The quality and quantity of stream habitat can have profound impacts on the distribution and abundance of aquatic species. Stream networks, however, are dynamic in their response to naturaland human-induced disturbance regimes, which results in spatially explicit patterns of temporal variability. Quantifying spatial patterns in habitat (temporal) variability across different sites and identifying those factors associated with different levels of variability are important steps for stream habitat assessments. We evaluated the temporal variability in stream habitat over a 9-year period for 47headwater streams of the interior Columbia River basin. We used repeat-measures analyses to calculate temporal variability as root mean square error for six habitat attributes at each site. Multiple linear regression analyses with root mean square error as the response were then used to quantify which landscape, climate, and disturbance attributes were associated with different levels of temporal variability among habitat attributes. Our results indicated a considerable range of temporal variability in physical stream attributes across sites and an almost fourfold difference in the overall variability at sites. Landscape factors affecting stream power, land management activities, and recent fire regimes were all factors associated with the different levels of temporal variability across sites; surprisingly, we found little association with the different climatic attributes considered herein. The observed differences in temporal variability across sites suggest that a “one-size-fits-all” approach to monitoring stream habitat in response to restoration and management activities may be misleading, particularly in terms of sampling intensity, required resources, and statistical power; thus, in situ measures of temporal variability may be required for accurate assessments of statistical power. The important influence of physical habitat on the distribution and abundance of stream biota is well documented (Southwood 1977; Minshall et al. 1983; Riley and Fausch 1995). Consequently, quantifying the status and trends of stream habitat is critical for understanding the factors that potentially limit populations (Nickelson and Lawson 1998), for quantifying restoration effectiveness (Bernhardt et al. 2005), for determining how land management activities impact stream ecosystems (Kershner et al. 2004b), and for directing future management *Corresponding author: email@example.com 1Present address: U.S. Geological Survey, Northern Rocky Mountain Science Center, 2327 University Way, Suite 2, Bozeman, Montana 59715, USA. Received July 21, 2010; accepted December 25, 2010 plans and restoration efforts (Burnett et al. 2007). However, robust evaluations of habitat status and trends can be difficult as multiple sources of variability can impede our ability to accurately evaluate the structure of stream habitat (Larsen et al. 2004). As such, identifying and minimizing sources of variability are important steps in the design of effective monitoring efforts and in generating expectations of the statistical power to detect changes (Urquhart et al. 1998; Larsen et al. 2001). 399 D o w n l o a d e d B y : [ U . S . G e o l o g i c a l S u r v e y L i b r a r y ] A t : 1 6 : 2 9 1 2 A p r i l 2 0 1 1 400 AL-CHOKHACHY ET AL. Much of the focus in habitat monitoring has addressed variability associated with sampling error, site-to-site differences, year effects, and site × year interactions. Numerous efforts have identified sampling error as a significant source of variability in habitat assessments (Kaufmann et al. 1999; Whitacre et al. 2007). Sampling error can vary considerably across stream habitat attributes, and selection of those habitat attributes (and protocols) that exhibit low levels of sampling error can reduce variability and improve the power to detect changes in physical habitat (Roper et al. 2010). Recent efforts have illustrated the need for repeated sampling of the same monitoring sites through time, as high site-to-site variability can limit inferences of change in habitat structure over relevant time frames (Urquhart et al. 1998; Larsen et al. 2004; Anlauf et al. 2011). Year effects (or other relevant time frames) are changes in habitat structure that result from yearly climatic patterns (e.g., pool scour during floods; Lisle and Hilton 1999) but do not necessarily result from the particular action of interest (e.g., channel restoration; Larsen et al. 2001). In general, year effects tend to be relatively insignificant across large spatial scales as climate patterns can vary within and across different ecoregions; however, year effects may be considerably higher for smaller regional analyses (Urquhart et al. 1998). Finally, site × year interactions can also result in substantial sources of variability as sites can respond differently to yearly climate patterns due to inherent site-specific characteristics (Larsen et al. 2001). Larsen et al. (2004) observed considerable site× year interactions, suggesting that sites may change differentially through time due to inherent differences in landscape characteristics (e.g., slope of watershed) and climate patterns. Despite rigorous assessments of stream monitoring designs and approaches (e.g., Kaufmann et al. 1999; Larsen et al. 2004), few efforts have evaluated how the temporal variability of stream habitat differs across sites. It is surprising that there have been relatively few assessments of the temporal variability of stream habitat, particularly given our understanding of the dynamic nature of stream networks (Giberson and Caissie 1998; Woodsmith et al. 2005). Yearly changes in habitat structure (i.e., temporal variability) can substantially affect our ability to accurately and precisely detect relevant changes in habitat status over time (Larsen et al. 2001). Thus, interactions between climate and geomorphic context suggest that even spatially proximate sites experiencing similar climate patterns may differ in their amounts of temporal variability due to inherent geomorphic characteristics (e.g., gradient). The ability to differentiate temporal variability from other aspects of residual error, such as sampling error (e.g., Larsen et al. 2004), will have important implications for sampling design at any given site (Gibbs et al. 1998). Therefore, understanding which landscape attributes, climatic patterns, and disturbances influence the temporal variability of habitat attributes across sites is an important step in identifying our expectations of monitoring (e.g., the power to detect change). The overarching goals of the present study were to quantify the levels of temporal variability across sites and to provide insight into the processes that cause streams to change over the short time scales associated with most monitoring efforts (i.e., 1–10 years; Marsh and Trenham 2008). To achieve these goals, we used a 9-year, spatially explicit data set to accomplish the following objectives. First, we estimated the total variance at sites and decomposed this variance into site, year, and residual components to better understand the relative importance of each component. Next, we quantified the temporal variability of different stream habitat attributes observed across sites to provide insight into how much this component varies. Finally, we evaluated which site-specific landscape, climate, and disturbance attributes were associated with higher levels of temporal variability in stream habitat across sites as a means to better understand how these factors may influence year-to-year changes in stream habitat.