A thesis submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the academic requirements for the degree of Doctor of Philosophy School of Animal, Plant and Environmental Sciences. Johannesburg, June 2018. / Lantana camara L. (sensu lato) (Verbenaceae) remains one of the worst invasive alien plants in most tropical and subtropical parts of the world, including South Africa. Despite a concerted biological control (biocontrol) effort, with 45 biocontrol agents released against the weed worldwide since the early 1900s to date, L. camara control is far from satisfactory in most areas, including the study area. In 2012, during the initial stage of this work, a plant-ecological survey was conducted in riparian areas along the Sabie River, across an altitudinal gradient, and also in the adjacent forest plantation areas, in the province of Mpumalanga (South Africa). As a follow-up to two separate previous studies in the same area (1996/7 and 2005), aimed at determining the effectiveness of the ‘Working for Water’s (WfW) invasive alien plant (IAP) control programme, this work is another milestone in a long-term monitoring study. However, despite 16 years (1996/7-2012) of integrated IAP-control operations in the area, the WfW programme was only able to successfully remove larger overstorey IAPs, which opened-up the canopy and reduced competition, creating a conducive growing environment for an amalgamation of understorey IAPs, including L. camara, whose spread and densification were still on the rise. Biocontrol is regarded as a better alternative for long-term, sustainable and environmentally friendly IAP control, compared to the conventional mechanical and chemical methods. Most L. camara biocontrol agents introduced into South Africa have not yet had their full impact quantified under field conditions. This work is novel in that, for the first time, it quantifies the combined impact of the ‘old plus new’ suite of L. camara biocontrol agents, on the growth, reproduction and biomass of the weed under field conditions, in an inland area, through an insecticidal exclusion experiment, using carbofuran. Five prominent biocontrol agents occur on L. camara at the study sites, namely the fruit-mining fly, Ophiomyia lantanae (Froggatt) (Diptera: Agromyzidae); the shoot-sucking bug, Teleonemia scrupulosa Stål (Hemiptera: Tingidae); the defoliating moth, Hypena laceratalis Walker (Lepidoptera: Noctuidae); the leaf-mining beetle, Octotoma scabripennis Guèrin-Mèneville (Coleoptera: Chrysomelidae); and the fungal leaf-spot pathogen, cf. Passalora sp. (Chupp) U. Braun & Crous var. lantanae. During the course of this study, an additional agent, the flower-galling mite, Aceria lantanae (Cook) (Acari: Trombidiformes: Eriophyidae), was released and successfully established at lower altitudes (~843 m), showing an affinity for the dark-pink L. camara variety over others in the study area, namely light-pink and red-orange. Agent impact was difficult to measure because the activity of carbofuran in exclusion plants (carbofuran-treated L. camara plants) was short-lived; and therefore the impact of biocontrol agents on L. camara, which appeared to be negligible, may have been underestimated. Despite failing to maintain the ‘exclusion’ plants biocontrol agent-free through the application of carbofuran, there were reductions of 28% in the number of side-stems per plant, 31% fewer seeds in the soil seedbank, and 29% lower seed production, in ‘biocontrol’ plants compared to ‘exclusion’ plants. Although these differences were not statistically significant, they suggest that the present suite of biocontrol agents slightly reduces the vegetative and reproductive growth of L. camara. To achieve significant biocontrol of L. camara in inland areas, it seems necessary to introduce additional agents, which are well adapted to inland climatic conditions. The effects of micro-environmental factors, namely altitude and the degree of shading, were also investigated. Some biocontrol agents, such as T. scrupulosa, exhibited feeding phenological plasticity, resulting in it maintaining its presence at different altitudinal levels throughout the seasons. The performance of the suite of biocontrol agents, except A. lantanae, was, also, not limited by plant varietal differences. Additional research on biological and integrated control of L. camara is required. Keywords: Biocontrol; Biological invasion; Carbofuran; Insecticidal exclusion; Invasive alien plants; Lantana camara; Post-release evaluation. / LG2018

Lantana camara L. (Verbenaceae) remains a highly invasive and ecologically damaging weed in South Africa, despite some 50 years of biological control efforts. Lack of success has been ascribed to varietal differences, climate and predation of agents but these have not been tested. In this study, the effects of ant predation were tested on populations of three biological control agents for L. camara. Colonies of two species, Crematogaster sp. 1 and 2 were investigated. Crematogaster sp. 1 colonies were offered no choice between immature stages of the agents Hypena laceratalis Walker (Lepidoptera: Noctuidae), Falconia intermedia Distant (Hemiptera: Miridae) or Teleonemia scrupulosa Stål (Hemiptera: Tingidae) on lantana shoots. Density-dependent predation on F. intermedia and T. scrupulosa nymphs on lantana shoots was tested using Crematogaster sp. 2 colonies. In choice experiments Crematogaster sp. 2 colonies were offered F. intermedia or T. scrupulosa nymphs on potted lantana plants. Preliminary food trials confirmed that colonies foraged for protein, thereby validating results of no-choice experiments. Crematogaster sp.1 foragers removed 50% of F. intermedia nymphs, followed by 45% of H. laceratalis larvae and only 9% of T. scrupulosa nymphs. Foragers recruited most actively to H. laceratalis larvae and significantly more H. laceratalis biomass was removed than either F. intermedia or T. scrupulosa. A trade-off existed in prey size selection because larger larvae provided considerably more biomass but required forager cooperation and a longer time to subdue than did smaller prey. This increases both forager energy expense and mortality risk by other predators. This study showed that all Crematogaster sp. 1 colonies removed small (≤10mm) H. laceratalis larvae more frequently than larvae larger than 10mm. Thus, of these biological control agents, predators probably prefer small H. laceratalis larvae. Significantly more F. intermedia than T. scrupulosa nymphs were removed by Crematogaster sp. 1, while Crematogaster sp. 2 colonies removed comparable numbers of both agent species. Falconia intermedia nymphs' fast movement triggered a predatory response by these ant species. In contrast, the relatively