Assumptions and limitations
In our modelling, we assume the human pressures (land use change and intensification, human population growth and landscape simplification) have caused the differences we see in biodiversity within each study. However, these are not the only drivers of biodiversity change.
We assume that the species at sites with minimally disturbed plants are similar to species in a pristine area as truly untouched environments are rare.
We also assume that all species found in these minimally disturbed sites are naturally present. But this is not always true as some of these species may be invasive. Usually, the PREDICTS database can't identify which species present are native or invasive.
In our modelling, we only consider landscape structure and landscape history in very basic ways, meaning that we won't capture all aspects of how these factors reshape biodiversity. We can try to account for additional pressures affecting sites by including more variables in the models.
Because we lack representative long-term data, we do not have true baseline sites with which we can make biodiversity comparisons.
While our data are more geographically representative than other biodiversity databases, there are still some geographical gaps in the data used to calculate the BII.
Although the mapping of land use change is continually advancing, there are still limitations. Filtering data within the Biodiversity Trends Explorer over broad areas and larger time periods will be more reliable than filtering for smaller areas and time periods.
In any model there is statistical and structural uncertainty. For example, when categorising each land use type there may be errors.
In the Biodiversity Trends Explorer, each country or region's BII value is the average across all its land with every square kilometre being equally important.
An alternative approach would be to weight each square kilometre by how ecologically active it is, so by its net primary productivity or how much life there is present and being produced. This would highlight that biodiversity intactness is more important in productive places such as rainforests than areas of low productivity such as deserts. Future releases of the BII will include weighted averages like this, as well as the area-weighted average in the current version.
Indicator: The Biodiversity Intactness Index, accessed through the Biodiversity Trends Explorer
Data set: Available through the NHM Data Portal
Modelling framework: Available through GitHub
Related Museum project: PREDICTS
Project and research leads: Professor Andy Purvis and Dr Adriana De Palma
Data last updated: October 2021