Thesis--University of California. / Bibliography: p. 328-334.

Spider wasps are solitary parasitoids that use one spider to lay a single egg. Even though their behavior seems homogeneous, the features pertaining to nesting and hunting behavior are diverse for different species. There are approximately 5,000 described species, in 120 genera, but there are probably many undescribed species. The systematics of Pompilidae has been studied in recent years, but only morphological data have been used for this purpose. Because of the morphological homogeneity of spider wasps, molecular data may prove promising for understanding the systematics of the group. Furthermore, dated molecular phylogenies calibrated with fossil data may allow studying the historical biogeography and evolution of the group. I used the nuclear molecular markers elongation factor–1 α F2 copy (EF1), long–wavelength rhodopsin (LWRh), wingless (Wg), RNA polymerase II (Pol2), the D2–D3 regions of the 28S ribosomal RNA (28S), and the mitochondrial Cytochrome C Oxidase I (COI) in a Bayesian and Maximum Likelihood framework, to reconstruct the phylogenies of four main Pompilidae groups: the subfamily Pompilinae, the tribe Aporini, the genus Psorthaspis, and the genus Drepanaporus. I also studied the fossil Pompilidae, and used those results to produce time-calibrated phylogenies of Pompilinae, Aporini, and Psorthaspis. Molecular phylogenetic results support the utility of the use of molecular markers for species delimitation and sex-associations in Pompilidae. In addition, the use of dated phylogenies supports the correlation of host use with diversification rate-shifts, the coevolution of mimicry between pompilids and velvet ants, and various biogeographical hypotheses never tested before for spider wasps.

A major area of investigation in evolutionary biology is the evolution of complex traits. The number of states, the order in which they arise, and the number of times a trait has evolved interest evolutionary biologists. Such studies are only made possible by reconstructing phylogenies in the context of the taxa. Biological investigations rely on accurate species designations and delimitations, and lack of well-defined taxonomic groups impedes scientific progress. Pompilidae (Insecta: Hymenoptera), popularly known as spider wasps, are predatory insects that provision their offspring with spiders as the sole food source. Adult female wasps attack spiders and paralyze them with venom, then place them in simple nests that are usually dug in the soil. Spider wasps form a large and cosmopolitan family with nearly 5,000 described species. Although all Pompilidae have similar biology, there is considerable variation in the nest construction and provisioning behavior; thus, this family could be useful for understanding the evolution of complex behavior. My study aims to evaluate and solve several taxonomic conflicts in spider wasps by reconstructing the complex evolution of behavioral patterns using a molecular phylogenetic framework. Early stages of sociality are found in spider wasps, such as communalism (females of same generation nesting together). My ultimate goal was to study the evolution of communalism in these wasps. I reconstructed relationships at the subfamily and tribal levels for the family, as well as generic and specific levels for pre-defined lineages using five nuclear markers (28S, EF, Pol2, LWRh, Wg), one mitochondrial marker (COI), and morphological characters. These studies comprise the first attempt to revise generic, tribal, and species delimitations for spider wasps, based on robust molecular evolutionary trees. Finally, by studying early stages of social evolution, my results will provide for a better understanding the evolution of social behavior in Hymenoptera as a whole.

