Westland petrels Procellaria westlandica breed only near Punakaiki on the West Coast of New Zealand. About 80 km offshore from their breeding colony, New Zealand's largest commercial fishery (for hoki Macruronus novaezelandiae) operates from mid June to early September, coinciding with the Westland petrel's breeding season. It has been assumed that Westland petrels feed extensively on fisheries waste and that this habit has been at least partly responsible for the increase in the Westland petrel population. Some seabird biologists have expressed concern that if a species comes to depend on scavenging at fishing vessels, such a species could experience a food crisis if fishing operations changed in a way that reduced the quantity of waste discharged. The aim of this research was to assess how dependent Westland petrels have become on fisheries waste for food. Diet studies showed that during the hoki fishing season, waste accounts for more than half by weight of the solid food Westland petrels bring back to the colony to feed their chicks. After the hoki season, waste contributes only about a quarter of their diet as birds switch to more natural prey and scavenge a wider variety of fish species presumably from smaller, inshore fishing vessels. Much of the fisheries waste eaten by Westland petrels was flesh which could not be identified using traditional techniques. The electrophoretic technique iso-electric focusing increased the number of fish samples that could be identified and consequently the diet was interpreted differently than it would have been had only traditional diet analysis been used. The survey of Westland petrel distribution off the west coast of the South Island, found that although hoki fishing vessels influence the distribution of Westland petrels, only a small proportion of the Westland petrel population appears to utilise this food resource at any one time. Westland petrels were tracked at sea by VHF radio telemetry and then by satellite tracking. Satellite tracking showed that there is considerable variation in the amount of time Westland petrels spend in the vicinity of fishing vessels. On average, satellite tracked birds spent one third of their time near vessels, but they foraged over much larger areas than that occupied by the West Coast South Island hoki fishing fleet. Although fisheries waste is an important component of the Westland petrel diet, it appears that the situation is one of opportunistic use of a readily available resource, rather than one of dependence. Several features of the Westland petrel's breeding biology and foraging ecology suggest that Westland petrels could compensate for a reduction in waste from the hoki fishery by switching to other sources of waste and increasing their consumption of natural prey. Nevertheless, much remains unanswered concerning the role of fisheries waste in the Westland petrel's diet. In particular, quantifying the waste available to seabirds, and the success of Westland petrels in acquiring that waste compared to other scavenging species, is needed in order to better predict the effect of a reduction in fisheries waste on Westland petrel population size.

This study investigated understorey management in Canterbury, New Zealand, apple orchards for the enhancement of populations of Dolichogenidea tasmanica (Cameron) (Braconidae) for leafroller (Lepidoptera: Tortricidae) biological control. The first objective was to determine the influence of understorey plants on the abundance of D. tasmanica and leafroller parasitism, and to investigate the mechanisms behind this influence. The second was to determine the most suitable understorey plants in terms of their ability to enhance parasitoid abundance, leafroller parasitism, parasitoid longevity, parasitoid fecundity and its ability to not benefit leafroller. Results from three consecutive field trials showed that buckwheat (Fagopyrum esculentum Moench), coriander (Coriandrum sativum L.), alyssum (Lobularia maritima (L.) Desv), and, to a lesser extent, broad bean (Vicia faba L.), enhanced parasitoid abundance and leafroller parasitism. The mechanisms behind the effects of understorey plants had previously been unexplored. However, results here showed that it was the flowers or the buckwheat that 'attracted' the parasitoid to the plant and not the shelter, aphids or microclimate that the plant may also provide. Providing flowering plants in the orchard understorey also increased immigration of parasitoids