Is structure or function a better measure of the effects of water abstraction on ecosystem integrity

1. Assessments of flow abstractions in streams often focus on changes to biological communities and in-stream physical characteristics, with little consideration for changes in ecosystem functioning. It is unclear whether functional indicators of ecosystem health may be useful for assessing the impacts of reduced discharge on small streams. 2. We used weirs and diversions to reduce stream discharge by over 89% in three small New Zealand streams (11-84 L s⁻¹), ranging in water quality from pristine to moderately impaired. 3. We used both structural (benthic invertebrates) and functional (drifting invertebrates, leaf breakdown, coarse particulate organic matter (CPOM) retention and primary productivity) measures of ecosystem integrity to compare responses to water abstraction in before-after, control-impact designed experiments during summer 2005. 4. At the pristine site, the density of invertebrates, taxon richness, Macroinvertebrate Community Index (MCI), Quantitative MCI, percentage of Ephemeroptera, Plecoptera and Trichoptera individuals and percentage of filter-feeders decreased in response to reduced flows. Only taxon richness decreased at the mildly impaired stream, and reduced discharge had no effect on the invertebrate community at the stream with the lowest water quality. 5. We found that reduced discharge had little influence on the breakdown rate of willow leaves in mesh bags over 1 month. Primary productivity was also relatively insensitive to water abstraction. However, CPOM retention increased with decreased flows. Drift propensity of invertebrates increased at two sites but only within the first few days after flow reduction. 6. Structural measures of ecosystem integrity suggested that the impacts of water abstraction differed among streams of varying water quality, probably because of differences in the sensitivity of invertebrate assemblages in the three streams. In contrast, the three functional measures tested were generally less sensitive to water abstraction impacts, although understanding how stream ecosystems respond to water abstraction clearly requires that both are considered.