diversity and densitiy together, the value keeps high (0,7). Just one single parameter, cover area, could explain the increase of richness (R2 = 0,676; p < 0,001) and diversity (R2 = 0,49; p < 0,001). With increase number of individuals, increase the richness in g1 (R2 g1 = 0,63; p = 0,002), in g2 (R2 g2 = 0,66; p = 0,004), but not in g3 (R2 g3 = 0,6; p = 0,07). The lack of significance in g3 is related to the few sample units of this group (6). The results with M. palustris seedlings showed a inverse relation (R2 = 0,46; p = 0,006) between the mean mensal mortality (mm) and the the mensal precipitation for all habitats in all slopes. The suvivorship curve decreased more abruptly in the top than north or south (p < 0,01), but not form the south to north (p = 0,263). This curve were also different (p < 0,01) amongst all habitats, which makes the chance of survival be higher in B than I and nule in H. The selective forces that cause seedlings death are distinct amongst the habitats (p < 0,01): by invertebrade herbivory higher in H, by vertebrade herbivory higher in I and, drought/disease higher in B and by fire higher in H. Based on this results, we found this conclusions: The floristic and spatial patterns of forest expansion by woody thickets reflect the successional process that ranges from an initial and simple stage (more pioneer) to an latter and complex stage (less pioneer) similar with the forest edge. This process is limited by micro-habitats distribution which are suitable for forest tree seedlings with dispersal by animals. The successional process within the thickets have a positive feedback, where the arrival and stablishment of new individuals from forest species atract more seed dispersors, accelerating the process. The floristic composition depends also from the thicket slope location. For the survivorship of M. palutris seedling, the mensal precipitation was main determinant parameter. The habitat distribution affects the forest expansion dynamics due the differents suvival rates amongst them. Even with a severious drought, there is forest expansion in the mosaic within more shade habitats. We concluded that the forest expansion over the herbaceous matriz is a phenomena dependent from the surrounding forest edges. Within the mosaic landscape, there are several potencial habitats available wich modified the chances of seedling survival and, therefore, had distinct contributions to communities and popualtion dynmics of woody species. Amongst habitats, the herbaceous matriz is the less suitable for seedlings of this specie, while the forest edge is the most one. The thickets has the intermediate conditions, where the most development are very similar to the edges. The mechanisms wich delay this dynamic, on the decreased importance order, is the water availability (all habitats), the vertebrate herbivore (thickets and herbaceous matrix), the fire (herbaceous matrix) and the invertebrade herbivory (herbaceous matrix).