The immediate aim of this study was to relate herbage availability, liveweight gain and stocking rate to one another under continuous and rotational grazing so that management recommendations could be formulated for Coastcross II Bermudagrass and kikuyu. In addition, these data would be used for an economic evaluation of growing out young beef animals on these two pasture species. The broad long term goal was to use this information to persuade farmers to make more use of planted pastures and thereby reduce the stocking pressure on the veld. Besides pursuing the primary objectives, the study offered an opportunity to make an observational assessment of the put-and-take technique for use in grazing trials. In relation to fixed stocking, variable stocking appeared to have no distinct advantage in this study. This conclusion is drawn from the fact that little success was achieved in applying put-and-take at a high level of precision. It was therefore recommended that future trials should make use of fixed stocking, except when variation in herbage availability is very large and predictable. In such cases the use of put-and-take would be warranted. Liveweight gain of animals was linearly related to herbage availability. In two seasons liveweight gain of animals on Coastcross was higher for continuous grazing than for rotational grazing at equivalent herbage availability, but in other seasons there was no difference between the two methods of grazing. At equivalent levels of herbage availability liveweight gain was higher in early summer than in late summer. Herbage availability decreased linearly witilin each season and on average, rotational grazing resulted in increasingly greater herbage availability than continuous grazing, as stocking rate was increased on Coastcross. However, no difference between the two grazing methods was evident on kikuyu. Finally, the relationship between liveweight gain and stocking rate was also described by a linear function. On average, the stocking rate at which maximum liveweight gain per ha occurred (SRmax) was higher for rotational grazing than continuous grazing on Coastcross, but not on kikuyu. However, no difference was evident between the two grazing methods in each season. During the study period annual rainfall varied from 506 mm to 990 mm. This offered a unique opportunity to examine the relation between some of the pasture production parameters measured and annual rainfall. The length of the grazing season and SRmax increased, but liveweight gain of animals at SRmax declined as annual rainfall increased. This caused seasonal liveweight gain at SRmax to increase initially, but then to reach a maximum and decrease as annual rainfall increased from 500 to 1000 mm. From this information it was possible to build two models which can be used to predict liveweight gain per ha and profit per ha from stocking rate and annual rainfall. These models can either be built into farm planning programmes or used directly by agricultural advisors. In tl1is thesis they have been used to show that there is a wide range in stocking rates and levels of annual rainfall within which it is possible to make substantial profits by grazing young beef animals on dryland pastures. This study has therefore provided forceful information which can be used to persuade farmers to make more use of planted pastures, and in so doing, conserve the veld. / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 1983.

Two methods of estimating the intake of a 25% leucaena : 75% grass hay diet by young male Boer goats were assessed, (a) The purine derivative technique which uses the urinary excretion of purine derivatives (expressed relative to creatinine concentration in the same sample) as an index of feed intake, and (b) the conventional marker method, utilizing chromic oxide (Cr₂O₃) contained in gelatin capsules and dosed twice daily, as the marker. Following a prerun the two techniques were compared in three runs of an indoor experiment. In each run 10 goats were randomly allocated to five feeding levels (500 to 1100 g fodder d ¯¹ on air dried basis). A preliminary and an adaptation period during which goats were dosed with the Cr₂O₃ and fed their daily feed allowance, was followed by a 4 day collection period during which spot samples of urine were collected and analysed for allantoin and creatinine (allantoin being used instead of total PDs) and faecal samples were collected for chromium