Quantifying species biases among multidata sources on illegal wildlife trade and its implications for conservation.

Unsustainable wildlife consumption and illegal wildlife trade (IWT) threaten biodiversity worldwide. Although publicly accessible data sets are increasingly used to generate insights into IWT, little is known about their potential bias. We compared three typical and temporally corresponding data sets (4204 court verdicts, 926 seizure news reports, and 219 bird market surveys) on traded birds native to China and evaluated their possible species biases. Specifically, we evaluated bias and completeness of sampling for species richness, phylogeny, conservation status, spatial distribution, and life-history characteristics among the three data sets when determining patterns of illegal trade. Court verdicts contained the largest species richness. In bird market surveys and seizure news reports, phylogenetic clustering was greater than that in court verdicts, where songbird species (i.e., Passeriformes) were detected in higher proportions in market surveys. The seizure news data set contained the highest proportion of species of high conservation priority but the lowest species coverage. Across the country, all data sets consistently reported relatively high species richness in south and southwest regions, but markets revealed a northern geographic bias. The species composition in court verdicts and markets also exhibited distinct geographical patterns. There was significant ecological trait bias when we modeled whether a bird species is traded in the market. Our regression model suggested that species with small body masses, large geographical ranges, and a preference for anthropogenic habitats and those that are not nationally protected were more likely to be traded illegally. The species biases we found emphasize the need to know the constraints of each data set so that they can optimally inform strategies to combat IWT.