Abstract
SummaryVegetation condition metrics are often used as a surrogate of biodiversity to support management decisions, conservation regulations and biodiversity markets. Vegetation condition metrics, which aggregate multiple attributes, are often criticised for simplifying the complexity of biodiversity. A particular challenge is substitution when high‐scoring attributes compensate for those with low scores (e.g. high vegetation cover compensating for low growth from diversity). The geometric mean is often suggested for aggregation to reduce these effects. In New South Wales, Australia, the Vegetation Integrity metric, calculated as the geometric mean of Composition, Structure and Function sub‐indices, measures the losses and gains in biodiversity values within the NSW Biodiversity Offsets Scheme. However, concern has been raised that Vegetation Integrity underestimates conservation values of derived native grasslands when the Function sub‐index (primarily tree‐related attributes) approaches zero. We explore this issue using two datasets and compare the current Vegetation Integrity metric with aggregation using the arithmetic mean, adopting a minimum value of 10/100 for the Function sub‐index and use of a grassland benchmark for derived native grasslands. Our evaluation draws on a large‐scale expert elicitation of conservation values in Critically Endangered Box‐gum Grassy Woodlands and 4018 Vegetation Integrity estimates undertaken during development assessments. We find that Vegetation Integrity underestimates conservation values of derived native grasslands and that the problem is widespread. Although evidence most strongly supports aggregation using the arithmetic mean, this change could be disruptive to the Biodiversity Offsets Scheme. Alternatively, a sub‐index minimum of 10/100 eliminates underestimation of derived native grasslands without substantial impacts in other circumstances. We found little evidence to support the use of a grassland benchmark, which tended to overestimate conservation values. This study highlights the need for sufficient flexibility in biodiversity policies and regulations to accommodate ongoing metric evaluation and revision to support more robust biodiversity outcomes.