We tested whether the density and spatial distribution of available nest-sites had an effect on community structure. We performed an experimental manipulation to examine whether nest-sites are limiting in this system and the effects on community structure. The density and spatial arrangement of available nest sites was manipulated using flagstone tiles that served as available nest sites. 
Results/Conclusions

Species richness increases nonlinearly with rock density (R² = 0.632, F_1,26 = 44.6225, df = 1, p <0.0001). There were significant differences in richness between sites (F_3,14 = 3.9407, p = 0.0313) and richness declined over the four summer months (F_3,12 = 21.2538, p<0.0001). Richness between high and low tile density plots was significantly different (F_1,14=5.9195, p=0.0290); on average, high density plots had more species (1.727 + 0.132, n=56) than low density plots (1.120 + 0.140, n=56). The spatial distribution of tiles had no effect on species richness. There was a significant interaction between the density of available tiles and their spatial arrangement (F_2,23=5.4017, p=0.0199). The greatest richness was found on plots with a high tile density in a dispersed configuration. On low tile density plots the greatest richness occurred when tiles were in a clumped distribution.
The proportion of tiles occupied also declined significantly as the summer progressed (F_2.722,62.598=68.5832, p<0.0001). Low tile density plots had a greater occupancy than high tile density plots (F_1,23=6.2662, p=0.0199). There was no effect of the spatial arrangement of tiles on occupancy. These data suggest there is competition for available nest-sites.
Inter-annual effects were examined by analyzing the differences in the proportion of tiles occupied by species. The percentage of tiles occupied was not different between 2006 and 2007 (r=0.206, t=-1.82, df=93, p=0.0719), or between 2007 and 2008 (r=0.44, t=1.74, df=122, p=0.0844).  In both years, there were significant differences among species in colonization ability (F_15,107=4.6590, p<0.0001), however the within species differences were not significant (F_15,107=1.2454, p=0.2639).
A principal components analysis of temperature and humidity underneath tiles revealed that the maximum temperature accounted for the most variation in principle component 1 and mean relative humidity in principle component 2. Principle component 1 was negatively correlated with occupancy of nest-sites (Spearman’s ρ=-0.943, p=0.0048).