Pollinators are believed to have played a key role in the radiation of flowering plants. The Grant-Stebbins model of pollinator-mediated speciation, in which evolutionary shifts between pollinators result in phenotypic diversification and enforce reproductive isolation, is one of the most compelling hypotheses for the rapid diversification of angiosperms. A key principle in this model is that plant pollination systems tend towards specialization, resulting in convergent suites of floral traits (syndromes) associated with particular types of pollinators. However, the expectation of pollination system specialization is not always supported by ecological data and has also been questioned on theoretical grounds. In this thesis, I examine pollination by Hemipepsis spider-hunting wasps (Hymenoptera, Pompilidae, Pepsinae) and use this system to address questions about levels and proximal mechanisms of floral specialization, floral shifts and convergent evolution of floral traits. Specialized pollination by Hemipepsis wasps is a newly described pollination system within the angiosperms. I document pollination by these wasps for the first time in 15 South African grassland plant species, including two species of Eucomis (Hyacinthaceae) and 13 asclepiads (Apocynaceae: Asclepiadoideae). In one of the asclepiads, Xysmalobium undulatum, I describe a bimodal pollination system involving both Hemipepsis wasps and a cetoniine beetle. I also describe an unusual and potentially antagonistic pollination mechanism whereby wasps are systematically dismembered during the insertion of pollinia in the two asclepiads Pachycarpus asperifolius and P. appendiculatus. I have used these and previous case studies to establish the existence of a new pollination guild, consisting of at least 21 plant species (across 10 genera and three families), that are reliant on four functionally similar species of Hemipepsis wasp for pollination. Plants in the guild are distributed throughout the moist grasslands of eastern South Africa and flower from September through until early May, peaking in December/January. The Hemipepsis-wasp pollination guild is characterized by high levels of functional specialization (17 of the 21 known guild members are pollinated exclusively by Hemipepsis wasps), despite the absence of morphological adaptations to prevent non-pollinating insects from accessing nectar. I used field and laboratory based experiments to explore the function of floral traits in enforcing specialization. These showed that Hemipepsis wasps primarily use scent, rather than visual cues, to locate flowers, but I was unable to firmly identify specific compounds responsible for the attraction of these wasps (compounds that elicited antennal responses in preliminary GC-EAD experiments did not attract wasps in bioassays). The chemical composition of the floral scents of guild members was examined for 71 individuals representing 14 species in addition to previous studies, and found to comprise complex blends of volatiles (usually containing between 30 and 50 compounds), typically dominated by aliphatics and monoterpenes with small amounts of aromatics. I also showed that the floral colours of guild members are similar to background vegetation, suggesting that floral colours are adapted for crypsis to avoid detection by non-pollinating insects. Palatability choice experiments with honeybees showed that non-pollinating insects find the nectars of at least three of the asclepiad guild members distasteful. Plants in this guild thus appear to achieve specialization through biochemical filters (scent as an attractant and differentially palatable nectar) and cryptic coloration. Pollinator-mediated convergence in floral traits is the fundamental basis for pollination syndromes, but has seldom been rigorously analyzed. Flowers in the Hemipepsis-wasp pollination guild share several qualitative traits, including dull greenish- or brownish-white colour, often with purple blotches, exposed sucrose dominant nectar with a relatively high sugar concentration (typically over 50% sugar by weight) and a sweet/spicy fragrance to the human nose. To test for convergent evolution in guild members, I compared scent, nectar and colour traits of guild members to those of congeners with different pollinators. Although traits often differed between guild members and their congeners, I found little evidence for overall convergence in floral scent profiles and nectar properties, but floral colours in the guild were significantly closer to the colour of background vegetation than those of congeners. At this stage, the lack of knowledge about specific floral volatiles that influence Hemipepsis-wasp behaviour and secondary nectar constituents that limit non-pollinator visits makes it difficult to identify the extent of biochemical convergent evolution within the guild. The directions and functional traits involved in evolutionary transitions between pollination by Hemipepsis wasps and other vectors are currently difficult to ascertain as there is limited phylogenetic data for the plant families concerned. In the genus Eucomis, fly and Hemipepsis-wasp pollinated species are very similar in floral morphology and colour, but differ strongly in floral scent. Using manipulative field experiments in conjunction with detailed analyses of colour, scent and morphology, I was able to show that a shift between wasp and fly pollination could be induced simply by manipulating oligosulphides in the scent emission from inflorescences. When considered in combination with other experiments highlighting the importance of scent as a pollinator attractant for all guild members, this suggests that scent properties may have played a key role in the evolutionary transitions between pollination by Hemipepsis wasps and other vectors. This research has established that pollination by Hemipepsis spider-hunting wasps is more geographically and phylogenetically widespread than was previously known, and has confirmed that these wasps are important and consistent pollinators in southern African grassland ecosystems. I have shown that a distinct guild of plants is specialized for pollination by these wasps. The high levels of specialization within this guild highlight the effectiveness of biochemical filters and cryptic coloration in limiting the spectrum of flower visitors. The major challenge ahead will be to identify the floral volatiles that attract Hemipepsis wasps and the non-sugar constituents that make the nectars of some guild members differentially palatable. These would both contribute greatly to our understanding of floral specialization and the mechanisms involved in the radiation of the angiosperms. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2011.