Limited evidence of compositional convergence of restored vegetation with reference states after 20 years of livestock exclusion

Abstract

AbstractNative pastures and rangelands may become degraded by intensive pastoral use through clearing of woody components, loss of sensitive native species, invasion of opportunistic exotics and soil erosion, compaction and nutrient imbalance. Livestock exclusion is commonly a first management step to restore ecosystems degraded by overgrazing. However, few studies evaluate the outcomes of livestock exclusion on plant species composition and soil chemistry, while those that do are inconclusive due partly to highly variable responses and to inexplicit performance criteria. We adopted a conceptual model that characterised responses to fencing in terms of divergence from initial states and convergence with targeted states that represent restoration goals. We used the model to design an investigation of fencing effects on grassy woodlands within livestock production landscapes in subhumid southeastern Australia. First, we asked whether fencing initiated divergence from continually grazed vegetation and convergence with reference states with respect to plant species composition and soil properties. Secondly, we asked how responses depend on livestock exclusion time, degradation of the initial state, and soil conditions. Compositional trajectories based on a 17‐year chronosequence showed that livestock exclusion initiated some divergence from unfenced controls but convergence with reference states was limited, variable among sites and unrelated to time‐since‐livestock‐exclusion. Fencing initiated the development of novel communities comprising exotic taxa and a subset of native taxa from the reference species pool. Initial state offered limited explanation for the variability in compositional divergence from unfenced controls. Soil properties did not converge with reference states over time, and initial state did not explain the degree divergence of fenced from unfenced plots. The findings suggest site selection is crucial to successful restoration and reinforce the need to consider supplementary management actions, such as planting, scalping, direct seeding and weed control, in addition to livestock grazing exclusion, to restore woodlands towards desired states.