Juan F. Silva

Determinants of Tropical Savannas

Tropical savannas, defined as ecosystems formed by a continuous layer of graminoids (grasses and sedges) with a discontinuous layer of trees and/or shrubs, are the most common vegetation type (physiognomy) in the tropics. Tropical savannas are found over a wide range of conditions: rainfall from approximately 200 mm to 1500 mm a year, temperature from subtropical regimes such as the South American Chaco and the South-African savannas with temperature seasonality and cold-month average temperatures below 10 °C, to low-latitude savannas with no temperature seasonality, and soils from volcanic soils such as in parts of the Serengueti plains in Tanzania to dystrophic soils such as in the Brazilian cerrados. The one constant climatic characteristic of tropical savannas is rainfall seasonality. Yet the duration of the dry season can vary from 3 to 9 months, with a mode of 5 to 7 months.

Savannas of northern South America: a steady state regulated by water-fire interactions on a background of low nutrient availability

Savannas in northern South America (the seedlings depends on the length of PAM and is closely Orinoco Llanos) are found on a variety of highly leached related to the phenological pattern of the species. substrates, from tertiary sediments to alluvial soils, with The duration of PAM regulates specific and phenological markedly seasonal rainfall, 800-2500 mm/year. Physio- diversity of the grass layer. An extension of the dry season gnomic types range from tree-less grasslands to woodland may impair populations of early growers, whereas an early type communities--Dominant tree species are evergreen end of the rainy season affects negatively late species. and sclerophyllous being favoured by low soil fertility These effects are both direct, and are mediated by species and fire against deciduous, mesophyllous trees. Water competitive interactions. Longer PAM periods determine appears not to be a limiting factor for established trees higher probability for successful growth and reproduction since leaf flushing and flowering take place in the dry of annual species. season. Regular occurrence of fire determines low diversity of Tree/grass ratios increase with soil water availability the tree layer and may affect its productivity if it occurs during the dry season. Areas with high water table, or in after leaf flushing and the initiation of flowering. Fire also which a large fraction of the previous rainfall is accessible appears to maintain vigour of the herbaceous layer, its to tree roots, have higher tree densities than savannas with exclusion leading to deleterious changes. Fire frequency is soils of low water retention capacity and/or deep water also associated with length of PAM. The interplay of dry, tables. Tree recruitment is dependent on their capability to fire-prone years, and wet, fireless years probably deter- withstand grass-root competition during early growth mines short-term changes in the composition of the herba- stages, and to reach deeper soil layers to guarantee water ceous layer. Grass productivity is enhanced by fire during availability during dry periods. Therefore tree seedling the middle of the rainy season, when there is still some establishment appears associated with flush germination water left in the upper soil layers, and the rainy season during sequences of humid years. Duration of the season begins before water reserves are exhausted.

Population responses to fire in a tropical savanna grass, Andropogon semiberbis:a matrix model approach.

SUMMARY (1) To study the population-level effects of fire on the savanna grass Andropogon semiberbis, size-classified matrix population models were constructed for an annually burnt population and for a population protected from fire. These models were used to examine the effects of fire on population growth rate, stable size distributions and reproductive value, and to simulate different fire regimes. (2) The burnt population is capable of increasing rapidly (kX = 1 2524, r = 0-2251), whereas the unburnt population is unable to persist (kX = 0 2762, r = -1.2886). Most of this difference is due to effects on the growth, survival and reproduction of the smallest two size classes, which are shown by elasticity analysis to be the most important to population growth in both populations. The stable size distributions and reproductive values are similar in the two populations. (3) Both deterministic and stochastic analyses reveal a critical frequency of fire (= 0.85) below which this species is unable to maintain itself. This apparent reliance on fire frequency suggests that the spread and evolution of this species has been closely related to human occupation of neotropical savannas.

Leaf-cutting ants and forest groves in a tropical parkland savanna of Venezuela: facilitated succession?

