We appreciate the concerns raised by Grover & Lawton (p. 484) regarding our recent work. The central issue raised by Grover & Lawton is the primacy of stability as a property which maintains the status quo-an issue with which ecologists and evolutionary biologists have grappled for a very long time. There is little doubt that thought about the nature of stability strongly influences the design and interpretation of ecological experiments, as well as the form of prevailing paradigms. As a result, investigations exploring different realms of a system's behaviour often expose dynamics and behaviour which cannot be reconciled within the framework of current ecological thought. Rather than being at odds with Grover & Lawton, we find our respective approaches complementary if not integral to the development of a broader framework of community phenomenology, particularly given the emerging characterization of the community assembly state space (Drake 1990, 1991; Drake et al. 1993; Luh & Pimm 1993). To address Grover & Lawton's concerns, we begin by defining an assembly map as a convoluted mesh of variously crossing, and sometimes intersecting, trajectories. Trajectories are defined by a given set of species and associated environmental effects moving through space and time. Hence, a trajectory is one of many sequences of community states or configurations which exist and move through time (cf. assembly, succession). The form any trajectory can take is initially limited by environmental and geological vagaries. Subsequently, however, the direction a community assembly trajectory can take is richly varied. Factors such as the sequence and timing of species invasion and reinvasion, and the evolutionary and real-time dynamical interactions which occur therein, are all capable of directing a system along different, though equally plausible, trajectories. The assembly map as described above has several specific geometric properties which tie together studies such as ours. First, there are unreachable portions of the assembly map (e.g. Diamond 1975; Sugihara 1985); these regions represent community configurations which simply cannot exist. Secondly, the map of trajectories contains alternative pathways (some shorter than others) between distant community states. The system can reach state C by traversing states A-B-D in sequence, but the sequence AD-C might also exist containing very different mechanisms responsible for the organization of state C. Some trajectories have the same solution, while others traverse very different portions of the assembly space despite identical initial conditions. Disturbance can reset a trajectory onto another or simply back up the current trajectory. So defined, we have a map which is woven together into a mesh of trajectories or permissible states. Viewed in this manner we believe that our respective approaches simply focus on different portions of the assembly space-an essential difference if we are to develop a more general framework of community organization. It appears as though Grover & Lawton believe that some aspects of the assembly space offer more information than others. If this were a general rule nature should have relatively few assembly trajectories with which to define community organization. In the laboratory and in nature we often find a wide range of trajectories producing both transitory and persistent species ensembles-even within the confines of very similar environments. We also wish to reiterate that the trajectories we observed were the results of invasion timing, dynamics, and the interplay between the two. On a more technical side, our experimental design involved a relatively rapid sequence of invasions without maintenance of constant nutrient levels. However, Grover & Lawton suggest that the slowest meaningful invasion schedule, invasion only at successive equilibria (implicitly maintaining nutrient levels) is preferable. One feature of natural freshwater systems, particularly those in temperate regions, is quite rapid seasonal variation. This feature was a major concern in the design of our experiment because maintaining relatively constant conditions for an extended period would surely reduce environmental variation from our assembly trajectories. We chose to focus on the historical assembly processes which are capable of altering the expression of the mechanisms (e.g. predation, competition) often posited as the causal agent. To suggest, as Grover & Lawton do, that only certain time intervals are interesting is simply not justifiable by any scientific evidence we are aware of. Rather than 'rapid' invasions being extreme we feel that such a schedule is more likely to be the norm than 488