Nitrogen (N) fixing trees are commonly used to promote forest restoration in disturbed areas because they can quickly recreate forest canopy structure. That structure in turn is hypothesized to attract animal seed dispersers and create enough shade to reduce undesirable species (particularly grasses). Yet N-fixers tend to increase soil N availability, which could facilitate the spread of nitrophilous invasive species. This dissertation evaluates the long-term consequences for understory community composition of establishing three N-fixing tree species (Acacia koa, Sophora chrysophylla, and Morella faya) after exotic grass-fueled fire in the seasonally dry subtropical woodland in Hawaii. To understand the restoration potential of these species, I compared discrete single-species stands of N-fixing trees in burned areas to both an intact native woodland and burned, open sites with no tree cover. Although N-fixing species are often assumed to be ecologically similar, trait variation among N-fixing trees in this system was strong enough to differentiate understory communities among stands of the three N-fixer species. To understand the mechanisms driving differences in understory composition among site types, particularly among N-fixing trees, I characterized the abiotic environment created by these species in terms of light and N availability, both of which were important drivers of understory community composition. High light and N availability were associated with greater exotic species cover and unique exotic species. Surprisingly, N availability was highest and N cycled fastest beneath the relatively slow-growing S. chrysophylla despite having much lower litter-N inputs than the faster-growing A. koa and M. faya. In this study, fast N-cycling was associated with high specific leaf area, high foliar N content and low foliar lignin:N. These traits are consistent with fast leaf economic spectrum traits in the general ecological literature, but this approach has not previously been applied to distinguish among N-fixing trees. Native Hawaiian dry forest understory recovery, particularly that of woody species, was limited throughout the burned area regardless of canopy cover. To determine what limits native shrub recovery, I sampled the seed bank and recorded natural seedling germination. I also planted native seedlings into the understory of all site types and either removed or left intact the invasive grass grasses present in the understory. I found that native shrubs were limited by both seed availability and competition with exotic grasses. Although outplant survival did not vary by N-fixer species identity, differences in the mechanisms by which each N-fixing species limited native seedling survival likely play a role in understory community assembly long-term. When restoration occurs in the context of secondary succession, prioritizing the creation of forest structure using N-fixing trees, particularly open-canopied fast-cycling species, such as S. chrysophylla, could make full community recovery more difficult by promoting rather than suppressing exotic grasses.