• long-term vegetation data;
  • mixed effects models;
  • non-equilibrium;
  • ordination;
  • plant–animal interactions;
  • riparian;
  • wetland


  • 1
    Spring-fed wetlands are nested within California's oak savanna–annual grassland, which is considered a non-equilibrium-type system because it shows little community-level response to changes in grazing intensity. This insensitivity to disturbance is thought to result from transient resource limitation and the annual life cycle tending to swamp the effects of defoliation. Because spring-fed wetlands receive relatively consistent and high inputs of water and nutrients, and they are dominated by perennial herbaceous vegetation, we hypothesized that these systems would respond to a grazing intensity gradient in a predictable manner, i.e. equilibrium dynamics would prevail.
  • 2
    We experimentally tested plant community responses of spring-fed wetlands to two levels of grazing intensity (light and moderate) and no grazing over 10 years. Wetland vegetation was tracked at two distinct geomorphological parts of the wetland system: spring heads, where emergent water formed marshy zones, and their resultant channelized creeks.
  • 3
    We used linear mixed effects models to estimate grazing intensity effects over time. A general result at both springs and creeks was that slope estimates of annual total herbaceous cover over time were negative under moderate grazing but positive under light grazing and grazing removal.
  • 4
    Diversity metrics were not affected by grazing treatments at springs. At creeks, Simpson's diversity index increased over time under moderate grazing. When averaged over 10 years, species evenness and Shannon's diversity index were greater under moderate grazing at creeks.
  • 5
    Species composition was highly variable from year to year at springs, with no separation of the first ordination axis (detrended correspondence analysis) amongst grazing treatments. In contrast, three relatively stable and distinct equilibria were evident for creeks. These results indicate that springs exhibited non-equilibrium dynamics while their creek counterparts, separated by several metres, behaved in a more equilibrium fashion, in relative terms.
  • 6
    Grazing removal significantly increased interannual variability in species composition at both geomorphological types, demonstrating that plant–animal interactions serve as the main control on community composition in these systems by reducing cover and promoting diversity and stability.
  • 7
    Synthesis and applications. Grazing managers in California's oak savanna–annual grassland have little control of species composition because the annual community is entrained by the weather. Our results show that first-order wetland–riparian communities nested within this annual grassland matrix can be manipulated via grazing intensity adjustments. However, marshy spring-fed wetland areas are less sensitive than channelized creeks to these manipulations and grazing removal will increase variability in species composition for both geomorphological types. Grazing management decisions are usually made at the landscape level based on the matrix vegetation, but nested ecosystems may respond differentially, requiring a more nuanced approach that includes site-specific information.