<p>Current research has emphasized the need to identify and quantify the effects of positive and negative interactions (both direct and indirect) between species, taking into account the influence of abiotic conditions and spatial scales. In this research it is particularly challenging to adequately assess and predict the impact of introduced species on native communities. This study examined interactions of introduced and native species on coastal sand dunes in New Zealand. Substantial areas of this habitat have become dominated by the highly competitive exotic sand-binder marram grass (Ammophila arenaria) and other exotic weeds, replacing native species, including the native eco-engineering spinifex (Spinifex sericeus). First, I examined direct interactions (competition and facilitation) between marram and spinifex along an abiotic stress gradient where experimental plots were subject to different restoration management techniques. In my large-scale dune experiment I planted 2475 spinifex seedlings in three different treatments (bare sand, live marram, dead marram) including an unplanted control, and monitored the plots for one year along exposed marram grass-dominated dune fields near Whanganui, West Coast, North Island/New Zealand. The stress gradient hypothesis predicts facilitation will be greatest where stress is most severe. I hypothesized facilitation of spinifex plantings and other self-colonizing plants in dead sprayed marram compared to live marram and a change of interaction between spinifex and marram grass along an abiotic stress gradient. Spinifex survival was not significantly different across treatments, but the interaction between treatment and location from the sea as well as pre-existing marram cover were significant predictors of plant growth in a linear mixed effect model. Exotic weeds such as Senecio elegans, Conyza canadensis, dandelions and legumes were facilitated by sprayed marram grass relative to abundances in live marram grass, while grasses other than marram grass and spinifex performed best in bare sand, in the absence of competitors. For S. elegans, abundances were higher closer to the sea. Spinifex growth was greatest in plots closest to the sea. Comparing spinifex growth in live marram grass and dead marram showed similar patterns at the fore, mid and back dune, but sprayed maram showed better facilitation of spinifex. For restoration plantings it is only recommended to plant into dead marram if the site is not weed prone as weeds were facilitated in the same way as spinifex and in some instances appeared to hinder spinifex growth. Crucial for a successful conversion from marram to spinifex is that the initial marram cover is not too high prior to spraying- lower densities allow for better spinifex growth. Second, I examined indirect competition with a survey of natural populations and a common garden experiment at a shingle beach. I was interested in determining the influence of plant density at different spatial scales as well as plant morphology on insect abundances. I studied indirect interactions by choosing native and introduced Senecio spp. as host plants for insect colonization and incorporated fine scale plant density (50cm radius circle, area = 0.8 m²) and coarse scale plant density (6m quadrat annulus, area = 32 m²) as predictors for insect colonization in addition to plant morphology. I surveyed Senecio spp. over a period of three growth seasons and conducted an experiment where I manipulated densities of pairs of species of either introduced Senecio elegans, S. skirrhodon and native S. lautus. My survey of natural populations and the manipulative field experiment show a negative impact of high conspecific and heterospecific plant density on the colonization of the seed head predator fly Sphenella fascigera. Along with plant density, plant size was a significant predictor of insect abundances. Higher densities of introduced S. elegans indirectly facilitated S. lautus at fine spatial scales by reducing the incidence of S. fascigera. This supports the resource dilution hypothesis which predicts higher insect herbivore numbers on isolated resource patches. Thus, my results provide empirical evidence for apparent facilitation of a native plant by an introduced plant via a shared herbivore. For future assessments of the impacts of invasive species it will be important to consider the net-outcome of direct and indirect competitive and facilitative interactions. In particular, for restoration purposes in stressful environments removal of invasive species may have to occur in a carefully controlled manner taking into account the abiotic conditions and spatial scales at which interactions occur.</p>