The flora of New Zealand has evolved largely in the absence of terrestrial mammals, the predominant frugivore guilds being birds and reptiles. The evolution of divergent fruit traits in New Zealand may be a consequence of different selection pressure by these two guilds, and two contrasting putative dispersal syndromes have been proposed for New Zealand fleshy fruited plants. Coprosma (Rubiaceae: subtribe Coprosminae) is a speciose and morphologically diverse genus, which is distributed throughout the South Pacific and which also produces variably coloured drupes. It was selected as a model genus to investigate the evolution of fruit traits within this context. For this purpose, a molecular phylogeny for the subtribe Coprosminae and Coprosma was inferred using parsimony, likelihood and split decomposition analysis on sequences from the 16rps intron of cpDNA and ETS and ITS region of nrDNA. Up to 32 species were included in the subtribal analyses, whilst 53 species of Coprosma were used in subgeneric studies. The basis for the variety of fruit colours seen in New Zealand was determined using histology and pigment extractions. To test the assumption that fruit traits among species evolved in concert under selection from frugivore guilds, fruit shape, size and nutrient composition were determined in order to test predictions that these would co-vary with fruit colour. In the Coprosminae, fleshy drupes have arisen from dry fruited schizocarps and with two possible reversals to semi-dry drupes. Within Coprosma, fruit colour was found to be evolutionarily labile and varied both among and within lineages, particularly within two main groups where fruit colour had shifted from orange to blue and white, or red, pink and black colours respectively. The evolution of novel (non-orange) fruit colours was restricted to New Zealand, as was the small-leaved, divaricate growth form, the combination of which has been associated with a putative reptile dispersal syndrome. Several trans-oceanic dispersals out of New Zealand were also inferred from the phylogeny, these extending into Australia, New Guinea and Hawai�i. In these instances, fruit colour did not appear change markedly after establishment in a new country. Within New Zealand, fruit sizes were found to vary with colour as predicted, although the majority of species produced small (< 8 mm), elliptical fruits. There was no clear association between fruit colour and fruit nutrient composition. It would appear that these small, succulent, carbohydrate-rich and lipid-poor fruits cater to a wide range of generalist frugivores. The variety of fruit colours in Coprosma stemmed from qualitative and quantitative differences in carotenoid and anthocyanin composition. The genetic basis for the control of these pathways is currently unknown, but it is assumed that a few regulatory genes can control a substantial amount of phenotypic variation. Considering the evolutionary history of Coprosma, it would appear that a history of recent and rapid speciation, hybridisation and reticulate evolution may have increased the tempo of fruit colour evolution in the genus.

The flora of New Zealand has evolved largely in the absence of terrestrial mammals, the predominant frugivore guilds being birds and reptiles. The evolution of divergent fruit traits in New Zealand may be a consequence of different selection pressure by these two guilds, and two contrasting putative dispersal syndromes have been proposed for New Zealand fleshy fruited plants. Coprosma (Rubiaceae: subtribe Coprosminae) is a speciose and morphologically diverse genus, which is distributed throughout the South Pacific and which also produces variably coloured drupes. It was selected as a model genus to investigate the evolution of fruit traits within this context. For this purpose, a molecular phylogeny for the subtribe Coprosminae and Coprosma was inferred using parsimony, likelihood and split decomposition analysis on sequences from the 16rps intron of cpDNA and ETS and ITS region of nrDNA. Up to 32 species were included in the subtribal analyses, whilst 53 species of Coprosma were used in subgeneric studies. The basis for the variety of fruit colours seen in New Zealand was determined using histology and pigment extractions. To test the assumption that fruit traits among species evolved in concert under selection from frugivore guilds, fruit shape, size and nutrient composition were determined in order to test predictions that these would co-vary with fruit colour. In the Coprosminae, fleshy drupes have arisen from dry fruited schizocarps and with two possible reversals to semi-dry drupes. Within Coprosma, fruit colour was found to be evolutionarily labile and varied both among and within lineages, particularly within two main groups where fruit colour had shifted from orange to blue and white, or red, pink and black colours respectively. The evolution of novel (non-orange) fruit colours was restricted to New Zealand, as was the small-leaved, divaricate growth form, the combination of which has been associated with a putative reptile dispersal syndrome. Several trans-oceanic dispersals out of New Zealand were also inferred from the phylogeny, these extending into Australia, New Guinea and Hawai�i. In these instances, fruit colour did not appear change markedly after establishment in a new country. Within New Zealand, fruit sizes were found to vary with colour as predicted, although the majority of species produced small (< 8 mm), elliptical fruits. There was no clear association between fruit colour and fruit nutrient composition. It would appear that these small, succulent, carbohydrate-rich and lipid-poor fruits cater to a wide range of generalist frugivores. The variety of fruit colours in Coprosma stemmed from qualitative and quantitative differences in carotenoid and anthocyanin composition. The genetic basis for the control of these pathways is currently unknown, but it is assumed that a few regulatory genes can control a substantial amount of phenotypic variation. Considering the evolutionary history of Coprosma, it would appear that a history of recent and rapid speciation, hybridisation and reticulate evolution may have increased the tempo of fruit colour evolution in the genus.

