In New Zealand, large-scale disturbance in the form of anthropogenic burning and clearance for agriculture began with Maori settlement in 1280 AD and intensified with European settlement after 1840. Successional vegetation in New Zealand is highly flammable when certain native trees (Leptospermum scoparium, Kunzea ericoides) and invasive species (Ulex europaeus) establish. Succession to forest, where abiotic conditions allow it, is slowed by seedling herbivory and reduced dispersal and is dependent on facilitative effects of the early stage canopy trees. The aim of this thesis was to quantify how herbivory, dispersal and facilitation interact to affect regeneration in Awarua wetland near Invercargill, and to test restoration techniques at the site. Drier parts of the Awarua wetland complex were previously in lowland native forest but are now dominated by L. scoparium and have been burnt repeatedly. I tested whether regeneration was dispersal limited using a vegetation survey at four sites in the wetland. I found dispersal limitation at multiple scales (among and within sites). This conclusion was strengthened by seed trapping around a forest remnant, where I found only three of 11 native forest species were being dispersed even small distances (< 150 m) from the remnant. I quantified the effects of competition, facilitation, and seed predation and herbivory on seedling establishment, survival, growth, and species diversity and evenness with a seed sowing experiment of 2.6 years duration. I found that abiotic conditions in the wetland were generally suitable for forest species to grow; that the L. scoparium was relatively young and should allow more forest regeneration as it ages; that the wetland was seed limited; and that excluding herbivory and seed predation doubled the number of surviving seedlings. Chionochloa tussock communities are naturally present within the wetland and increase in abundance after fire, in the absence of which they are invaded by L. scoparium. I used a seedling and seed sowing experiment to test whether forest seedlings could successfully establish in tussock areas. I found seedling establishment was limited by pH in both tussock and control (L. scoparium) areas, and confirmed that biotic constraints (herbivory and seed predation) were slowing regeneration. I also tested two relatively novel methods in New Zealand. I conducted three pest surveys of the wetland using chew cards and propose guidelines for nationally consistent use that reduce errors due to interference between mammal species. I tested the cost effectiveness of bird perches for speeding succession in restoration ecology, and found they will rarely, if ever, be the best option. Finally, I assessed factors affecting the speed of forest regeneration at a landscape scale with a spatially explicit vegetation model, parameterised from my field experiments and the literature. This model suggests that the shortage of existing forest fragments is the key factor slowing regeneration within the landscape, and interacts with seed predation and seedling herbivory. The Ramsar convention on wetlands requires maintaining ecological character in listed wetlands such as Awarua. My results contribute to this planning by highlighting the current vulnerability of the wetland to fire, showing the importance of seed limitation, and quantifying which actions would best facilitate forest regeneration.