and enhanced parasitoids and enhanced parasitoid longevity and fecundity in the laboratory. In contrast, the understorey plants had no influence on the female:male ratio of D. tasmanica. Although coriander enhanced leafroller parasitism three-fold in field experiments compared with controls, it failed to enhance the longevity of both sexes of D. tasmanica in the laboratory compared with water-only. Broad bean significantly enhanced parasitoid abundance three-fold and significantly increased parasitism from 0% to 75% compared with the controls on one leafroller release date. However, laboratory trials showed that of male D. tasmancia but it did not enhance female longevity. Also, female D. tasmanica foraging on broad bean produced a total of only three parasitoid cocoons, but this result was based on an overall 6.5% survival of larvae to pupae or to parasitoid cocoon. Furthermore, results suggested that extrafloral nectar secretion decreased as the plants matured. Phacelia (Phacelia tanacetifolia Benth.) did not significantly enhance parasitism rate in the field compared with controls, and numbers of D. tasmanica captured by suction sampling were significantly lower in phacelia treatments compared with alyssum, buckwheat and control plots. Also, laboratory experiments showed that survival of D. tasmanica on phacelia flowers was equivalent to that on water-only and significantly lower than on buckwheat. These results suggest that phacelia does not provide nectar to D. tasmanica, only pollen, and therefore is not a suitable understorey plant for D. tasmanica enhancement in orchards. Buckwheat and alyssum showed the most potential as understorey plants for the enhancement of natural enemies. Buckwheat not only increased numbers of D. tasmanica seven-fold, but also increased numbers of beneficial lacewings (Micromus tasmaniae (Walker)) and hover flies (Syrphidae) captured on yellow sticky traps compared with the controls. It significantly increased leafroller parasitism by D. tasmanica from 0% to 86% compared with the controls (on one date only), and in the laboratory enhanced D. tasmanica longevity and increased fecundity compared with water-only. Similarly, alyssum significantly increased parasitism rate compared with controls, and two-fold more D. tasmanica were suction sampled in these plots compared with controls. It also enhanced longevity of both sexes of D. tasmanica compared with water, and showed the most favourable characteristics in terms of being of no benefit to leafrollers. This is because it was not preferred over apple by leafroller larvae and when they were forced to feed on it, it caused high mortality (94.3%) and low pupal weight (15 mg). Furthermore, alyssum did not enhance the number of fertile eggs produced by adult leafrollers compared with water only. However, further research is required to address the overall effect of buckwheat and alyssum on crop production and orchard management, including effects on fruit yield and quality, frost risk, disease incidence, soil quality, weeds and other pests. Also, research into the ability of these plants to survive in the orchard with little maintenance, and into the optimal sowing rates, would be useful. Sampling natural populations of leafroller within each treatment showed that damage from leafrollers and the number of leafroller larvae were respectively 20.3% and 29.3% lower in the flowering treatments compared with the controls. Furthermore, field trials showed up to a six-fold increase in leafroller pupae in controls compared with buckwheat and alyssum. This suggests that increasing leafroller parasitism rate from understorey management in orchards will translate into lower pest populations, although neither larval numbers/damage nor pupal numbers differed significantly between treatments. Trapping D. tasmanica at a gradient of distances showed that this parasitoid travels into rows adjacent to buckwheat plots, indicating that growers may be able to sow flowering plants in every second or third row of the orchard, and still enhance leafroller biocontrol while minimising the adverse effects of a cover crop. Sowing buckwheat and alyssum in orchard understoreys may enhance biological control of apple pests in organic apple production and reduce the number of insect growth regulators applied in IFP programmes. However, the challenge still remains to investigate whether conservation biological control can reduce leafroller populations below economic thresholds.