The vegetation of the island of Rangitoto was examined in order to determine the current vegetation pattern and to identify the factors which have influenced the development of this pattern. Information about the order and dates of eruptive events was reviewed to gauge the length of time that the various surfaces of the island have been available as a substrate for the development of a vegetation covering. Available geological information, dating evidence, historical accounts, tree ages, and Maori evidence all point to a single period of eruptive activity, probably only several years in duration, around the mid- to late-1300's (A.D.). The order of eruption was probably production of the ash that covers neighbouring islands, followed by cinder cone building, followed by eruption of the lava flows. The lava flows were found to consist of clinkery aa flows, blocky aa flows, and a flow type referred to as Rangitoto slab flows. These flow types could be partially distinguished from one another, using discriminant analysis, on the basis of the length of the longest fragment on a 5x5m plot and the number of fragments on a lxlm subplot. A new transition sequence of flow types from pahoehoe to aa lavas is proposed for viscous lavas undergoing low rates of shear strain. The vegetation pattern of the lava fields was examine using TWINSPAN and CANOCO analysis of foliage cover percentage information gathered from 125 5x5m plots. It was found to consist of a successional sequence of vegetation arranged in a mosaic. The mosaic was found to relate strictly to the underlying lava flow surface. Large Metrosideros grow in crevices on slab flows and next to large boulders on aa flows. Smaller Metrosideros grow on small slabs on slab flows. Mixed species scrub is found growing on unbroken slab surfaces. The relative rate of colonisation of different types of flows under different climatic conditions was considered. The Metrosideros hybrid swarm was studied using morphometric information. The putative parent species of the swarm were found, using Principal Components Analysis, to be Metrosideros excelsa and Metrosideros robusta. The Rangitoto Metrosideros population was found, using Canonical Variates Analysis, to be the result of hybridisation, followed by introgression towards M. excelsa . The probable F1 hybrids grow in early successional stage vegetation. The major geographical trend is the tendency for backcrossed hybrids to grow on the eastern side of the island, suggesting eastern origins of M. robusta seeds. The impact of browsing animals on the vegetation was studied over five years in exclosures and corresponding control sites. Metrosideros foliage recovered slightly. Griselinia lucida and Cyathodes juniperina seedlings were recruited into the shrub layers in exclosures, but not in control sites. Astelia seedlings also benefitted from the absence of browsing pressure, as did Thelymitra longifolia. The distribution of browsing animal populations in relation to the vegetation pattern was studied using faecal pellet recruitment data gathered by the New Zealand Forest Service in 1984. Both wallabies (Petrogale penicillata penicillata) and possums (Trichosurus vulpecula) were found to be distributed principally according to the amount of palatable foliage available to each species in each vegetation type. / Appendices restricted at the request of the author

The vegetation of the island of Rangitoto was examined in order to determine the current vegetation pattern and to identify the factors which have influenced the development of this pattern. Information about the order and dates of eruptive events was reviewed to gauge the length of time that the various surfaces of the island have been available as a substrate for the development of a vegetation covering. Available geological information, dating evidence, historical accounts, tree ages, and Maori evidence all point to a single period of eruptive activity, probably only several years in duration, around the mid- to late-1300's (A.D.). The order of eruption was probably production of the ash that covers neighbouring islands, followed by cinder cone building, followed by eruption of the lava flows. The lava flows were found to consist of clinkery aa flows, blocky aa flows, and a flow type referred to as Rangitoto slab flows. These flow types could be partially distinguished from one another, using discriminant analysis, on the basis of the length of the longest fragment on a 5x5m plot and the number of fragments on a lxlm subplot. A new transition sequence of