immobile behaviour of T. scrupulosa nymphs was identified as a highly effective predator avoidance strategy. Since T. scrupulosa nymphs are unable to escape predators by moving, they appear to depend on the presence of alternative prey attracting predator attention. At high agent and/or forager density, T. scrupulosa nymphs attempted escape, but foragers identified them as prey once they moved and caught them. Predation on F. intermedia was also density dependent in that at high nymph and/or forager densities, escape routes were congested and nymphs were more easily caught. Survival of F. intermedia and T. scrupulosa nymphs in particular was low on ant-accessed shrubs in choice experiments and high on ant-excluded shrubs. It is likely that ants significantly depress F. intermedia populations in the field since besides predation, ant foragers probably interrupt F. intermedia feeding and ovipositioning. The combination of parasitism and predation on early instar larvae may explain why H. laceratalis occurs across lantana's range in South Africa but populations remain low. It is unlikely that T. scrupulosa nymphs are habitually preyed on by ant species unless they attract attention by being mobile. Although biological control of L. camara is influenced by climate and physiological defence mechanisms, this study has shown that predation by two ant species severely impacts leaf-feeding agents for L. camara. Thus, it is recommended that future selection of additional agents to control lantana should exclude leaf-feeding .

The exotic plant Lantana camara L. has invaded the riparian areas of the Victoria Falls World Heritage Site in Livingstone, southern Zambia, threatening native plant communities which support populations of species of special concern. I trialled the mechanical control method of manual uprooting and 3 different herbicides applied through paint brushing of an imazapyr concentrate at 250g. l¯¹, spraying on cut stumps with metsulfron methyl at 600g.l¯¹, and foliar spraying on re-emergent lantana foliage with glyphosate at a dosage of 166g. l¯¹ in July 2008 in 20 100m2 treatment plots, 5 invaded control plots and 5 uninvaded controls. Follow-up treatments for re-sprouting lantana stumps and emerging seedlings were undertaken in June 2009. I measured effectiveness of the methods using adult lantana mortality in June 2009 and lantana seedling density in the different treatment plots during the follow-up exercise. The cost of the various methods and human labour applied were compared across the four treatments at initial clear and at follow-up. All treatments recorded a high adult lantana mortality rate, though there were no significant differences in lantana adult mortality amongst the treatments. Overall, uprooting had the highest adult mortality, followed by imazapyr, metsulfron and lastly glyphosate. Germination of lantana seedlings after clearing was high for all treatments but with no significant differences occurring between the treatments. Both adult lantana mortality and seedling density were however significantly different from the control. With labour included, chemical costs were far higher relative to uprooting, though uprooting costs were the highest when it came to the follow-up because of the emerging seedlings and some resprouting stumps. The effects of mechanical and chemical treatments on vegetation composition in the cleared areas were also assessed in order to detect any non-target and medium term effects of treatments. Contrary to expectation, none of the chemicals showed any significant effects on vegetation composition in the short and medium-term and no significant differences were found in plant species richness, diversity and seedling density between invaded and uninvaded plots at baseline, in October 2008 and in September 2009. In order to determine potential for unaided vegetation recovery in the riparian areas of the study site after lantana clearing, I conducted an investigation of soil seed banks and seed rain using 60 seed bank samples measuring 1800m³ collected from 30 invaded and uninvaded plots. Using the seedling emergence method, 1, 991 seedlings belonging to 66 species representing 27 families germinated from the seed bank. Sedges (Cyperaceae family) were the most abundant taxa in the seed banks from invaded areas, followed by Ageratum conyzoides, lantana, Triumfetta annua and Achyranthes aspera which also occurred in the uninvaded soil seed banks. The seed banks from uninvaded plots were dominated by the grass Oplismenus hirtellus. Overall, species richness, diversity and seedling density from seed banks in invaded areas did not differ significantly from seed bank in uninvaded areas and there was a low similarity in species composition when above ground vegetation was compared to seed banks from invaded and uninvaded areas. It would appear if natural regeneration occured from the current seed bank in disturbed areas, future vegetation would largely comprise of short lived, early successional species in the short term as the seed bank is dominated by non-native herbaceous weedy species. From the seed traps investigating seed rain, a total of 27 species numbering 623 individual seeds were found in the thirty 1m² seedtraps distributed in invaded and uninvaded areas at the five sites, over an intermittent period of three months. Lantana had the highest monthly arrival rate in the seed traps followed by Phoenix reclinata and Ricinus communis. The number of species with invasive potential found in the seed traps located in invaded areas was more than that found in seed traps under native vegetation cover by far. Considerable forest remnants still occur around the invaded sites, and these could serve as an important source for long-term natural re-establishment of native vegetation if seed availability by animals and wind dispersal continues, while the re-invasion of lantana is prevented by ongoing follow-ups and futher clearing of lantana invaded areas. It is concluded that while uprooting and other treatments are effective in the control of lantana, its successful control in the Victoria Falls World Heritage Site will require extensive clearing to keep it from reinvading infested areas after clearing as shown by the seed rain data. The high seedling density of lantana in the seed banks and in the cleared areas shows the need for ongoing follow-up in order to deplete soil stored seed banks. There is need for longer term research to establish what the exact follow-up requirements are in order to contain lantana re-infestation and create favourable micro-sites for native species to establish. It is predicted that ongoing lantana control in the cleared plots will most likely initiate long-term community recovery.