Among the species of Diomedea albatrosses, diverse foraging strategies during breeding have been described, indicating species differences in foraging ecology and behaviour. Foraging strategies of Southern Royal Albatrosses, Diomedea epomophora (SRA) breeding on Campbell Island were studied in January – early February 1999 during the latter half of incubation. Movements and activity of ten birds were monitored using satellite transmitters and wet-dry activity recorders. Three birds from a pilot tracking study in February 1997 were also included in some analyses. Foraging strategies, zones used, factors influencing the duration of foraging trips, and the influence of wind conditions were investigated. Foraging activity took place at sites with bathymetric characteristics associated with high productivity: outer shelf and shelf-break zones, with a concentration of activity on a shelf contour south of the Snares Islands. This is in contrast to Wandering (D. exulans) and Gibson’s (D. gibsoni) albatrosses, typically deep oceanic foragers, but is similar to Northern Royal Albatross (D. sanfordi). The maximum distance of foraging trips from the colony was 1250 kilometres (mean 584 +351(SD)). This was closer than for incubating Wandering and Gibson’s Albatrosses but more distant than for Northern Royal Albatross from the Otago Peninsula. The mean duration of 77 foraging trips from 52 nests was 10.11 days for females and 8.76 for males (ns). Foraging trips became shorter as incubation progressed. Foraging trips were shorter, but not significantly so, when the median wind speed throughout the foraging trip was higher. No significant relationship was found between bird mass and duration of foraging trips. The mean cumulative distance flown by the ten birds tracked in 1999 was 4262 km + 1318 (SD). Eight of the ten SRA employed a ‘commute, forage, commute’ foraging strategy, and the other two alternated short bouts of commuting and foraging. Commuting phases were characterised by rapid directional flight with a straight-line distance (range) of 180 km to 800 km between positions 24 hours apart. Foraging phases were characterised by a range of less than 180 km per 24 hour interval and frequent tight turns. Displacement rate between successive uplinks was significantly higher during commuting phases (28.6 kph + 1.93 SE) than foraging phases (15.1 kph + 1.4 SE). Wind strength and direction influenced the timing of the return commute to the colony. SRA covered greater distances at more favourable wind angles relative to flight track (broad reach and close reach) than in head, tail or direct side winds. Birds of low mass (< 8kg) made fewer landings in winds above 40 kph than in lighter winds, whereas heavier birds had a similar level of landing activity across all wind speed bands. One bird was delayed for several days by light winds, and another flew off course during strong winds. Two birds exploited the same window of wind conditions to return to the colony, each flying a similar course in both timing and route. These results define the foraging strategies of SRA during incubation, and demonstrate the influence of wind conditions and other factors on the overall duration of foraging trips and on the timing of commuting and foraging phases.

There are many factors that have been proposed to contribute to plant invasiveness in nonnative ecosystems. Traits of invading species are one of them. It has been proposed that successful species at a certain invasion stage share particular traits, which could be used to predict the behaviour of potentially invasive plants at the respective stage. Three main stages of invasion are distinguished: introduction, naturalization, and invasion. I conducted a stageand trait-based analysis of available data for the invasion of New Zealand by the flora of southern Africa. Using 3076 southern African native vascular plant species introduced into New Zealand, generalised linear mixed model analysis was conducted to assess association of several species traits with the three invasion stages. The results showed that plant traits were significantly associated with introduction but fewer traits were associated with naturalization or invasion, suggesting that introduction can be predicted better using plant traits. It has been also hypothesized that species may become invasive in non-native ecosystems because they are removed from the regulatory effects of coevolved natural enemies (Enemy Release hypothesis). A detailed field study of the succulent plant Cotyledon orbiculata var. orbiculata L. (Crassulaceae) was conducted in the non-native New Zealand and native Namibian habitats to compare the extent of damage by herbivores and pathogens. C. orbiculata is a southern African species that is currently thriving in New Zealand in areas seemingly beyond the climatic conditions in its native range (occurring in higher rainfall areas in New Zealand than are represented in its native range). As hypothesised, C. orbiculata was less damaged by herbivores in New Zealand but, contrary to expectation, more infected by pathogens. Consequently, the plant was overall not any less damaged by natural enemies in the non-native habitat than in its native habitat, although the fitness impacts of the enemy damage in the native and invaded ranges were not assessed. The results suggest that climatic conditions may counteract enemy release, especially in situations where pathogens are more prevalent in areas of higher rainfall and humidity. To forecast plant invasions, it is concluded that species traits offer some potential, particularly at the early stage of invasion. Predicting which introduced plants will become weeds is more difficult. Enemy release may explain some invasions, but climatic factors may offset the predictability of release from natural enemies.