analysis and percentage dry matter determination. For the first two runs, two grab samples per day for each goat were bulked and analysed for chromium content. For the last run, the total daily faecal collection was subsampled and analysed for chromium. Work was done in metabolic crates to determine the effect of time of collection on the ratio of allantoin : creatinine (A/C) in spot urine samples and it was found to non-significant (P>0.05). Linear regressions of: (a) feed intake expressed per unit metabolic mass (g.d ¯¹.LW ¯°∙⁷⁵ (I_mmass)) against A/C ratio; (b) faecal output (g d ¯¹) against feed intake (g d ¯¹); and (c) faecal chromium concentration (mg kg ¯¹) against faecal output (g d ¯¹) were fitted to the data. During the prerun, only regression (a) was fitted and was non-significant (P>0.05), showing no trend at all. For the first true run, the regression of I_mmass against A/C ratio was significant and the correlation was high (P≤ 0.001, R² [A] 0.715, n = 10) but for the second and third runs, the correlations only became significant when the apparent outliers were discarded from the data. (Run 2: P≤ 0.001, R² [A] 0.824, n = 8; Run 3: P≤ 0.05, R² [A] 0.430, n = 9). It was concluded that the relation between I_mmass and A/C ratio is not well enough defined to be used for predictive purposes. When regression (b) was investigated, all the runs produced significant results (P≤0.001, P≤0.01, P≤ 0.001 for runs 1,2 and 3 respectively) however the correlations were not as high as expected (R² [A] being 0.714, 0.565 and 0.863 respectively). For the regression of faecal Cr concentration against faecal output (regression c), all runs showed significant relations (P< 0.001, P< 0.0001, P:s 0.001 for runs 1,2 and 3 respectively) and the correlations were high (R²[A] being 0.836, 0.837 and 0.912 respectively). The data from the three runs were pooled and single equations established for regressions (b) and (c) to allow for the prediction of intake from faecal chromium concentration. Faecal output = feed intake * 0.448 + 19.341 (P≤ 0.001, r 0.853, R² [A] 0.718, SE 25.664, n - 30) Faecal chromium concentration = faecal output * -241.547 + 1.315E+05 (Ps 0.001, r 0.904, R² [A] 0.811, SE 5603.788, n = 30). In vitro figures were determined for a range of leucaena : hay mixes but no apparent trend was found between percentage leucaena in the mix and the digestibility of the mix. These results compared favourably with in vivo results obtained for a 25% leucaena : 75% hay mix. Neither technique proved entirely satisfactory, but the external marker method was found to be more effective than the purine derivative technique. More work is required especially with respect to the latter method. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 1997.

An area and a point-based technique were used together at each of the same 75 sampling sites (stands), on a Sourish-Mixed Bushveld farm, to collect data for the classification and mapping of the vegetation. Both sets of data were synthesized using the same computer program package and the efficacy of the resulting classifications as well as the efficiency of the two field sampling techniques was compared. Following this, a continuous 7 752 point (1 m apart) transect was carried out, traversing the farm, in order to determine the optimum scales at which to sample Sourish-Mixed Bushveld so as to increase classification efficacy and improve community boundary recognition. The results indicated that (1) the arbitrarily chosen sampling scale of 1:8 000 was too large for "farm-scale" studies; (2) the area-based method proved to be satisfactory in that the classification and vegetation map produced with this method were verified spatially and environmentally; (3) the point-based method was deficient as a classificatory and mapping tool at large scales, since too few species were recorded with this method to make any sense of the classification and mapping of the vegetation was not possible; (4) less time per species was spent using the area-based method but because more species per stand were recorded with this method, the point-based method was quicker per stand; (5) the area-based method was easier to use in dense vegetation and irregular terrain; and (6) the optimum sampling scales for Sourish-Mixed Bushveld, as indicated by the synthesis of the continuous transect data, are about 1:12 000, 1:50 000 and 1:250 000. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 1995.