flow types from pahoehoe to aa lavas is proposed for viscous lavas undergoing low rates of shear strain. The vegetation pattern of the lava fields was examine using TWINSPAN and CANOCO analysis of foliage cover percentage information gathered from 125 5x5m plots. It was found to consist of a successional sequence of vegetation arranged in a mosaic. The mosaic was found to relate strictly to the underlying lava flow surface. Large Metrosideros grow in crevices on slab flows and next to large boulders on aa flows. Smaller Metrosideros grow on small slabs on slab flows. Mixed species scrub is found growing on unbroken slab surfaces. The relative rate of colonisation of different types of flows under different climatic conditions was considered. The Metrosideros hybrid swarm was studied using morphometric information. The putative parent species of the swarm were found, using Principal Components Analysis, to be Metrosideros excelsa and Metrosideros robusta. The Rangitoto Metrosideros population was found, using Canonical Variates Analysis, to be the result of hybridisation, followed by introgression towards M. excelsa . The probable F1 hybrids grow in early successional stage vegetation. The major geographical trend is the tendency for backcrossed hybrids to grow on the eastern side of the island, suggesting eastern origins of M. robusta seeds. The impact of browsing animals on the vegetation was studied over five years in exclosures and corresponding control sites. Metrosideros foliage recovered slightly. Griselinia lucida and Cyathodes juniperina seedlings were recruited into the shrub layers in exclosures, but not in control sites. Astelia seedlings also benefitted from the absence of browsing pressure, as did Thelymitra longifolia. The distribution of browsing animal populations in relation to the vegetation pattern was studied using faecal pellet recruitment data gathered by the New Zealand Forest Service in 1984. Both wallabies (Petrogale penicillata penicillata) and possums (Trichosurus vulpecula) were found to be distributed principally according to the amount of palatable foliage available to each species in each vegetation type. / Appendices restricted at the request of the author

The vegetation of the island of Rangitoto was examined in order to determine the current vegetation pattern and to identify the factors which have influenced the development of this pattern. Information about the order and dates of eruptive events was reviewed to gauge the length of time that the various surfaces of the island have been available as a substrate for the development of a vegetation covering. Available geological information, dating evidence, historical accounts, tree ages, and Maori evidence all point to a single period of eruptive activity, probably only several years in duration, around the mid- to late-1300's (A.D.). The order of eruption was probably production of the ash that covers neighbouring islands, followed by cinder cone building, followed by eruption of the lava flows. The lava flows were found to consist of clinkery aa flows, blocky aa flows, and a flow type referred to as Rangitoto slab flows. These flow types could be partially distinguished from one another, using discriminant analysis, on the basis of the length of the longest fragment on a 5x5m plot and the number of fragments on a lxlm subplot. A new transition sequence of flow types from pahoehoe to aa lavas is proposed for viscous lavas undergoing low rates of shear strain. The vegetation pattern of the lava fields was examine using TWINSPAN and CANOCO analysis of foliage cover percentage information gathered from 125 5x5m plots. It was found to consist of a successional sequence of vegetation arranged in a mosaic. The mosaic was found to relate strictly to the underlying lava flow surface. Large Metrosideros grow in crevices on slab flows and next to large boulders on aa flows. Smaller Metrosideros grow on small slabs on slab flows. Mixed species scrub is found growing on unbroken slab surfaces. The relative rate of colonisation of different types of flows under different climatic conditions was considered. The Metrosideros hybrid swarm was studied using morphometric information. The putative parent species of the swarm were found, using Principal Components Analysis, to be Metrosideros excelsa and Metrosideros robusta. The Rangitoto Metrosideros population was found, using Canonical Variates Analysis, to be the result of hybridisation, followed by introgression towards M. excelsa . The probable F1 hybrids grow in early successional