ECOHYDROLOGY Ecohydrol. 2, 213–225 (2009) Published online 26 May 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/eco.62 Streamside herbaceous vegetation response to hydrologic restoration on the San Pedro River, Arizona Gabrielle L. Katz 1 * J. C. Stromberg 2 and M. W. Denslow 3 1 Department of Geography & Planning, Appalachian State University, Boone, North Carolina, USA 2 School of Life Sciences, Arizona State University, Tempe, Arizona, USA 3 Department of Biology, Appalachian State University, Boone, North Carolina, USA ABSTRACT Recovery of physical conditions and processes is increasingly emphasized in ecosystem restoration. We examined a restoration project aimed at recovery of groundwater levels and base flows on an undammed desert river. We sampled streamside plant communities and hydrology annually (2003–2008) at six restoration sites at two farms (Three Links Farm and H&E Farm) after groundwater pumping was curtailed for restoration purposes, and at six reference sites. Vascular plant cover was recorded, and species were classified into functional groups based on water needs and life history. Synthetic vegetation metrics and community composition were compared between restoration sites and perennial-flow reference sites representing target conditions. Perennial sites had higher cover, species richness, relative cover of hydric perennials and hydric annuals, and lower wetland indicator scores than non-perennial sites, but did not differ in relative cover of non-native species. Perennial sites had distinct species composition and high species constancy among years. Streamside vegetation was similar at Three Links Farm and perennial reference sites, indicating restoration success; streamside vegetation at H&E Farm differed from target conditions according to most measures. Hydrology is a key factor shaping desert streamside plant communities. However, single river sites may respond differently to hydrologic restoration. Contrasts in site response were likely due to differences in hydrogeomorphic context, exacerbated by drought. Understanding the constraints on restoration response is critical for setting realistic restoration goals and anticipating time frames of ecosystem change. Copyright  2009 John Wiley & Sons, Ltd. KEY WORDS desert; groundwater; hydrology; riparian; wetlands Received 24 February 2009; Accepted 15 April 2009 INTRODUCTION Recovery of physical conditions and processes has been increasingly emphasized in ecosystem restoration. This approach assumes that once the physical environmental conditions thought to underpin ecosystem characteristics are restored, many of the biotic ecosystem components will recover spontaneously without human intervention. It also relies upon speculation that physical and biolog- ical ecosystem changes are reversible and will follow predictable trajectories over relatively short time scales (i.e. ‘the rubber band model of stress and recovery’, Sarr, 2002; ‘successional recovery’, Suding et al., 2004). This strategy has yielded positive results in many set- tings, but for a variety of reasons expected results may not always be achieved. For example, legacies of past land use may over-ride the effects of changed manage- ment practices (e.g. Harding et al., 1998; van Dijk et al., 2007), or insufficient reduction of ecosystem stresses may cause relapse to degraded conditions following restora- tion (Sondergaard et al., 2007). Restoration of specific historic conditions may be essentially unachievable in systems that have shifted to a new environmental regime. New species (whether native or non-native), absence of * Correspondence to: Gabrielle L. Katz, Department of Geography & Planning, Appalachian State University, Boone, North Carolina, USA. E-mail: katzgl@appstate.edu seed sources in a changed landscape, fluvial-geomorphic changes, substrate alteration and climate change might severely inhibit restoration to past conditions (Lindig- Cisneros et al., 2003; Suding et al., 2004; Hobbs, 2007; Klimkowska et al., 2007). Understanding the causes of diverse or unpredicted outcomes is a key challenge in restoration research. For both non-riparian and riparian wetlands, the hydro- logical regime is a key physical factor underlying ecosys- tem conditions. The hydrological support for depressional wetlands is typically groundwater inflow, and floodplain meadows similarly depend on shallow alluvial ground- water levels (Wright and Chambers, 2002; Dwire et al., 2004; van Diggelen et al., 2006). Thus groundwater level recovery is usually considered a pre-requisite for restoration of drained wetlands. In riparian meadow sys- tems along incised mountain streams in the northern Sierra Nevada, California, USA, restoration projects have manipulated fluvial geomorphology (i.e. filled incised channels and constructed new smaller channels) as a means to raise riparian water tables (Loheide and Gore- lick, 2007; Hammersmark et al., 2008). In Europe, where historic drainage is a primary cause of changes to wet meadow vegetation, re-wetting (raising riparian water tables) is a common approach to restoration and has been most successful when combined with topsoil removal and diaspore transfer (Klimkowska et al., 2007; van Dijk Copyright  2009 John Wiley & Sons, Ltd.