The long-tailed bat (Chalinolobus tuberculatus) is a threatened species endemic to New Zealand. Historical anecdotes indicate that long-tailed bat populations have declined. However, it is unknown if all populations have declined and if declines are historical or ongoing. Thus, the development and implementation of a national network of long-tailed bat monitoring sites is a priority of the Department of Conservation's Bat Recovery Plan. Potentially, information gained from a national monitoring programme would assist conservation managers to target resources towards those areas where bat populations are declining and provide baseline information to assist managers to gauge the impact of management techniques on bat populations. Of critical importance is that unless it can be demonstrated that long-tailed bat populations have declined and that, that decline is real, management will not be initiated. The aim of this research was to investigate aspects of the ecology of long-tailed bats that would influence the development of a monitoring programme. The distribution, roost selection, habitat use, and activity patterns of a long-tailed bat population that persisted in the Waitakere Ranges, Auckland, were investigated. A study of the Waitakere Ranges long-tailed bat population was significant because (1) the Waitakere Ranges is the northern most location at which long-tailed bats have been researched; (2) the study was the first to be conducted on a long-tailed bat population that persisted in kauri Agathis australis dominated forest remnants; (3) the long-tailed bat population in the Waitakere Ranges is the only known extant population in close proximity to a major urban area; and (4) the factors that are attributed to long-tailed bat population declines (i.e., forest clearance, predation and urbanisation; O'Donnell, 2000) are likely to be ongoing and intensified in the Waitakere Ranges. Twenty roosts were located. Most roosts (85%) were in kauri, 2 were in mature rimu (Dacrydium cupressinum) and 1 was in a kahikatea (Dacrycarpus dacrydioides). All roosts were in large, live, emergent trees. Mean height of roost trees was 38.4 ± 1.3 m and average DBH was 186 ± 12 cm. The entrances of six roost cavities were identified all were located in minor lateral branches in the crown of the tree and were primarily near the tip of branches. Roosts were a mean height of 24.6 ± 3.7 m above ground level. It was argued that roosts in the crowns of kauri were inaccessible to terrestrial mammalian predators. Twenty-eight roost watches were conducted. The average number of bats counted leaving roosts was 10.0 ± 1.5 (maximum = 24). Roosts were occupied by radio-tagged bats for an average of 2.0 ± 0.4 days, and 11 (55 %) were occupied for only one day. Roost size was the lowest reported for long-tailed bats. Roost switching also appeared higher than in other populations that have been studied. It was argued that morepork predation may have a significant impact on the population viability of the population. As in other studies long-tailed bats were found to forage over modified habitats including over farmland, dwellings, orchards and along streams and roads with little vehicular traffic. Long-tailed bats foraged throughout the Waitakere Ranges and their foothills. Bat activity was highly variable. Of the environmental variables analysed, temperature was found to have the greatest influence on bat activity. There were seasonal and habitat influences on bat activity. The relationship between sample sizes, variation in bat detection rates and desired statistical power using automatic bat detectors to monitor populations of bats was explored. A power analysis on activity data collected with automatic bat detectors indicated that declines in bat populations would need to be reflected in declines of greater than fifty percent in bat activity before monitoring programmes would have sufficient power to detect declines in activity. It was recommended that monitoring programmes should concentrate on intensive presence – absence surveys rather than long-term studies at a few sites.

