This thesis explains two aspects of animal spatial foraging behaviour arising as a direct consequence of animals' need to drink water: the concentration of animal impacts, and the response of animals to those impacts. In semi-arid rangelands, the foraging range of free-ranging large mammalian herbivores is constrained by the distribution of drinking water during the dry season. Animal impacts become concentrated around these watering sites according to the geometrical relationship between the available foraging area and the distance from water, and the spatial distribution of animal impacts becomes organised along a utilisation gradient termed a "piosphere". During the dry season the temporal distribution of the impacts is determined by the day-to-day foraging behaviour of the animals. The specific conditions under which these spatial foraging processes determine the piosphere pattern have been identified in this thesis. At the core of this investigation are questions about the response of animals to the heterogeneity of their resources. Aspects of spatial foraging are widely commented on whilst explaining the consequences of piosphere phenomena for individual animal intake, population dynamics, feeding strategies and management. Implicated are our notions of optimal foraging, scale in animal response, and resource matching. This thesis addressed each of these. In the specific context of piospheres, the role of energy balance in optimal foraging was also tested. Field experiments for this thesis showed a relationship between goat browsing activity and measures of spatial impact. As a preliminary step to investigating animal response to resource heterogeneity, the spatial pattern of foraging behaviour/impacts was described using spatial statistics. Browsing activity varied daily revealing animal assessment of the spatial heterogeneity of their resources and an energetic basis for foraging decisions. This foraging behaviour was shown to be determined by individual plants rather than at larger scales of plant aggregation. A further experiment investigated the claim that defoliation has limited impact on browser intake rate, suggesting that piospheres may have few consequences for browser intake. This experiment identified a constraining influence of browse characteristics at the small scale on goat foraging by relating animal intake rate to plant bite size and distribution. Computer simulation experiments for this thesis supported these empirical findings by showing that the distribution of spatial impacts was sensitive to the marginal value of forage resources, and identified plant bite size and distribution as the causal factors in limiting animal intake rate in the presence of a piosphere. As a further description of spatial pattern, piospheres were characterised by applying a contemporary ecological theory that ranks resource patches into a spatial hierarchy. Ecosystem dynamics emerge from the interactions between these patches, with piospheres being an emergent property of a natural plant-herbivore system under specific conditions of constrained foraging. The generation of a piosphere was shown to be a function of intake constraints and available foraging area, whilst piosphere extent was shown to be independent of daily energy balance including expenditure on travel costs. A threshold distance for animal foraging range arising from a hypothesised conflict between daily energy intake and expenditure was shown not to exist, whereas evidence for an intermediate distance from water as a focus for accumulated foraging activity was identified. Individual animal foraging efficiency in the computer model was shown to be sensitive to the piosphere, while animal population dynamics were found to be determined in the longer term by dry season key resources near watering points. Time lags were found to operate in the maintenance of the gradient, and the density dependent moderation of the animal population. The latter was a direct result of the inability of animal populations to match the distribution of their resources with the distribution of their foraging behaviour, because of their daily drinking requirements. The result is that animal forage intake was compromised by the low density of dry season forage in the vicinity of a water point. This thesis also proposes that piospheres exert selection pressures on traits to maximise energy gain from the spatial heterogeneity of dry season resources, and that these have played a role in the evolution of large mammalian herbivores.

