Ecological context shapes the interpretation of biodiversity intactness metrics
Developing biodiversity metrics that are both operational for decision-making and ecologically meaningful remains challenging. Mean Species Abundance (MSA) is widely used to quantify biodiversity intactness in applications including biodiversity scenario modeling, spatial assessment, policy and planning, and life cycle assessment (LCA). MSA compares species abundance under disturbed conditions with their abundance under a reference condition, but the pressure-impact relationships used to calculate it are generalized across studies and may not adequately represent variation among ecological contexts or taxonomic groups. This is particularly relevant when MSA is transferred across regions or used to support comparisons and environmental decision-making. We compiled biodiversity data from 7,064 forest fragments containing 21,761 species of arthropods, amphibians, birds, mammals, squamates, and plants across 11 biomes worldwide. We evaluated how MSA relates to broader ecological outcomes, including local and regional species richness and ecosystem productivity, and whether these relationships vary across ecological contexts (vegetation condition, landscape structure, and biome) and taxonomic groups. MSA was positively associated with local species richness and ecosystem productivity across all sites, but these associations weakened, disappeared, or reversed depending on ecological context and taxonomic group. For example, the MSA-productivity association held in only five of 11 biomes, and the MSA-regional richness association was negative in temperate conifer and boreal forests but positive in four other biomes. We propose a simple decision guide linking assessment goals, contextual evaluation, and interpretation to help identify when MSA can be used directly, when ecological context should guide its interpretation, and when complementary ecological information is needed. These findings may extend to other biodiversity metrics that compare impacted conditions with reference states and apply generalized impact relationships across ecological contexts and taxonomic groups.