In the late 1980s the endangered Chatham Island oystercatcher (Haematopus chathamensis) (CIO) was estimated at less than 110 individuals. Endemic to the Chatham Islands, New Zealand, it was feared to be declining and, based on existing productivity estimates, in danger of extinction within 50-70 years. These declines were thought to be caused by numerous changes since the arrival of humans, including the introduction of several terrestrial predators, the establishment of marram grass (Ammophila arenaria) which changes dune profiles, and increased disturbance along the coastline. The New Zealand Department of Conservation has undertaken recovery planning and conservation management to increase CIO numbers since the late 1980s. Recovery planning raised some key research questions concerning the population dynamics, habitat selection, and breeding biology of Chatham Island oystercatcher (CIO), and the critical factors currently limiting the population. The objectives of this study were to collect and interpret data on: 1) population size, trends, and distribution across the Chathams, 2) basic breeding parameters, 3) recruitment and mortality rates, 4) habitat selection at the general, territorial and nest-site levels, 5) habitat factors that are correlated with territory quality, and 6) cues that elicit territorial behaviour in CIO.

Minimising the potential impact of forest management requires an understanding of the key elements that maintain forest diversity and its role in ecological processes. Invertebrates are the most diverse of all biota and play important roles in maintaining forest processes. However, little is known about invertebrates in New Zealand's beech forests or the degree to which selective beech harvest might impact on their diversity and ability to carry out ecosystem processes. Studying ecosystem responses to disturbance is considered vital for understanding how ecosystems are maintained. One of the main objectives of this research was to assess whether litter-dwelling invertebrates were susceptible to the impacts of selective harvest and, if so, whether they could be used as indicators of forest health. Changes in invertebrate diversity could have important implications for nutrient cycling and primary production in forests. Litter-dwelling invertebrates contribute to the process of decomposition by increasing the surface area of the leaves, mixing soil organic matter and by infecting leaf particles with soil microbes. This investigation into the function of invertebrates in beech forest was carried out in the context of ecological theories which relate diversity to ecosystem stability and resilience. A replicated study was established in Maruia State Forest (South Island, New Zealand) to assess the potential biotic and abiotic impacts of sustainable beech harvest. Litter-dwelling invertebrates and environmental factors were monitored during 1997, before harvest, to determine how much variability there was between study sites. Specifically, litter pH, light intensity, litter fall, litter temperature, moisture as well as invertebrate abundance and diversity were compared before and after selective harvest. On 17 January 1998, two to three trees were selectively harvested from three of the nine study sites. On 15 February 1998 a similar number of trees were winched over or felled manually to create artificial windthrow sites. The remaining three undisturbed sites were used as controls. Invertebrates belonging to the detritivore guild were assessed from litter samples and a series of litter-bags containing pre-weighed leaf litter which were placed in each of the sites to assess rates of litter decomposition. Millipedes (Diplopoda: Polyzoniidae, Schedotrigonidae, Dalodesmidae, Habrodesmidae, Sphaerotheridae), earthworms (Oligochaeta: Annelida), tipulid larvae (Diptera: Tipulidae), weevils (Coleoptera: Curculionidae), moth larvae (Lepidoptera: Oecophoridae, Tortricidae and Psychidae), slaters (Isopoda: Styloniscidae), Oribatid mites (Acarina: Cryptostigmata) and landhoppers (Crustacea: Amphipoda) were extracted from the litter-bags and their abundance and diversity was compared between the three treatments. Weight loss from the litter-bags and the carbon and nitrogen content of litter were used to measure the rate of decomposition in each treatment. An additional study investigated whether exclusion of invertebrates from leaf litter resulted in reduced rates of decomposition. The results indicated that there was an increase in light intensity and a small increase in temperature following selective harvest and artificial windthrow. There was no significant difference in litter moisture or the amount of litter fall between the treatments. Invertebrate abundances were significantly affected by season but did not appear to be affected by selective harvest or artificial windthrow. The diversity of invertebrates remained relatively constant throughout the year, as did the rate of decomposition. When invertebrates were excluded from the leaf litter there was no consequential effect on the rate of litter decomposition. This suggests that there may be compensatory mechanisms taking place between the trophic levels of the food web to maintain processes and that direct links between invertebrates and decomposition are relatively weak. In conclusion, it appears that the effects of selective beech harvest on forest-floor processes were minimal and are comparable to those created by natural windthrow disturbance. It also appears that macroclimatic effects such as seasonal climatic effects have a large effect on forest biota. As none of the invertebrates studied appeared to be detrimentally affected by selective harvest and as there was no direct link demonstrated with decomposition, it was considered inappropriate to advocate the use of this group of invertebrates as indicators of sustainable forest management. The results from this study provide information which may help inform decisions on the future management of diversity in beech forest ecosystems.