In South Africa, the unsustainable use of natural resources by domestic livestock has led to resource depletion and serious land degradation. Rangeland degradation, especially bush encroachment and soil erosion, is particularly acute in the North-West Province, where all districts show signs of desertification and a loss of biodiversity resulting in a deterioration of human and animal health. This has a major impact on livestock productivity and the economic viability of livestock farming with serious consequences for the livelihoods of pastoral communities. It is important to recognise ecological change before irreversible changes occur. The aim of this study, which falls within the Global Environmental Facility Desert Margins Programme (GEF-DMP), was to investigate to what extent vegetation in combination with ant communities can be used as indicators of ecosystem change due to anthropogenic human induced land-use patterns and how can this information be used in land degradation management and biodiversity conservation in the semi-arid western rangelands of Southern Africa. Sites, representing a degradation gradient (relative poor and relative good rangeland condition extremes) within each of three Tribal-, three Commercial- and three Reserve areas, were surveyed. The impacts of these land uses on the herbaceous species composition, woody-, soil- and ant components were evaluated. Both the woody and herbaceous species components reflected the existence of a rangeland condition/degradation gradient across the larger study area. The herbaceous species composition reflected similar degradation tendencies within the Commercial and Reserve land uses, with sites being associated with low rangeland as well as high rangeland condition scores. The tendencies differed between these two land uses based on the woody degradation gradient. The entire Tribal herbaceous- and woody species components showed a transitional shift towards another state, which differed significantly from the Commercial and Reserve land uses. Both the Tribal herbaceous and woody components were associated with low to intermediate rangeland condition ranges, with no significant rangeland condition gradient existing within the Tribal land use. Understanding and quantification of the soil-vegetation dynamics hold important implications for rangeland degradation management. This study provided criteria for selecting the most appropriate measures when incorporating the soil parameters as additive data in the multivariate analyses with the vegetation, ant and nominal environmental data. Different land use practices resulted in different soil patterns, with significant gradients pertaining to the soil stratum and openness/woodiness groups. There was a significant though neglectable difference pertaining to the rangeland condition/degradation gradient based on the soil component. Ants have been extensively used as bio-indicators, also with regard to the monitoring of the environmental effects of rangeland pastoralism. Ant species compositional patterns and functional groups displayed congruent clustering and diversity patterns as those of the vegetation and soil components. In contrast to the vegetation components, ant assemblages did not reflect a degradation gradient, but rather reflected environmental changes (modifications) to the habitat structure and - heterogeneity as a result of different land use disturbances. Both vegetation and ant diversity measures were mainly associated with the Tribal land use. These diversity indices were indicators of habitat complexity, heterogeneity and moderate disturbance, rather than indicators of a rangeland condition/degradation gradient. The diversity patterns are best described by a dichotomy between the humped-shaped productivity/diversity and the habitat complexity/heterogeneity models. Vegetation and ant diversity measures for this study should be considered as environmental indicators of habitat disturbance rather than as biodiversity indicators. It is suggested that vegetation, soil and ant patterns are best described by the state-and-transition model, which encompasses both equilibrium and non-equilibrium systems. The resilient nature of these rangelands, typical of non-equilibrium systems, was reflected by the low to intermediate differences between land uses with regard to the herbaceous, woody, soil and ant components. However, densitydependent coupling of herbivores to key resources resulted in transitional shifts and modification of the vegetation composition and structure within and between land uses, displaying the equilibrium dynamics pertaining to these rangelands. Small disturbances in these rangelands may result in detrimental “snowball” interactive biotic-biotic /abiotic cascades. Spatial heterogeneous patterns within and between land uses as displayed by the vegetation, soil and ant parameters, necessitate that monitoring and management at patch, paddock and landscape scale should be conducted, cautioning against the extrapolation and over simplification of management strategies across all land uses. Because these arid rangelands are linked socio-ecological systems, it is not possible to address biophysical issues associated with land degradation without including the human dimensions. A “Key assessment matrix” is provided for monitoring and management purposes pertaining to land degradation and diversity aspects within and between the different land uses, and can be used by the land user, extension officer and scientist. / Thesis (Ph.D. (Botany))--North-West University, Potchefstroom Campus, 2007.