stage vegetation. The major geographical trend is the tendency for backcrossed hybrids to grow on the eastern side of the island, suggesting eastern origins of M. robusta seeds. The impact of browsing animals on the vegetation was studied over five years in exclosures and corresponding control sites. Metrosideros foliage recovered slightly. Griselinia lucida and Cyathodes juniperina seedlings were recruited into the shrub layers in exclosures, but not in control sites. Astelia seedlings also benefitted from the absence of browsing pressure, as did Thelymitra longifolia. The distribution of browsing animal populations in relation to the vegetation pattern was studied using faecal pellet recruitment data gathered by the New Zealand Forest Service in 1984. Both wallabies (Petrogale penicillata penicillata) and possums (Trichosurus vulpecula) were found to be distributed principally according to the amount of palatable foliage available to each species in each vegetation type. / Appendices restricted at the request of the author

The vegetation of the island of Rangitoto was examined in order to determine the current vegetation pattern and to identify the factors which have influenced the development of this pattern. Information about the order and dates of eruptive events was reviewed to gauge the length of time that the various surfaces of the island have been available as a substrate for the development of a vegetation covering. Available geological information, dating evidence, historical accounts, tree ages, and Maori evidence all point to a single period of eruptive activity, probably only several years in duration, around the mid- to late-1300's (A.D.). The order of eruption was probably production of the ash that covers neighbouring islands, followed by cinder cone building, followed by eruption of the lava flows. The lava flows were found to consist of clinkery aa flows, blocky aa flows, and a flow type referred to as Rangitoto slab flows. These flow types could be partially distinguished from one another, using discriminant analysis, on the basis of the length of the longest fragment on a 5x5m plot and the number of fragments on a lxlm subplot. A new transition sequence of flow types from pahoehoe to aa lavas is proposed for viscous lavas undergoing low rates of shear strain. The vegetation pattern of the lava fields was examine using TWINSPAN and CANOCO analysis of foliage cover percentage information gathered from 125 5x5m plots. It was found to consist of a successional sequence of vegetation arranged in a mosaic. The mosaic was found to relate strictly to the underlying lava flow surface. Large Metrosideros grow in crevices on slab flows and next to large boulders on aa flows. Smaller Metrosideros grow on small slabs on slab flows. Mixed species scrub is found growing on unbroken slab surfaces. The relative rate of colonisation of different types of flows under different climatic conditions was considered. The Metrosideros hybrid swarm was studied using morphometric information. The putative parent species of the swarm were found, using Principal Components Analysis, to be Metrosideros excelsa and Metrosideros robusta. The Rangitoto Metrosideros population was found, using Canonical Variates Analysis, to be the result of hybridisation, followed by introgression towards M. excelsa . The probable F1 hybrids grow in early successional stage vegetation. The major geographical trend is the tendency for backcrossed hybrids to grow on the eastern side of the island, suggesting eastern origins of M. robusta seeds. The impact of browsing animals on the vegetation was studied over five years in exclosures and corresponding control sites. Metrosideros foliage recovered slightly. Griselinia lucida and Cyathodes juniperina seedlings were recruited into the shrub layers in exclosures, but not in control sites. Astelia seedlings also benefitted from the absence of browsing pressure, as did Thelymitra longifolia. The distribution of browsing animal populations in relation to the vegetation pattern was studied using faecal pellet recruitment data gathered by the New Zealand Forest Service in 1984. Both wallabies (Petrogale penicillata penicillata) and possums (Trichosurus vulpecula) were found to be distributed principally according to the amount of palatable foliage available to each species in each vegetation type. / Appendices restricted at the request of the author