The primary objective of this study was to increase net farm income in the Highland Sourveld of Natal by developing pasture based fat lamb production systems. Secondary objectives included (a) determination of the most suitable pasture species or species mixture; (b) determination of the optimum level of N fertilization; (c) establishment of biological and economic optimum stocking rates; (d) projection of expected profit and (e) verification of proposals on a farm scale. Midmar Italian ryegrass was compared to selected legumes alone, and in mixtures with ryegrass in a small plot trial conducted over two years at a 3- and 6-week harvesting frequency under irrigation. Ryegrass in combination with red and white clover provided the highest yields for both harvesting intervals in both years. When this pasture was subsequently compared to ryegrass alone under grazing, no significant difference could be detected in lamb gains at 20 and 28 ewe-lamb units/ha, but at 36 ewe-lamb units/ha the lambs on ryegrass started losing weight at 30 kg, while those on ryegrass + clover continued to gain until a market weight of 40 kg was achieved. In a small plot trial little response was detected to added N above 175 kg N/ha on ryegrass + clover, and under grazing, no significant difference was observed between 275; 375 and 4,75 kg N/ha over two single year periods. Relationships between average daily gain (ADG) and stocking rate, ADG and disc meter height, and stocking rate and disc meter height were mostly well described by linear functions. Expected pre-weaning AOG was 150 g higher (p ≤ 0,01) than post-weaning ADG (100 g) in the first year, and a similar difference was observed in the second year. For the period common to all stocking rates an average maximum gain/ha of 824 kg was achieved at a stocking rate of 33,5 lambs/ha and a disc meter height of 4,5 cm. By lambing in April, lambs could be marketed in September at 40 kg and more weaned lambs could be grazed on the pasture and marketed in December, leading to an expected gain/ha of 1400 to 2200 kg, and expected profit/ha of R 1254 to R 2771, depending on stocking rate. When evaluated under farm conditions, this system produced 2060 kg live weight/ha and a profit of R 3206/ha. It is concluded from the present study that the fat lamb production system based on irrigated ryegrass + clover pastures can be recommended with confidence to farmers in the Highland Sourveld of Natal. / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 1986.

Four grazing trials to characterise cultivated pastures, in terms of beef production, were conducted in Bioclimate 3 of Natal. Dual purpose and British beef type cows were run on kikuyu at stocking rates from 2,81 to 7,30 cows plus calves per ha. During the eight seasons of the trial the seasonal rainfall varied from 580 to 933 mm. There was a positive linear relationship between rainfall and pasture yield with maximum yield of kikuyu being recorded during February - March. Stocking rate affected pasture yields only during favourable rainfall seasons. Crude protein (CP) and crude fibre (CF) of kikuyu fluctuated markedly within and between seasons. However, CP increased and CF decreased as stocking rate increased. There were significant relationships between stocking rate and (a) calf performance, (b) calf livemass gain, (c) period required to attain maximum mass, (d) period on pasture for the cows, and (e) cow mass change: Weaners were run on irrigated Italian ryegrass at 5, 7 and 9 weaners per ha for four seasons. Stocking rate had little effect on the growth pattern of the pasture but affected dry matter yields. Reducing the stocking rate resulted in increased pasture yields and CF content but reduced CP levels of material on offer. Steers exhibited higher gains than heifers but lower carcass grades and stocking rates for maximum gain per ha (SRmax). Livemass gains of 1315 and 1224 kg per ha can be expected at SRmax of 6,85 and 9,54 for steers and heifers respectively. Yearling heifers run at four stocking rates on kikuyu for one season showed a negative linear relationship between stocking rate and gain and a positive linear relationship between pasture height and gain. A SRmax of 8,85 allows for a livemass gain of 1 040 kg per ha. The effect of feeding concentrates on foggaged kikuyu was evaluated. Foggaged kikuyu can be used as a source of roughage for fattening steers. However, as the steers became adapted to the concentrate the intake of kikuyu declined from 39 to 19% of their daily intake. Regressions derived from the characterisation trials allow for developing beef systems for different situations. / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 1985.

