Plant traits as predictors of performance in ecological restoration.
There are few studies of the performance of species in restored vegetation communities. Here we report the results of a meta-analysis of 25 experiments concerned with species-rich grassland restoration on ex-arable land and agriculturally improved grasslands situated at a wide range of locations throughout lowland Britain. Differences in species' performance were related to 38 physiological and morphological traits. An experiment-adjusted performance index was calculated for each of the 58 species (13 grasses and 45 forbs). The performance index was calculated for the first 4 years after establishment together with a temporal trend. Individual species showed large differences in performance indices. However, grasses consistently out-performed forbs. We examined the linkage between species' performance and traits according to four non-exclusive hypotheses. The ability to establish and persist in restored vegetation communities requires: (H1) good gap colonization ability; (H2) strong competitive capability; and (H3) ability to undergo vegetative regeneration. (H4) Successful species are generalists associated with fertile habitats. Trait analyses supported all four hypotheses. Within the forbs, good establishment in the first year was linked to traits determining colonization ability: ruderality, percentage germination of seeds and autumn germination. However, traits linked to competitive ability, vegetative growth and seed bank persistence became increasingly important determinants of success with time. Species with generalist habitat requirements, and especially those associated with fertile soils, performed increasingly well with time. This reflects the development of a closed vegetation in which the ability to grow vigorously and out-compete other established plants is important. Stress-tolerators, habitat specialists and species of infertile habitats performed badly. This may reflect high residual fertility in restored grasslands and particular niche requirements of these species. This may be a problem as grassland restoration often targets communities characterized by species with these traits and many are food plants of invertebrates of conservation value. There were few significant correlations between the performance of the grasses and traits reflecting their overall good performance in comparison with the forbs. This study has important implications for practical restoration programmes and policies. Efficiency might be increased by introducing only species with good performance, but this would lead to uniformity among restored grasslands and would diminish the benefits of habitat restoration for national and regional biodiversity. Synthesis and applications. Future work should focus on practical methods to increase the successful establishment of the poor performing but desirable species, by (i) targeting restoration to low fertility soils, (ii) changing the abiotic environment or (iii) the 'phased introduction' of species several years after restoration, when both the plant community is more stable and the environmental conditions are more favourable for establishment.