The discovery and taxonomic history of fluttering shearwater (Puffinus gavia (Forster) and Hutton's shearwater (Puffinus huttoni Mathews) are reviewed. Taxonomic theory, where appropriate to this thesis, is discussed. The external morphology of P. gavia and P. huttoni is compared. No single external measurement or plumage character separates more than 60% of birds examined. The best system of identification is to compare the ratio of different body parts within an individual bird. The distribution of P. gavia and P. huttoni is compared. Hutton's shearwater feeds further out to sea and it is believed to be a migrant species wintering in north west Australian waters. The fluttering shearwater is believed to be a semi-migrant species with only the juveniles spending time in south east Australia. The red cell enzymes of P. gavia, P. huttoni and P. griseus are compared. There are differences in two esterase loci between gavia and huttoni, while P. griseus is more distantly related. Nei's genetic identity values are calculated. The systematic value of electrophoretic data is discussed. The relationship of an undescribed subfossil shearwater to P. gavia and P. huttoni is discussed. An outgroup analysis to other shearwater species is carried out according to phylogenetic (cladistic) theory. The subfossil shearwater is most closely related to the fluttering shearwater, and these two form a sister group to Hutton's shearwater. These three species are a sister group of P. opisthomelas. The relationship between the many P. assimilis subspecies, the black-backed Manx shearwaters, and the gavia, huttoni and opisthomelas group was not resolved. Puffinus nativitatis is more closely related to the Manx and the little shearwaters than to the P. griseus, P. tenuirostris group.

There is anecdotal evidence that populations of the New Zealand endemic red admiral butterfly Bassaris gonerilla (F.) have declined since the early 1900s. This decline has been associated with the introduction of the generalist pupal parasitoids Pteromalus puparum (L.) and Echthromorpha intricatoria (F.). The former was deliberately introduced for the biological control of the cabbage white butterfly (Pieris rapae (L.)); the latter is an adventitious arrival from Australia. The objective of this thesis was to quantify, using population models, the effect that P. puparum is having on B. gonerilla abundance. Population monitoring and a phenology model (based on temperature-related development rates) indicated that B. gonerilla has two full generations and one partial generation per summer in the Banks Peninsula region of New Zealand. B. gonerilla abundance was greatly reduced in drought summers, which was probably due to the negative effects of drought on the quality and quantity of the larval host plant Urtica ferox Forst. A life table study showed that egg parasitism by the unidentified scelionid Telenomus sp. was the largest mortality factor for the pre-imaginal stages of B. gonerilla, followed by "disappearance" mortality (predation and dispersal) in the larval stages. Pupal mortality due to P. puparum was lower compared with that caused by E. intricatoria, with 1-19% and 20-30% of pupae being parasitised by P. puparum and E. intricatoria, respectively. Collection of B. gonerilla pupae from the Christchurch, Dunedin and Wellington areas confirmed higher rates of percentage parasitism by E. intricatoria. B. gonerilla collected from the Banks Peninsula had a 50: 50 sex ratio and lifetime fecundity was estimated in the laboratory as 312 eggs per female. There was no evidence of density-dependent parasitism of B. gonerilla pupae by P. puparum in the field, although there was a significant positive relationship between life table estimates of E. intricatoria parasitism and B. gonerilla pupal abundance. Larval dispersal from the host plant showed a positive relationship with larval instar but no relationship with larval density. Rates of change in B. gonerilla adult abundance between generations within a year showed evidence of density dependence, and this negative feedback was stronger in a drought year. A discrete-time model for B. gonerilla population dynamics was constructed which had two summer generations per year and a partial overwintering generation. The model showed that the presence of this overwintering generation provides a