Although controlled burning has been used to manage Highland Sourveld grasslands, little was known of its effects on the vegetation. This study examined the effects of past fire treatments on veld condition, species composition, dry matter production, quality and canopy recovery growth rates of these grasslands. Also, six techniques of estimating the species composition of grasslands were compared in order to decide on a standard technique for monitoring these grasslands. From this work it was concluded that the wheel point method is the most satisfactory. Veld condition scores were significantly lower in grassland protected from fire than where veld had been burnt or burnt and grazed at regular intervals. Frequent defoliation was found to maintain the grassland composition largely unchanged over a period of 30 years. Individual species were, however, found to react strongly to defoliation frequency. Plant demographic studies were therefore carried out to explain this differential response to burning. Three Decreaser and two Increaser I species were studied. In all species examined, recruitment of secondary tillers was stimulated by regular burning, each species being well adapted to a regular fire regime. Differential responses to burning were best explained by the combined effects of the different reproductive capacities and mortality rates of tillers of these species. A biennial spring burning regime was shown to be most suitable for maintaining the most important grass species at their present levels of abundance. Annual winter and biennial spring burning did not result in significant differences in dry matter production. Maximum net productivity was approximately 230 g/m² in both treatments, placing them amongst the more productive areas of Southern Africa. Examination of canopy recovery growth rates showed that there is little difference in the percentage canopy cover at the end of the growing season when veld is burnt annually in winter or biennially in spring. However, differences in season of burn resulted in exposure to erosive forces at different times of the year. The results of this investigation have highlighted the importance of regular burning during the dormant period in the montane grasslands of Natal. / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 1985.

Abstract can be viewed in PDF document. / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 1994.

A predictive understanding of plant community response to various environmental influences (e.g. type, timing and frequency of disturbance, site productivity, fertilization, etc.) is a general goal of plant ecology. This study sought to further understanding of mesic grassland dynamics in KwaZulu-Natal using long-term field experiments (> 50 years) and short-term pot and plot competition experiments. The specific objectives were to: 1) examine the effects of long-term burning of grassland on soil organic matter content because of its potential impact on nitrogen cycling and community composition, 2) examine patterns of community composition and species richness in response to different type, timing and frequency of disturbance (burning and mowing) in a long-term grassland burning and mowing experiment and to different type and amounts of fertilizer application in a long-term grassland fertilization experiment, 3) develop hypotheses concerning the response of different species to disturbance and fertilization, and test these hypotheses using pot and plot competition experiments, and 4) provide a general synthesis of the results of the various field, pot and plot experiments that may be used to develop a predictive theoretical framework for mesic grassland dynamics. Total soil nitrogen was lowest in sites burnt annually, intermediate in sites burnt triennially and highest in sites protected from disturbance and sites mown annually in the dormant-period (spring or winter). Winter burning reduced soil organic carbon and total soil nitrogen more than spring burning. Mineralizable nitrogen was reduced by burning. The different effects of type, timing and frequency of disturbance on total soil nitrogen appeared to be an important determinant of community composition and species richness. Short-grass species (Themeda triandra, Eragrostis capensis, Heteropogon contortus, Diheteropogon amplectens and Eragrostis racemosa) were most abundant in annually burnt sites (especially winter burnt sites), whereas medium and tall-grass species (Eragrostis curvula, Cymbopogon spp., Hyparrhenia hirta and Aristida junciformis) were most abundant in triennially burnt sites, sites protected from disturbance and sites mown annually in the dormant-period, all of which had higher total soil nitrogen than annually or biennially burnt sites. Species richness and short-grass species declined with increasing levels of nitrogen fertilization in the fertilizer experiment and declined with increasing productivity and nitrogen availability in both the fertilizer and burning and mowing experiments. Thus, it was hypothesized that the type, timing and frequency of disturbance resulted in different compositional states through different effects on soil resources (especially nitrogen), which affected the competitive balance between short and tall species. The hypothesis that composition was determined by disturbance-mediated soil nitrogen availability was supported by competition experiments, which revealed that shortgrass species were most competitive in low-nutrient/low-productivity treatments and tall-grass species most competitive in high-nutrient/high-productivity treatments. The fertilizer experiment and a competition experiment revealed that tall broad-leaved species were most competitive in sites of highest productivity, fertilized with both nitrogen and phosphorus, whereas tall narrow-leaved species were most competitive in sites of intermediate productivity, fertilized with nitrogen only. It was