In South Africa, the unsustainable use of natural resources by domestic livestock has led to resource depletion and serious land degradation. Rangeland degradation, especially bush encroachment and soil erosion, is particularly acute in the North-West Province, where all districts show signs of desertification and a loss of biodiversity resulting in a deterioration of human and animal health. This has a major impact on livestock productivity and the economic viability of livestock farming with serious consequences for the livelihoods of pastoral communities. It is important to recognise ecological change before irreversible changes occur. The aim of this study, which falls within the Global Environmental Facility Desert Margins Programme (GEF-DMP), was to investigate to what extent vegetation in combination with ant communities can be used as indicators of ecosystem change due to anthropogenic human induced land-use patterns and how can this information be used in land degradation management and biodiversity conservation in the semi-arid western rangelands of Southern Africa. Sites, representing a degradation gradient (relative poor and relative good rangeland condition extremes) within each of three Tribal-, three Commercial- and three Reserve areas, were surveyed. The impacts of these land uses on the herbaceous species composition, woody-, soil- and ant components were evaluated. Both the woody and herbaceous species components reflected the existence of a rangeland condition/degradation gradient across the larger study area. The herbaceous species composition reflected similar degradation tendencies within the Commercial and Reserve land uses, with sites being associated with low rangeland as well as high rangeland condition scores. The tendencies differed between these two land uses based on the woody degradation gradient. The entire Tribal herbaceous- and woody species components showed a transitional shift towards another state, which differed significantly from the Commercial and Reserve land uses. Both the Tribal herbaceous and woody components were associated with low to intermediate rangeland condition ranges, with no significant rangeland condition gradient existing within the Tribal land use. Understanding and quantification of the soil-vegetation dynamics hold important implications for rangeland degradation management. This study provided criteria for selecting the most appropriate measures when incorporating the soil parameters as additive data in the multivariate analyses with the vegetation, ant and nominal environmental data. Different land use practices resulted in different soil patterns, with significant gradients pertaining to the soil stratum and openness/woodiness groups. There was a significant though neglectable difference pertaining to the rangeland condition/degradation gradient based on the soil component. Ants have been extensively used as bio-indicators, also with regard to the monitoring of the environmental effects of rangeland pastoralism. Ant species compositional patterns and functional groups displayed congruent clustering and diversity patterns as those of the vegetation and soil components. In contrast to the vegetation components, ant assemblages did not reflect a degradation gradient, but rather reflected environmental changes (modifications) to the habitat structure and - heterogeneity as a result of different land use disturbances. Both vegetation and ant diversity measures were mainly associated with the Tribal land use. These diversity indices were indicators of habitat complexity, heterogeneity and moderate disturbance, rather than indicators of a rangeland condition/degradation gradient. The diversity patterns are best described by a dichotomy between the humped-shaped productivity/diversity and the habitat complexity/heterogeneity models. Vegetation and ant diversity measures for this study should be considered as environmental indicators of habitat disturbance rather than as biodiversity indicators. It is suggested that vegetation, soil and ant patterns are best described by the state-and-transition model, which encompasses both equilibrium and non-equilibrium systems. The resilient nature of these rangelands, typical of non-equilibrium systems, was reflected by the low to intermediate differences between land uses with regard to the herbaceous, woody, soil and ant components. However, densitydependent coupling of herbivores to key resources resulted in transitional shifts and modification of the vegetation composition and structure within and between land uses, displaying the equilibrium dynamics pertaining to these rangelands. Small disturbances in these rangelands may result in detrimental “snowball” interactive biotic-biotic /abiotic cascades. Spatial heterogeneous patterns within and between land uses as displayed by the vegetation, soil and ant parameters, necessitate that monitoring and management at patch, paddock and landscape scale should be conducted, cautioning against the extrapolation and over simplification of management strategies across all land uses. Because these arid rangelands are linked socio-ecological systems, it is not possible to address biophysical issues associated with land degradation without including the human dimensions. A “Key assessment matrix” is provided for monitoring and management purposes pertaining to land degradation and diversity aspects within and between the different land uses, and can be used by the land user, extension officer and scientist. / Thesis (Ph.D. (Botany))--North-West University, Potchefstroom Campus, 2007.