Some savanna areas in the Orinoco Llanos region in Venezuela are characterized by the abundance of small forest groves, resulting in a parkland landscape. We hypothesized that forest groves are the result of colonization of the open savanna by forest species, facilitated by the activity of leaf-cutting ants. In this paper we present results from a field study conducted on an ecotonal belt between a deciduous gallery forest and an open savanna in the western Llanos of Venezuela. In seven transects, trees, groves and leaf-cutting ant nests were counted, size parameters evaluated and distance from grove to forest measured. Soil chemical analysis and microclimatic measurements were performed in the forest, grassland and groves. Floristic composition of the woodland component of the savanna (scattered trees) was very dissimilar from that of the forest, but the composition of the groves was essentially a mixture of both woodland and forest species. Woodland trees were present in all groves, whereas forest trees were present only in larger groves. The size structures of populations of the two types of trees differed significandy according to the size of the grove; also, woodland tree population structure differed significantly between open savanna and groves. The frequency of groves with Atta laevigata nests increased with the size of the grove, and groves with forest trees were positively associated with the presence of nests. A. laevigata nests improved the soil of groves, increasing N, Mg, Ca and organic carbon, but other soil properties were not modified. Temperatures at ground level in the dry season were ameliorated in the groves compared with the grassland, and this effect increased with the size of the grove. We concluded that a dynamic process of grove formation is facilitated by the aggregation of a few woodland trees, followed by the successful invasion of forest trees and a leaf-cutting ant nest. The results did not show any clear sequence of arrival of the ants or the forest trees, but it seems clear that these two would act synergistically, improving conditions for each other.

Fire, growth and survivorship in a Neotropical savanna grass Andropogon semiberbis in Venezuela

Two cohorts of seedlings of Andropogon semiberbis were tagged and monitored for up to three years. Several variables such as survivorship, plant size, seedling density, fire damage and distance to the nearest adult were measured or estimated. Two burning experiments to measure survivorship before and after fire, and two seedling-growth experiments to assess competitive interference from adult plants of three grass species were performed

Seed production and soil seed bank in three evergreen woody species from a neotropical savanna

Results of a 3-y study on the populations of three evergreen woody species (Byrsonima crassifolia, Palicourea rigida and Bowdichia virgilioides), from a regu- larly burnt savanna, in Venezuela, are reported. Tree density, size structure, fruit- seed production per tree of different size classes and soil seed bank were estim- ated. Viable seed production was one order of magnitude higher in B. crassifolia, which was also the only species with a permanent soil seed bank. Seeds of B. virgilioides germinated readily after dispersal, whilst seeds of P. rigida showed an innate dormancy synchronously released after a year. Both of the latter species seemed to maximize the first growth season by germinating at the beginning of the favourable season. In B. crassifolia, germination proceeds with an early small burst of germination, possibly followed by a second peak at the beginning of the next rainy season and little or no germination thereafter. The results showed that despite the high fire frequency, the studied species produce a substantial number of viable propagules, and have the capacity to germinate in field conditions (readiness depending on the species). Sexual reproduction is expected, therefore, to play an important role, as suggested by their reproductive investment, in long- term population maintenance.

Savanna Biodiversity and Ecosystem Properties

The overall question addressed here is the effect of different degrees of biodiversity on the function of savanna ecosystems. Function can be interpreted in two different ways. It can refer to the flow of energy and nutrients through an ecosystem or to the flow of species populations through time, i.e., the persistence of species populations and their properties, which we call the structure of the system. Here we discuss the effect of biodiversity on ecosystem function in this second sense. The role of biodiversity in the flow of energy and nutrients is addressed in Chapter 10.

Seed dispersal in Palicourea rigida, a common treelet species from neotropical savannas

