Analytical guidelines to increase the value of community science data: An example using eBird data to estimate species distributions

Abstract

Aim Ecological data collected by the general public are valuable for addressing a wide range of ecological research and conservation planning, and there has been a rapid increase in the scope and volume of data available. However, data from eBird or other large‐scale projects with volunteer observers typically present several challenges that can impede robust ecological inferences. These challenges include spatial bias, variation in effort and species reporting bias. Innovation We use the example of estimating species distributions with data from eBird, a community science or citizen science (CS) project. We estimate two widely used metrics of species distributions: encounter rate and occupancy probability. For each metric, we critically assess the impact of data processing steps that either degrade or refine the data used in the analyses. CS data density varies widely across the globe, so we also test whether differences in model performance are robust to sample size. Main conclusions Model performance improved when data processing and analytical methods addressed the challenges arising from CS data; however, the degree of improvement varied with species and data density. The largest gains we observed in model performance were achieved with 1) the use of complete checklists (where observers report all the species they detect and identify, allowing non‐detections to be inferred) and 2) the use of covariates describing variation in effort and detectability for each checklist. Occupancy models were more robust to a lack of complete checklists. Improvements in model performance with data refinement were more evident with larger sample sizes. In general, we found that the value of each refinement varied by situation and we encourage researchers to assess the benefits in other scenarios. These approaches will enable researchers to more effectively harness the vast ecological knowledge that exists within CS data for conservation and basic research.

Integrating citizen‐science and planned‐survey data improves species distribution estimates

“…Other studies develop models with more explicit descriptions of the complex variation of sampling effort that is characteristic of citizen-science datasets (e.g. August et al, 2020;Johnston et al, 2021). We opted for a more general approach that, for example, carries no information about individual observer behaviour.…”

Estimating the movements of terrestrial animal populations using broad-scale occurrence data

“…eBird ( www.ebird.org ) has been a leading example, capitalizing on already dedicated birding groups and hobbyists, and developing a platform that mimics the checklist format already popular among birdwatchers [ 36 ]. New analytical methods and procedures have been developed to leverage the information provided by eBird to generate reliable estimates of species occurrence [ 66 ]. Other datasets focus on different taxonomic groups and geographic regions, but are increasingly providing the quality, density, and frequency of human-observed data necessary to assess population-level movements (Table 1 ).…”

“…Prior to statistical analysis, occurrence data frequently require cleaning and processing. Data processing methods may differ among datasets and for distinct research questions, but general challenges include estimating occupancy from presence-only data [ 109 ], standardizing sampling intensity by subsampling observations (e.g., [ 67 , 110 ]), accounting for low detection probability of certain species or in certain time periods or habitats [ 102 ], the potential for false positive occurrences [ 96 ], and accounting for detection or sampling biases related to human behavior [ 66 ]. Clear guidelines or code to aid in processing the data may be available for some datasets, or may require more technical knowledge to navigate, especially for WSR, acoustic, or image data [ 89 ].…”

Habitat suitability models based on opportunistic citizen science data: Evaluating forecasts from alternative methods versus an individual‐based model

“…Thus, detection-only data are often analysed by pairing them with locations without records of a species or with a sample of locations in the study area (pseudo-absences or background, Phillips et al, 2009). Pseudoabsences or background data represent habitats that can be occupied or unoccupied, and this approach can lead to predicted species distributions that are less accurate compared to distributions based on high-quality non-detection data (Bradter et al, 2018;Johnston et al, 2021). Nevertheless, detection-only data are widespread in many platforms such as GBIF, and SSOS until 2018, and are often the only information available for many study systems.…”

“…Point process models integrate opportunistic and systematic data, and explicitly model a possible sample selection bias (Fithian et al, 2015). In contrast to the previous three methods, occupancy models use repeated sampling to account for imperfect detection of species (MacKenzie et al, 2003) and can reduce problems due to uneven survey effort in opportunistic data (Johnston et al, 2021;. In our previous work, we established that these four methods with models based on opportunistic data could successfully produce species distributions that were similar to results from systematic monitoring data (Bradter et al, 2018).…”