Not all data are equal: Influence of data type and amount in spatial conservation prioritisation

Decisions about land use significantly influence biodiversity globally. The field of spatial conservation prioritisation explores allocation of conservation effort, including for reserve network expansion, targeting habitat restoration, or minimising ecological impacts of development. Inevitably, the utility of such planning depends on the quantity and quality input data, including spatial information on biodiversity, threats, and cost of action. In this work we systematically develop understanding about the significance of these different data types in spatial conservation prioritisation. We clarify the common ways different data types enter an analysis, develop mathematical models to understand the effects of data in spatial prioritisation, and survey literature to establish typical quantities of different types of data used. We use Jackknife analysis to derive the expected change in site values, when a single new data layer is added to a prioritisation. We validate mathematical formulae for expected impacts using simulations. A survey of scientific literature reveals that typical spatial prioritisation analyses include hundreds of biodiversity feature layers (species, habitat types, ecosystem services), but the count of cost, threat or habitat condition layers is typically 0-5. Due to these differences, and the mathematical formulations commonly used to combine data types, the influence of a single cost, threat, or habitat condition data layer can be an order or two higher than the influence of a single biodiversity feature layer. In a classical cost-effectiveness formulation (benefits divided by costs, B/C) the influence of a single cost layer can even be as large as the joint influence of thousands of species distributions. We also clarify how changes in data impact site values and spatial priority rankings differently, with the latter being further influenced by data correlations, the spread of numeric values inside data layers and other data characteristics. For example, costs influence priorities significantly if cost is positively correlated with biodiversity, but the correlation is the other way around for biodiversity and habitat condition. This work helps conservation practitioners to direct efforts when collating data for spatial conservation planning. It also helps decision makers understand where to focus attention when interpreting conservation plans and their uncertainties.