Debates about the nature of ecological communities have shaped many aspects of the science. Ecologists of the late 19th and early 20th centuries generally saw communities as highly interdependent systems or even superorganisms whose composition was shaped by biotic interactions and coevolution. The contrary view, that the physical environment in a location pretty much controlled what plants lived there, gained ascendency later. This idea, modified to included the effects of chance and some competition, is still dominant today. However, mutualism is finally beginning to get some recognition as an important factor determining community composition. (The word "mutualism" refers to an interaction between two species that benefits both of them -- what laypeople typically think of as "symbiosis".) The debate is not merely academic. It has real consequences for conservation and restoration of damaged ecosystems.
The May 2006 issue of Frontiers in Ecology and the Environment contains a review paper by Francisco M. Padilla and Francisco I. Pugnaire about the use of nurse plants in ecological restoration. This approach involves planting the plants you want next to larger plants already growing in the area. (Yes, I get paid for each use of the word "plant".) The nurse plants may shield the target species from harsh weather, bring up water and nutrients, protect against grazers and attract pollinators. However, this approach doesn't always work. Environment matters, as does the identity of the nurse and target plants.
Nurse plants are most helpful in harsh environments like chapparral and alpine tundra. In easier conditions, competition between the two plants trumps any beneficial effects. For the same reason, nurse plants are more useful in dry years than wet years. This counterintuitive effect may occur because the deep roots of the nurse plant bring up water and the shade it casts helps the establishing seedling conserve water.
The species of the nurse and target plants also makes a difference. Plants that release growth inhibitors into the soil don't make good nurse plants. (Duh.) Late successional species benefit more from having nurse plants than early successional ones. And, although the authors don't discuss this, I'd bet pairing close relatives will increase competition and reduce beneficial effects.
The importance of this synthesis goes beyond practical restoration work to providing key insights for community ecology. If nurse plants matter in restoration succession, they probably also make a difference in natural succession. Since these interactions show at least some species specificity, we can draw a similar conclusion about species interactions in succession and ecosystem development. Finally, both biotic interactions and physical conditions make a difference; indeed, physical conditions can determine the nature of the interaction between nurse and target plants. The same effect can be expected in the wild.