Global Cumulative Human Impacts to Marine Ecosystems

A comprehensive list of 38 categories of anthropogenic drivers of change in marine ecosystems developed through expert workshops (S1) was assessed for data availability on a global scale. We intentionally did not further subdivide these categories into unique drivers (e.g. each specific type of pollutant) as this would lead to over-emphasis of certain activities when impacts are summed. We limited our analyses to anthropogenic drivers with pre-existing global coverage or those for which we could assemble or develop global coverage. Although many regional-scale data and data with a global scope but incomplete coverage exist for a variety of specific human activities, inclusion of these data would bias global comparisons and so were excluded. Anthropogenic drivers that could not be included are, among others: hypoxic zones, coastal engineering (piers, rock walls, etc.), non-cargo shipping (ferries, cruise ships, etc), aquaculture, disease, recreational fishing, changes in sedimentation and freshwater input, and tourism. For all of these drivers data exist for one or more regions, but none have full global coverage. We also assessed availability of global data for 23 different marine ecosystems and were granted access to or developed spatial data for 20 of these. Data that were included in our analyses are listed in Table S1 and described in detail below. Our cumulative impact model follows a 4-step process (Fig. S1). We first assembled the global data for each anthropogenic driver (Di) and each ecosystem (Ej). Second, we log[X+1]-transformed and rescaled between 0-1 each driver layer to put them on a single, unitless scale that allows direct comparison, and converted ecosystem data into 1 km2 presence/absence layers. Third, for each 1 km2 cell of ocean we multiplied each driver layer with each ecosystem layer to create driver-by-ecosystem combinations, and then multiplied these combinations by the appropriate weighting variable (uij) These weighting variables come from an expert survey that assessed the vulnerability of each ecosystem to each driver on the basis of 5 ecological traits (S1). The weighting values (Table S2) represent the relative impact of an anthropogenic driver on an ecosystem within a given cell when both exist in that cell, and do not represent the relative global 1 impact of a driver or the overall status of an ecosystem. The sum of these weighted driver-by-ecosystem combinations then represents the relative cumulative impact of human activities on all ecosystems in a particular 1 km2 cell. Finally, to provide ecological meaning to these relative cumulative impact scores, we used empirical data on the condition of ecosystems to groundtruth our scores. The sensitivity of our results to key steps in this process and further details on the groundtruthing method are provided below. All open-access data and analytical code used in this project can be downloaded at http://www.nceas.ucsb.edu/GlobalMarine