1. Recent studies have demonstrated that there is generally no unambiguous relationship between plant species composition and specific environmental conditions in rivers. Nevertheless, indices of environmental pressures based on macrophytes are flourishing, because of the requirements of the Water Framework Directive (WFD).
2. We first reviewed nine such indices against 13 criteria for bioindicators. Then, using data from France and England, we tested whether the IBMR (Macrophyte Biological Index for Rivers) and LEAFPACS (predictions and classification system for macrophytes) methods could reliably indicate nutrient and hydromorphological pressures. Finally, we used an improved bootstrapping method to estimate accuracy.
3. Currently, most indices lack ecological meaning for a variety of reasons, including partial sampling (backwaters are excluded); reliance on list of taxa (there are identification difficulties) rather than structure and functions; correlation rather than causation; application within a limited biogeographical area; reliance on ‘expert’ judgement; high precision but poor accuracy; poorly defined reference conditions; lack of independent tests; and an inability to discriminate reliably between the target pressures of interest from confounding background variables.
4. IBMR was a far better indicator of pH (or HCO3-pCO2) than it was of soluble reactive phosphorus, SRP (or SRP-NH4). While there was a highly significant correlation between IBMR and SRP after removing the effect of pH, the relationship was weak (r2 = 0.08, n = 215, P < 0.001).
5. LEAFPACS is a multi-metric method summing up five individual indices, each compliant with the WFD. Its individual metrics were not better correlated with nutrient and hydromorphological pressures (with r2 < 0.1, n = 62, P < 0.05) than was the IBMR. The meaning of the overall metric is questionable.
6. There are problems in determining the precision of the indices, owing to uncertainties in recording, but they are less than the uncertainties in determining accuracy (because species optima and tolerances are sometimes poorly known).
7. Reliable information is needed to improve the state of our rivers. Macrophyte indices are able to detect statistically significant pressures from a large population of sites but cannot be applied at specific sites, as required by the WFD, owing to large uncertainties and low explanatory power. Typically, more than 90% of the variability in macrophyte indices is attributed to factors other than human pressure. The WFD would be better served by a simpler, holistic approach based on our current mechanistic understanding of river processes. These findings are likely to apply also to other taxonomic groups (macroinvertebrates, diatoms, fish) used in the assessment of purported ecological quality and to palaeolimnological measures of reference status.