Data Package Metadata   View Summary

Survey design. The survey requested experts' estimates of global biodiversity loss and its consequences for ecosystem functioning and nature's contributions to people. It aimed to extend previous related research (Sala, et al. 2000, Schlapfer, et al. 1999) by including more experts and by being more quantitative, in terms of estimates and their uncertainty. An international team of biodiversity experts collaboratively developed the survey questions to ensure that they were clear and accessible to respondents worldwide. Questions were iteratively revised to increase clarity and avoid ambiguity. To avoid constraining responses and the functional forms of relationships between biodiversity loss and its consequences, quantitative questions allowed responses of any integer from 0 to 100%. To help quantify the large uncertainties around estimates, respondents were asked to provide lower and upper bounds for each estimate provided. The survey was designed in collaboration with survey professionals from the Minnesota Center for Survey Research in the Office of Measurement Services at the University of Minnesota. The survey was developed and disseminated using Qualtrics Survey Software. This project was reviewed and approved by the Institutional Review Board at the University of Minnesota. Respondents were instructed to (1) skip questions where they were unable to provide estimates; (2) indicate levels of uncertainty by a range for estimates; (3) use knowledge of the scientific literature, but provide current best estimates, rather than recall previously published estimates; and (4) hover over or click on underlined text to see clarifications. Definitions of terms and other clarifications were provided in hover over text (WebTable 1). Terminology was made as consistent as possible with that used in a recent global report of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (Díaz, et al. 2019).

Survey content. The first section of the survey asked respondents to identify areas of expertise. Display logic was used to ask follow-up questions. That is, experts first indicated the type of ecosystem they know best: terrestrial, marine, or freshwater. Depending on this response, they were then asked which terrestrial biome (Olson, et al. 2001), marine realm (Spalding, et al. 2007, Spalding, et al. 2012), or freshwater major habitat type (Abell, et al. 2008) they know best. Experts were also asked to indicate the direct driver(s) of biodiversity loss they know best. Experts were required to choose only one taxonomic group, and only one of the 18 major groups of nature’s contributions to people (Díaz, et al. 2018) or one of two ecosystem functions (i.e., primary productivity or decomposition), they know best. The second section of the survey requested estimates of global biodiversity loss and its consequences for ecosystem functioning or nature’s contributions to people. Global biodiversity loss was defined as the threatening or extinction of species, following the IUCN’s definitions for threatened species. Respondents were reminded to consider only the type of ecosystem, taxa, and function or contribution that they know best, as indicated in the first section of the survey. Experts were also asked to indicate uncertainty for estimates by providing the likely lower and upper bounds for the mean estimate. Hover over text clarified that sources of uncertainty may include those arising from a low quantity or quality of evidence, low level of agreement of evidence, and any gaps in personal knowledge of the literature. Experts who indicated that global biodiversity loss was expected to either decrease or increase ecosystem functioning or nature’s contributions to people were asked to provide estimates and uncertainty for specified amounts of global biodiversity loss. The third section of the survey asked respondents to rank direct and indirect drivers from those that currently threaten biodiversity most to least, and asked whether synergistic, antagonistic, or no interactions were expected between co-occurring drivers and between specific pairs of direct drivers. Experts who indicated expertise in land/sea use change and/or climate change in the first section were asked to estimate the magnitudes of global biodiversity loss expected to occur under specified magnitudes of changes in these direct drivers. Respondents were again reminded to consider only the type of ecosystem, taxa, and function or contribution that they know best. The final section of the survey asked questions about respondent demographics and perspectives on conservation. Demographic questions included the highest level of related education, years of related work experience, place of work, country of residence (which were grouped into regions and income groups according to the World Bank’s classification), and self-identified gender. Respondents were asked to place their perspectives along non-mutually exclusive continuums for values related to biodiversity conservation, indicate the spatial scales at which they conduct the majority of their biodiversity research, recommend allocation of conservation budgets, and indicate levels of global biodiversity loss that they would consider to be a planetary boundary with respect to sustaining the integrity of nature or its contributions to people.

Survey dissemination. We used a repeatable search to identify biodiversity experts as corresponding authors of articles and reviews published in scientific journals over the last decade on the topic of biodiversity. On the 1st of January 2020, we searched the Web of Science Core Collection for articles or reviews published in English on the topic of biodiversity during the decade from 2010 to 2019. Specifically, we used the following search terms: (WC=(Biodiversity & Conservation OR Ecology OR Multidisciplinary Sciences) AND TS=biodiversity) AND LANGUAGE: (English) AND DOCUMENT TYPES: (Article OR Review). This search returned 37,779 results. We downloaded the full record for each of these publications, which included the contact information for corresponding authors. We then used the ex_email function in the qdapRegex package in R to extract email addresses for all corresponding authors. This returned 50,950 email addresses. Unsurprisingly, many of these email addresses were duplicates, given that many biodiversity experts have been a corresponding author on more than one paper during the past decade. After removing duplicates, there were 35,970 unique email addresses. On the 8th of January 2020, an invitation to complete the survey was sent to each of these email addresses. Two reminders to complete the survey were sent on the 15th and 27th of January. Subsequently, to encourage additional responses globally, the survey invitation was translated by coauthors into Spanish, Japanese, and Chinese, and requests for distribution were made to the Ecological Societies of Argentina, Chile, China, Japan, Mexico, and the UK, and to Institutes of the Chinese Academy of Sciences. Only 6% of responses came from this additional dissemination effort. The survey was closed on the 25th of March 2020.

Survey responses. We received 3,331 responses from biodiversity experts who live in 113 countries and research biodiversity in 187 countries, including all terrestrial biomes and major habitats in freshwater and marine ecosystems. Respondents were allowed to retake the survey if they wished to provide responses for different taxa or habitats. However, only 1.7% of email addresses submitted multiple responses, and only 3.9% of IP addresses submitted multiple responses, some of which may have been responses from different experts. Only 0.7% of responses came from duplicate combinations of email and IP addresses. Due to small sample sizes, archaea (N = 8) were grouped with bacteria and seaweeds (N = 46) were grouped with protists. The demographics of survey responses matched some of the previously reported demographic biases in authors of the related scientific literature (Maas, et al. 2021, Tydecks, et al. 2018). In particular, most of our survey responses came from experts who identify as men or who live in Europe or North America. We are unable to quantify the total number of experts invited to complete the survey or the percentage of experts who responded. Our methods used de-identified email addresses of corresponding authors, which are not associated with names of individuals in the exported Web of Science records. Note that many experts have used multiple email addresses over the last decade as they have advanced their careers or changed workplaces for other reasons. Thus, the number of biodiversity experts invited to complete this survey is likely substantially lower than the number of unique email addresses (35,970) invited to complete the survey. An unknown number of the non-responses are due to inactive emails formerly used by experts who may have responded to the survey using a current email address. The median time for survey completion was 21 minutes, with 30% of responses taking more than 30 minutes and 10% of responses taking less than 10 minutes. More than 88% of the responses finished the survey. The remaining 12% of responses ranged from 23-98% complete.