My dissertation research focused on the behavior, movement patterns, and foraging ecology of dusky dolphins (Lagenorhynchus obscurus) off the east coast of New Zealand's South Island. Information on growth, morphometrics, parasites and life history was also collected. Movement patterns and foraging behavior of New Zealand dusky dolphins were much different from those of dusky dolphins observed off the Argentine coast by Bernd Wursig, the only other study of dusky dolphin behavior. Unlike the Argentine dolphins, which cooperatively herd anchovy to the surface and contain them there for feeding, New Zealand dusky dolphins behave and forage more like Hawaiian spinner dolphins (Stenella longirostris). In summer and fall, New Zealand dusky dolphins remain near shore in morning and early afternoon hours, then move into deeper water with greatly increased activity levels in late afternoon. In winter and spring they remain farther from shore at all times of day, are found in somewhat larger groups, and appear to travel along shore more often than in other seasons. In summer, dive times of radio-tagged dolphins also varied on a daily cycle, most long dives occurring during crepuscular and night periods. Stomach contents of incidentally-netted and beachcast dolphins contained primarily a demersal fish and a few types of mesopelagic fishes and squid. Acoustic surveys along the east coast of South Island show a dense layer of mesopelagic fishes and squid that move to within 50-100 m of the surface at night. Like Hawaiian spinners, New Zealand dusky dolphins feed primarily on prey in and associated with the vertically migrating layer, probably as a means of increasing foraging efficiency. External measurements of L. obscurus specimens were analyzed using canonical variate analysis, which revealed measurements useful in discrimination of age/sex classes, including dorsal fin dimensions, and positioning of dorsal fin and flipper insertions. Tooth-section age analysis of specimens allowed construction of growth curves; life span maximum was about 35 years. The very large size of active testes (over 1 kg each) during summer breeding represents a large proportion of total body weight. Along with observations of group composition variability, this suggests a promiscuous mating system and a fluid, extended-group social system.

New Zealand forests have been reduced and degraded by gross removal, logging, and the effects of mammals introduced by Polynesian and European settlers. These changes increase the value of the remaining forests, so information on the effects of these disturbances will be useful to inform the management of forest protection. Integrated measurements of C and N cycling within forests can be obtained using foliar stable isotope ratios, which may detect differences between forests resulting from natural or anthropogenic disturbances. This thesis characterises the stable isotopic composition distribution and likely drivers of isotopic variation of vegetation in several central South Island forests, and provides a baseline for future ecological New Zealand studies of present and pre-human vegetation. The largest detected stable isotope variation in modern leaf material was that of δ15N values between the eastern and western podocarp-broadleaf forests. This variation was probably controlled by the lower soil N availability associated with the high rainfall of western forests causing low δ15N values (-8.5 ± 3.5 ‰) relative to an eastern forest (+1.6 ± 1.3 ‰) and global temperate forests (average -2.8 ± 2.0 ‰ (Martinelli et al. 1999)). The significant but slightly higher mean δ15N (0.6 ‰) of a historically selectively logged forest (Saltwater Forest) in comparison to the mean in an unlogged forest (Okarito Forest), on the West Coast, could be attributed to either alteration to N cycling from logging, site differences in topography, or local soil N differences between the forests. Although δ13C showed no significant geographical variation, the well-described ‘canopy effect’ was observed in all modern forests, manifested as a positive covariation between δ13C and vegetation height. Similarly, large taxon-specific differences were observed between δ15N and δ13C values in both modern and fossil leaves. Well-preserved fossil leaves, from sediments c. 4500 years B.P in Pyramid Valley, North Canterbury, had higher δ13C (4.2 ‰) and δ15N (2.5 ‰) values than modern vegetation from Riccarton Bush, Christchurch. The difference between ecosystems spanning several millennia probably reflects ecosystem-scale changes in C and N cycling within New Zealand forests following human arrival, particularly from the degradation caused by invasive animals.