Monitoring is a major component of restoration efforts and adaptive management. But monitoring is often neglected because it can be expensive and time-consuming. Accordingly, it is valuable to determine whether the temporal extent of sampling alters the validity of inferences about the response of diversity measures to environmental variables affected by restoration actions. Non-native species alter ecosystems in undesirable ways, frequently homogenizing flora and fauna and extirpating local populations of native species. In the Mojave Desert, invasion of salt-cedar (Tamarix ramosissima Ledeb.) and human efforts to eradicate salt-cedar have altered vegetation structure, vegetation composition, and some measures of faunal diversity. We examined whether similar inferences about relationships between plants and butterflies in the Muddy River drainage (Nevada, U.S.A.) could have been obtained by sampling less intensively (fewer visits per site over the same period of time) or less extensively (equal frequency of visits but over a more limited period of time). We also tested whether rank order of butterfly species with respect to occurrence rate (proportion of sites occupied) would be reflected accurately in temporal subsamples. Temporal subsampling did not lead to erroneous inferences about the relative importance of six vegetation-based predictor variables on the species richness of butterflies. Regardless of the temporal scale of sampling, the species composition of butterflies was more similar in sites with similar species composition of plants. The rank order of occurrence of butterfly species in the temporal subsamples was highly correlated with the rank order of species occurrence in the full data set. Thus, similar inferences about associations between vegetation and butterflies and about relative occurrence rates of individual species of butterflies could be obtained by less intensive or extensive temporal sampling. If compromises between temporal intensity and extent of sampling must be made, our results suggest that maximizing temporal extent will better capture variation in biotic interactions and species occurrence.