Resource constraints can affect species on multiple levels. In this dissertation, I combine laboratory experiments, an ecosystem-level manipulation experiment and statistical modeling to examine how resources maintain and constrain cave biodiversity and structure cave communities. Chapter I examines how N-limitation may drive morphological adaptations of cave arthropods. By analyzing free amino acid contents, I show that, in comparison to cave-transient millipedes, cave-obligates have decreased concentrations of essential, nonessential and N-rich amino acids, and amino acids associated with pigmentation and cuticular development. Chapter II tests the hypothesis that stoichiometric mismatches impose growth constraints on cave animals. Although results show that cave resource quality is similar to surface leaves, I do show that millipedes experience a strong mismatch to their food. Also, cave-obligate millipedes have lower %P and RNA/DNA (protein synthetic capacity) compared to cave-transient millipedes. Results from these chapters suggest that cave adaptations may reflect stoichiometric challenges of caves. Chapter III describes the manipulation experiment, wherein I removed all organic material from 12 caves, and, while excluding all natural subsidies, I added standardized quantities of leaf packs or rodent carcasses. For 23 months, I monitored the recipient communities to see how subsidies influence species abundance, diversity, and community dynamics. I observed 19,866 arthropods representing 102 morphospecies. Rat treatments supported greater abundances, but the treatments did not differ in richness. Multiple community-level analyses demonstrated that community composition differed drastically depending on treatment. Lastly, the communities changed directionally over time, diverging faster in caves receiving leaves. Chapter IV uses annual bioinventories of 65 caves to investigate occupancy patterns of terrestrial invertebrates. I estimated richness using classical estimators in concert with estimators that incorporate detection. I also used multispecies occupancy models to examine relationships between estimated richness and physical cave characteristics; demonstrating the importance of cave length, entrance geometry (a surrogate for energy input), and connectivity. The results show how inventory data, even if incomplete, can provide valuable information about the distribution of rare species. Resource availability can affect cave ecosystems on multiple levels. Here I illustrate how the biochemical composition, community dynamics, and occupancy patterns of cave species are influenced by resource constraints.