A highly variable invasive shrub, Lantana camara L. (Verbenaceae), has been notoriously difficult to control thus far despite a well established biological control programme in South Africa. A promising leaf-feeding biological control agent, Falconia intermedia (Distant) (Hemiptera: Miridae), released to control this invasive plant eventually crashed at three out of five sites in the Eastern Cape Province. In the Mpumalanga Province, after initially colonising and building up high numbers on the L. camara stands the agent populations crashed. Several reasons for these population crashes have been suggested, but induced plant defences have not been investigated. Although plants face the challenge of herbivory by various organisms while remaining immobile, some plants may possess the ability to induce physical and/or chemical defensive responses following feeding and thus prevent further plant tissue damage and loss. Laboratory trials were conducted to determine the existence, nature and effect of physical and chemical feeding-induced responses of L. camara on the performance of the leaf-feeding biological control agent, F. intermedia. Lantana camara plants used in the study were obtained from five localities in the Eastern Cape Province, South Africa, while the insect culture was established from field populations. Plants from all varieties on which F. intermedia was released significantly increased the toughness of their leaves compared to control treatment plants. In addition, plants from three localities: Lyndhurst Farm, East London and Port Alfred, significantly increased trichome density after prolonged feeding by F. intermedia. On the three varieties showing increases in these two factors (i.e. leaf toughness and trichome density), oviposition, survival and feeding damage by the mirid agent was significantly lower on previously damaged plants. A significant negative correlation between trichome density and population numbers was found (R²= 0.52, p < 0.0003), suggesting that an increase in trichome density strongly contributes to a reduction in F. intermedia's growth. The growth and reproduction of the resistant plants was not significantly impacted by F. intermedia feeding. The defensive responses were found to be plant systemic and rapidly induced as they were elicited and expressed throughout the plant in both damaged and undamaged leaves within five weeks after insect release. Leaf toughness and trichome density were not significantly increased after feeding on plants from Whitney Farm and Heather Glen Farm. On the contrary, mirid individuals performed significantly better on plants from Whitney Farm and Heather Glen Farm than on plants of other varieties, indicating their susceptibility and suitability to the agent and the lack of induced resistance against the agent. Plants from all localities besides East London showed some level of tolerance and overcompensated for feeding damage by increasing plant growth and reproductive factors on plants fed upon. This was however only significant in two variables of the more susceptible localities, Whitney Farm and Heather Glen Farm. This increase in plant fitness did however indicate an induced defence response by these plants to feeding, a response designed to lessen the effects of agent feeding. Headspace volatile analysis was used to investigate any volatile chemical responses by L. camara due to F. intermedia feeding at two of the five localities chosen: East London and Whitney Farm. There was no significant difference in headspace volatiles emitted by leaves of plants from the East London insect infested and control treatment plants. On the Whitney Farm damaged plants however there was a 2.5 fold increase in the emission intensity of one of the three main compounds, later identified as Beta-caryophyllene. Three major chemical constituents which were found to be common to leaf volatiles of the two varieties were identified through gas chromatography-mass spectrometry (GC-MS) from the damaged and undamaged leaves of these two varieties. The methods used in collecting leaf volatiles were shown to be significant in the strength of chromatogram peaks. Using general authentication methods and purified standards, one of these was identified as the sesquiterpene, Beta-caryophyllene (C₁₅H₂₄). This compound is one of the major constituents found in isolations of L. camara varieties worldwide. This is the first such work done on a variety of L. camara in South Africa, and hopefully the beginning of more in-depth studies of the volatile organic chemicals from the numerous naturalised varieties of L. camara. It is suggested that the sum of these responses may play a role bigger than is currently understood in this plant-insect relationship. It is also argued that feeding induced plant defences may play an important role in attempts to control alien plants using insect agents.