Main content area

Utility and limitations of climate‐matching approaches in detecting different types of spatial errors in biodiversity data

Simões, Marianna V. P., Peterson, A. Townsend
Insect conservation and diversity 2018 v.11 no.5 pp. 407-414
Chrysomelidae, biodiversity, biogeography, citizen scientists, data quality, databases, models, niches, quality control
The increase of digitally available primary biodiversity data has been a positive result of sharing initiatives in the natural history museum community and among citizen scientists. Owing to the heterogeneity of sources, however, limitations related to data quality control emerge, as incomplete and/or erroneous information at different stages of input must be overcome. To facilitate detection of spatial errors, species distribution modelling (SDM) has been suggested, but its efficiency in detection of different types of spatial errors has not been assessed. We investigate the utility of SDM‐based assessments in detection of two types of spatial errors found in large biodiversity databases, random errors versus errors of misidentification as congeneric taxa. We used available distributional data for five closely related species of the tortoise beetle genus Mesomphalia (Coleoptera, Chyrsomelidae, Cassidinae) to test the suitability values associated with simulated erroneous points mimicking the two error types. Overall, we observed that habitat suitability values associated with random outliers were lower than those for congeneric outliers, fitting expectations based on the idea of niche conservatism. Also, detecting outliers in small datasets is more challenging, whereas in larger datasets, the detection of random outliers should be more efficient. Our results indicate that SDM tools can be useful in detection of outliers more efficiently when erroneous points fall outside the ecological niche profile of the species, as in the case of random typographical errors, but not as effective with errors of misidentification. This paper explores a potential tool to promote better assessment of the quality of biodiversity data.