hypothesized that summer mowing increased the abundance of short-grass species and decreased the abundance of tall-grass species in the burning and mowing experiment by increasing the competitive ability of short-grass relative to tall-grass species, rather than the tall-grass species being less tolerant of mowing. A competition experiment revealed that tall-grass species (Hyparrhenia hirta and Panicum maximum) were as tolerant of cutting as a short-grass species (Themeda triandra). Themeda triandra was shown to become extremely competitive in cutting treatments, reducing the biomass of most other species relative to their monoculture biomass, showing that its dominance of mown sites in the burning and mowing experiment was a result of its superior competitive ability rather than greater tolerance of mowing. However, many tall erect herbaceous dicots appeared to be intolerant of summer mowing, probably because their meristems are aerial and easily removed by mowing, whereas short creeping herbaceous dicots were increased by summer mowing probably because their meristems were below the mowing height. Further, these short species would be vulnerable to shading in unmown sites. Thus, species with basal meristems (hemicryptophytes) or meristems near the soil surface (geophytes) appear to be more tolerant of mowing than species with aerial meristems (phanerophytes), but the tradeoff is that a low meristem height renders them vulnerable to shading in unmown sites. Very high litter accumulation in the sites protected from disturbance appeared to have a direct influence on community composition and species richness. Species that dominated these sites (e.g. Tristachya leucothrix & Aristida junciformis) initiated tillers below-ground and had sharp erect shoots that appeared to be an adaptation for penetrating litter. Species that initiate tillers below-ground are probably less vulnerable to the effects of shading by litter because their tiller initiation is not dependent on high light availability. The fact that Aristida junciformis was shown to have very low competitive ability in two competition experiments, suggests its dominance of protected sites was through tolerance of high litter levels rather than competitive exclusion of other species. Low grass species richness in these sites was probably a result of an inability of many species to tolerate these high litter levels. This study has revealed that inherent site productivity and its interaction with the effect of disturbance on soil resources and litter levels is a major determinant of community composition and species richness. The effect of type, timing and frequency of disturbance on soil nitrogen was able to account for the principal changes in community composition. Thus, the influence of disturbance on soil nitrogen is a unifying principle in plant ecological theory that enables greater understanding of disturbance-composition relationships. However, intolerance of certain forms of disturbance (e.g. mowing) by species with aerial meristems, or intolerance of accumulating litter in the absence of disturbance by species without sharp erect shoots, may also have important influences on composition. In addition, this study has revealed that plant traits (height, leaf width, position of tiller initiation, shoot morphology and position of meristems) were well correlated with the various effects of disturbance and fertilization on community composition, indicating that a plant trait-productivity-disturbance framework has great potential for understanding and predicting species response to disturbance and multiple limiting nutrients. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2003.

Observations by local people in the Mier area, southern Kalahari, South Africa, indicated that degraded rangeland does not recover within a time frame acceptable to landowners. Pristine vegetation in this linear dune system consists largely of a herbaceous layer dominated by perennial grasses. Woody vegetation is sparse on dunes and interdunes in good condition. The dunes and interdunes react differently to disturbance, probably because of differences in substrate stability, soil particle size distribution and consequent differences in soil nutrient distribution and moisture content. Degraded dunes are devoid of any vegetation, except for Stipagrostis amabilis, a rhizomatous grass which remain in small clumps, and the tree Acacia haematoxylon. The latter increases in numbers probably due to the high moisture content in degraded dunes. Moisture content in degraded dunes remain high even during prolonged dry periods. Interdunes are more susceptible to degradation and are invaded by the long-lived shrub Rhigozum trichotomum and the annual grass Schmidtia kalahariensis. Both these species compete with perennial grasses for moisture. Additionally, depleted seed banks and increased seed predation by ants (Messor capensis) may also affect the re-establishment of perennial grasses in the interdunes. The main conclusion from this study is that degraded southern Kalahari rangeland cannot recover spontaneously at the landscape scale because of a negative feedback mechanism that prevents establishment and growth of seedlings. The hypothesis put forward is that rangeland ceases to react to rainfall as an ecosystem driver as it becomes degraded. Once degraded, wind controls the dynamics of the system and recovery, irrespective of rainfall, is almost impossible. In contrast, detailed studies at the level of small isolated populations of S. amabilis and S. ciliata on degraded dunes indicate high rates of population growth. The scales at which these processes exist are contradictory and may not be contradicting. Rangeland managers should take cognisance of the influence of the spatial and temporal scale at which they operate, and on which they base their decisions. / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 2000.