The dispersal system of Palicourea rigida (Rubiaceae), a common woody species of neotropical savannas, was studied in a seasonal Venezuelan sav- anna. Production and consumption of fruits, dispersal agents, feeding behaviour, transport and effects on seed viability, were investigated by field observations, cage experiments and viability tests. To compare different dispersers, quantitative and qualitative dispersal parameters were calculated. About 59% of the fruit crop was dispersed successfully by birds. Although 10 different bird species consumed the fruits of Palicourea rigida, three bird species dispersed 79% of the fruits (Tyrannus melancholicus, Molothrus bonariensis and Elaenia chiriquensis). T. melancholicus (Tyrannidae) was the most important dispersal agent, with more than 39% of dispersed seeds. In contrast to the predictions of the specialization vs. generaliza- tion paradigm, this system achieves high effectiveness by means of predominantly unspecialized traits in both the plant and the dispersers. RESUMEN. Se estudioel sistema de dispersiond ePalicourea rigida (Rubiaceae), una lenosa comun de las sabanas neotropicales, en una sabana estacional de Vene- zuela. Mediante observaciones de campo y experimentos en jaulas y en el laborato- rio, se estudiaron parametros tales como produccion, transporte y consumo de frutos, los agentes dispersores y su comportamiento y los efectos sobre la viabilidad de las semillas. Para comparar los agentes de dispersion se estimaron parametros cualitativos y cuantitativos. Cerca del 59% de la cosecha total de frutos fue disper- sada existosamente por pajaros. Aunque los frutos fueron consumidos por diez especies de aves, el 79% de los frutos fueron dispersados por solamente tres espec- ies (Tyrannus melancholicus, Molothrus bonariensis y Elaenia chiriquensis). T. melancholicus (Tyrannidae) fue el mas importante agente dispersor con mas del 39% del total de semillas dispersadas. En contraste con el paradigma de especializacion vs. gen- eralizacion, este sistema alcanza alta efectividad pero se caracteriza por rasgos poco especializados, tanto en la planta como en los dispersores.

Architecture, seasonal growth and interference in three grass species with different flowering phenologies in a tropical savanna

In a field experiment we studied the relationships between architecture, seasonal aerial biomass growth and interference from neighbours in three savanna grass species differing in time of flowering: a precocious species (Elyonurus adustus), an early species (Leptocoryphium lanatum) and a late species (Andropogon semiberbis). To detect the neighbour interference upon architecture and seasonal regrowth, we measured the leaf interception of a plane at different heights, when the species grew alone, in pairs and in groups of three. Although the three species differed widely in the spatial and temporal patterns of occupation of above-ground space, important levels of interference among neighbouring plants were detected. The species differ in: a) the pattern of space occupation when growing alone; b) the magnitude of the interference effect by the companion species; c) the change in pattern as a consequence of interference; d) their reaction to fire. The interference is not symmetric and it is not related to phenological similarities based on flowering season.L. lanatum was both the most impaired and the least impairing of the three species and the opposite is valid forA. semiberbis. The basal species (E. adustus andL. lanatum) showed a higher potential to produce leaf surface during the growth season than the tall species (A. semiberbis), but the stronger interference from the latter tended to equate their growth when the species grew in mixtures. In all three species interference led to a reduction of the growing period. Based on the analysis of growth at the various heights and the architectural peculiarities of the species we concluded that neighbour interference is probably reducing both tillering and leafing in the basal species but only culm elongation and leafing in the erect species. The species also differed in their reaction to fire.A. semiberbis andE. adustus showed a pulse of regrowth after the passing of fire which is missing inL. lanatum. In all cases the neighbour interference affected both the intensity of growth and the length of the growing period, but it did not affect either the height of the plants nor the reproductive phenology of these three grass species.

Competition effects and responses to variable numbers of neighbours in two tropical savanna grasses in Venezuela

In a garden experiment we monitored tillering and flowering for target plants of two grass species, surrounded by one to five neighbours and compared them with control plants growing alone. We analysed the results using a non-linear regression. The species differed markedly in their seasonal patterns of growth but these differences faded when growing with two or more neighbours. The presence of neighbours had a significant depressing effect on both tillering and flowering in both species. Effects and responses were strongly asymmetrical concerning species and mode of growth. Andropogon semiberbis is in disadvantage against Trachypogon plumosus concerning vegetative growth. However, T. plumosus is strongly affected in its reproductive growth by competition not only from A. semiberbis but also from conspecific neighbours. We conclude that the complexity of interactions between effects and responses as well as vegetative and reproductive growth could have a net result of competitive equivalence or compensating competitive abilities of these two species.