Monitoring quality thresholds change the Living Planet Index through shifts in representation, not systematic bias in population trends

Quantifying biodiversity change using indicators is a crucial part of evaluating progress towards internationally agreed biodiversity targets. Many indicators rely on publicly available species monitoring data which is characterised by uneven geographic and taxonomic representation. The Living Planet Index (LPI) is a global biodiversity indicator which inherits these data biases, resulting in well-documented gaps in population data coverage. However, the relationship between taxonomic and geographic representation and the length and completeness of population abundance time series, a measure of data quality, has not been assessed. Given that LPI outcomes can be influenced by these factors, we explore how temporal data quality varies across taxa and regions and whether it relates to the direction of population abundance trends. We also examine relationships between temporal data quality and covariates: system, region, species range size and a proxy for species interest. We find that removing temporally lower quality data produced less negative aggregated trends but showed no relationship between temporal data quality and average population abundance change. Temporally lower quality data is concentrated in herptiles, particularly in Latin America and the Caribbean and is also related to: range size for amphibians; range size and region for mammals; and region for birds and reptiles. Whilst removing data with lower temporal quality changes the global LPI trend, it simultaneously increases existing geographic and taxonomic biases. Groups which are more likely to be in decline are also more likely to be removed from the LPI when temporal quality thresholds are applied. This highlights the inherent complexity of using species monitoring data in aggregated indicators and that both representation and temporal quality should be considered in data exclusion decisions.