Mistletoes intimately connect to their host trees with a haustorium that allows them to access nutrients and water. Mistletoes in South Africa vary greatly in their degree of host specificity. Most species occur on a wide range of host families, while others are restricted to a single host family or—at the extreme—to a single host species. Mistletoes that are host generalists at a larger spatial scale may become host-specific at a local scale. One of the challenges in mistletoe biology is determining the factors that maintain local host specificity. Birds potentially reinforce the mistletoe–host interactions by direct dispersal. However, many mistletoe species coexist while parasitising different co-occurring host species. This suggests that host trees may impose more selection pressure than birds in determining host specificity. Thus, my thesis examines the role of host trees as ecological and physiological filters that influence the infection patterns and determine host specificity of mistletoes in South Africa. The second chapter of this thesis synthesises the literature on host specificity in mistletoes. I then present the results of four field and laboratory experiments that were used examine the features affecting host specificity in representatives of two families of mistletoes (Loranthaceae and Viscaceae) in South Africa. My main research objectives focus on host abundance and morphology, host compatibility, host water and nutrient content, abiotic influences on mistletoe seedling survival and growth and mistletoe–host stomatal morphology in relation to water potential that affect nutrient acquisition by mistletoes from their host trees. The geographic mosaic approach was explored as a potential explanation for the mistletoe–host interactions that direct host specificity in mistletoes. I synthesised the available literature on the mechanisms and factors that direct mistletoe host specificity. This was supported by data analysed from South African herbarium collections, books describing the South African flora and field observations in South Africa. I suggest that host abundance (host availability through time and space) and host compatibility (as determined by genetic, morphological, physiological and chemical factors) play a primary role in determining host specificity in South African mistletoes, while differential bird dispersal strengthens or weakens mistletoe–host interactions. Analysis of the network structure of mistletoe–host interactions at different levels (e.g., at the level of population, species and genus) followed by genetic and reciprocal germination experiments may reveal the patterns and mechanisms of host specificity in mistletoes. I quantified the mistletoe–host composition, height of potential host trees and nutrient and water content of mistletoes and their hosts at Pniel Estates. Surveys of the study site revealed a single mistletoe species, Viscum rotundifolium, parasitising only Ziziphus mucronata and Ehretia rigida. Both parasitised host species were not the most abundant trees, were not the tallest trees and did not have the highest water or nutrient content of trees in the area, although these factors have been found to be good predictors for mistletoe parasitism in other studies. Subsequently, I tested mistletoe–host compatibility by conducting a germination experiment in the greenhouse by inculcating seeds of V. rotundifolium on freshly cut branches of nine available potential host trees. I found that mistletoe seeds had a greater chance of attachment and subsequent survival on branches of E. rigida and Z. mucronata as compared with seeds on co-occurring Acacia and other potential host species. This suggests that host compatibility plays a role in directing the host specificity of V. rotundifolium at Pniel Estates. I found that individuals of V. rotundifolium had more negative water potentials than their host trees and, by doing so, they passively maintain the flow of nutrients. In addition, I found evidence that the mistletoe uses active uptake to access nutrients from host phloem because the leaf tissue of a mistletoe had a nitrogen-to-calcium ratio (N:Ca) >1. Conventionally, a high N:Ca ratio (>1) in the leaf tissue of a mistletoe is taken as evidence of active uptake from host phloem because N is highly phloem-mobile while Ca is a large molecule and is phloem-immobile. This method has shortcomings discussed at greater length in the chapter but my findings suggest that the mistletoe V. rotundifolium uses a combination of passive and active nutrient uptake. I quantified the mistletoe–host community composition and host physical features (height and diameter at breast height) in two sites in KwaZulu-Natal, South Africa—Highover and