Background/Question/Methods    Plant community structure varies temporally and spatially.  One effective method for understanding the sources of this variation is to separate the effects of regional processes, such as species pool and climate, from the effects of local processes such as species interactions.  We previously showed that both local effects of species interactions and year-to-year climate variation reduced diversity in two communities in stabilized sand dunes, but that only regional differences in species pools explained the greater species richness at the more arid site.  Here, we present results from an experiment designed to further test the effects of species interactions and abiotic conditions on species richness, diversity, and composition in two annual plant communities.  In each community (coastal and desert), resource levels (irrigation) and potential for species interaction (initial planting density) were manipulated in a completely crossed design.  Observed species richness and evenness were compared to values expected if all plants occurred at random.  Species composition patterns were examined with redundancy analysis.

Results/Conclusions    Species richness and evenness responded to source community, irrigation, and planting density.  Because of a smaller regional species pool, the richness and evenness showed a greater deviation from the random expectation in the coastal community than in the desert community.  At high planting densities, intense competition reduced richness and evenness, but at low planting densities an absence of facilitative interactions also reduced diversity; for evenness, the facilitative effect was stronger in the coastal community.  Finally, filtering of species richness and evenness was greatest under the low irrigation treatment.
    Several dominant species occurred more frequently than expected in particular treatments, indicating that they were good or poor competitors, or good or poor drought tolerators.  Considering all species simultaneously, there were consistent differences in species composition between communities with low and high planting density.  Therefore, there is some systematic filtering of species in response to interaction intensity.  On the other hand, total community composition did not respond strongly to irrigation, suggesting that many of the less abundant species respond randomly to precipitation.