Abstract. This paper reviews the literature on neighbour manipulation experiments on species intemctions in natural of semi-natural vegetation. Three major approaches have been used: 1) introductions into vegetation, 2) trenching, and 3) vegetation removals. Some studies have used a combination of approaches, especially 1) and 3). The removal approach has been used most commonly and is given the greatest emphasis in this paper accordingly.
Details of over 50 such studies are summarized here, 48 of which have appeared within only the last two decades. We compare and contrast the experimental designs of these studies and examine their contributions to the understanding of species interactions in vegetation. Evidence for competition is virtually universal among these studies. Several studies have also detected evidence of beneficence between plant species. Numerous factors may confound interpretations in removal experiments including: life stage dependent species responses, species-dependent timing and speed of response, inadequacy of controls due to temporal and/or spatial variability in site quality, indirect effects of treatment on soil moisture content or nutrient levels, or on the activities of predators or decomposers, and several constraints inherent in particular experimental designs. Another level of complexity arises as a consequence of several variables related to plant attributes that directly determine the nature of interactions between neighbours (e.g. relative competitive abilities, the magnitude of beneficial interactions, the extent to which neighbours make demands on the same resource units). These may interact in a complex manner to affect the response of a ‘target’ plant to the removal of neighbours. Recommendations for future studies are considered and a neighbourhood experimental design is proposed which enables analysis of the extent to which the fates of naturally established individuals following vegetation removal can be accounted for (in multiple regressions) by several variables that reflect different properties and circumstances of interaction with immediate Thiessen neighbours in the field.