Mtontwane. The mistletoe Agelanthus natalitius (Loranthaceae) is common at both sites, parasitising the most abundant host species—Acacia karroo—and the second most abundant host tree—Acacia caffra. Prevalence of mistletoe infection (percentage of trees parasitised) was positively correlated with tree size (height and diameter at breast height). The two host species did not differ significantly in height. At Highover the host species A. caffra and A. karroo had a similar prevalence of mistletoe infection but at Mtontwane a significantly higher percentage of A. caffra trees was parasitised in comparison with A. karroo. However, the intensity of mistletoe infection (mean number of mistletoes per tree) was lower for A. caffra (Highover: 0.66 ± 0.01, Mtontwane: 0.89 ± 0.04) than for A. karroo (Highover: 0.73 ± 0.04, Mtontwane: 1.03 ± 0.64). There were two highly infected big trees in Highover and one in Mtontwane where many mistletoe-dispersing birds were nesting which inflated the numbers for intensity of mistletoe infection in A. caffra, however. I tested mistletoe–host compatibility by conducting a reciprocal transplant experiment in the two study sites. I applied a paired design, using one local and one non-local mistletoe seed in each pair, with seed pairs placed on the two main host species at the different sites. Except in Highover where an unidentified pathogen retarded growth and survival, mistletoe seeds placed on the same substrate and in the same site as their source host grew a longer hypocotyl and had greater survival. Regardless of source, mistletoes placed on A. karroo had longer hypocotyls and greater survival than mistletoes on A. caffra. These results suggest that there may be adaptation of the mistletoe Agelanthus natalitius to the most frequently encountered host species, Acacia karroo. To simulate the conditions encountered by mistletoes during the dry and cold South African winter, mistletoe seedlings were monitored at different levels of microclimate (light, temperature and moisture) in a growth chamber. I found that higher light availability (20% and 40% shade versus 80% shade), cool temperatures (15°C and 20°C versus 25°C) and continuous moisture availability improved seedling development and subsequent survival of two mistletoe species (Viscum rotundifolium and Agelanthus natalitius). I studied the leaf stomata of two host–mistletoe pairs (Acacia karroo–Agelanthus natalitius and Vitex obovata–Erianthemum dregei) using a scanning electron microscope to investigate some of the underlying mechanisms that enable mistletoes to maintain more negative water potentials than their host trees and at the same time control water loss. In addition, I examined the response of mistletoes to the application of abscisic acid (ABA), a plant growth regulator that controls stomatal closure. I found that the mistletoes had a higher density of stomata and had larger stomata than their host trees. In addition, both mistletoe and host leaves closed their stomata during midday and in response to exogenous ABA. The ability of mistletoes to control water loss in this way may be one reason why mistletoes rarely kill their host trees, which would be maladaptive. The mistletoes used in my studies are known to be host generalists at a larger spatial scale but I found that they were host specific at a local scale. The results of my research suggest that host abundance and compatibility play a role in directing host specificity, while host nutrient and water status have little effect on host specificity at this local scale. The interactions between the generalist mistletoes used in my studies and their hosts are likely to vary over the geographic ranges of the mistletoe and alternate among different hosts. This may create multiple locally host–specific mistletoe populations and produce a complex geographic mosaic of mistletoe–host combinations across space and time. I suggest that mistletoe populations in South Africa may comprise numerous lineages incapable of parasitising the full range of host species, which could potentially lead to the formation of distinct host races over time. In the future, it would be interesting to document the infection patterns of these generalist mistletoe species across their entire geographic ranges in southern Africa, with particular focus on the patterns of mistletoe infection in places where the host abundance changes among sites. Host preferences may vary with changes in host frequency and host community composition. This could be paired with reciprocal transplant germination experiments in several sites to ascertain whether the mistletoe species have higher fitness on the most locally abundant hosts. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2013.