temporal refuge from high levels of E. intricatoria parasitism. Removal of parasitoid mortality from the model suggested that P. puparum was suppressing B. Gonerilla populations on the Banks Peninsula by 5% and E. intricatoria by 30%. An important assumption of the model was that parasitism rates were independent of B. gonerilla density. This assumption appears valid for P. puparum parasitism, but may not be valid for E. intricatoria; therefore the estimated suppression levels due to this adventive parasitoid should be viewed with some caution. It is too soon to generalise on what determines the magnitude of non-target effects by arthropod biocontrol agents, this being only the second study to quantify effects at a population level. However, in this case retrospective analysis has shown that the impact of non-target parasitism by P. puparum on B. gonerilla abundance has been small. There is anecdotal evidence that populations of the New Zealand endemic red admiral butterfly Bassaris gonerilla (F.) have declined since the early 1900s. This decline has been associated with the introduction of the generalist pupal parasitoids Pteromalus puparum (L.) and Echthromorpha intricatoria (F.). The former was deliberately introduced for the biological control of the cabbage white butterfly (Pieris rapae (L.)); the latter is an adventitious arrival from Australia. The objective of this thesis was to quantify, using population models, the effect that P. puparum is having on B. gonerilla abundance. Population monitoring and a phenology model (based on temperature-related development rates) indicated that B. gonerilla has two full generations and one partial generation per summer in the Banks Peninsula region of New Zealand. B. gonerilla abundance was greatly reduced in drought summers, which was probably due to the negative effects of drought on the quality and quantity of the larval host plant Urtica ferox Forst.. A life table study showed that egg parasitism by the unidentified scelionid Telenomus sp. was the largest mortality factor for the pre-imaginal stages of B. gonerilla, followed by "disappearance" mortality (predation and dispersal) in the larval stages. Pupal mortality due to P. puparum was lower compared with that caused by E. intricatoria, with 1-19% and 20-30% of pupae being parasitised by P. puparum and E. intricatoria, respectively. Collection of B. gonerilla pupae from the Christchurch, Dunedin and Wellington areas confirmed higher rates of percentage parasitism by E. intricatoria. B. gonerilla collected from the Banks Peninsula had a 50: 50 sex ratio and lifetime fecundity was estimated in the laboratory as 312 eggs per female. There was no evidence of density-dependent parasitism of B. gonerilla pupae by P. puparum in the field, although there was a significant positive relationship between life table estimates of E. intricatoria parasitism and B. gonerilla pupal abundance. Larval dispersal from the host plant showed a positive relationship with larval instar but no relationship with larval density. Rates of change in B. gonerilla adult abundance between generations within a year showed evidence of density dependence, and this negative feedback was stronger in a drought year. A discrete-time model for B. gonerilla population dynamics was constructed which had two summer generations per year and a partial overwintering generation. The model showed that the presence of this overwintering generation provides a temporal refuge from high levels of E. intricatoria parasitism. Removal of parasitoid mortality from the model suggested that P. puparum was suppressing B. Gonerilla populations on the Banks Peninsula by 5% and E. intricatoria by 30%. An important assumption of the model was that parasitism rates were independent of B. gonerilla density. This assumption appears valid for P. puparum parasitism, but may not be valid for E. intricatoria; therefore the estimated suppression levels due to this adventive parasitoid should be viewed with some caution. It is too soon to generalise on what determines the magnitude of non-target effects by arthropod biocontrol agents, this being only the second study to quantify effects at a population level. However, in this case retrospective analysis has shown that the impact of non-target parasitism by P. puparum on B. gonerilla abundance has been small.