The Fishes of Texas Project - a Regional Museum-Based Species Occurrence Database and Its Applications

The Fishes of Texas Project (www.fishesoftexas.org) compiled Texas fish species occurrence records from 42 museum collections. All records were processed through a rigorous quality control and data normalization/standardization process to result in 124,415 records collected between 1851 and 2010 by 5,924 collectors. Nearly all Texas inland records were manually georeferenced with estimates of placement errors, resulting in 88,348 records from 7,868 unique localities. Another 8,460 records from the Gulf of Mexico and 18,923 inland records from neighboring Mexican and U.S. states have been partially processed. Georeferenced records were plotted and 4,107 species geographic outliers flagged as potential identification or location errors. The process is ongoing, but most flagged specimens, and often related original documentation, have been examined and identifications corrected or confirmed by trained staff. Relative inadequacy of prior knowledge of the state’s fish biodiversity, and the value of vouchering collections and compiling and normalizing all data, were quickly demonstrated by discovery via the database of 31 species occurrences in major river basins where they were previously not known to occur. The database also revealed strong indications of distribution changes over time, and the occurrence data were used to produce detailed and powerful species distribution models that have proven useful in diverse applications and greatly increase the value of the raw occurrence data. The models have now been used in exploring possible species’ responses to future climate scenarios and in systematic conservation planning exercises. Preliminary results from exploration of use of the models in statewide bioassessment efforts look very promising. We aspire to link this database to the mostly unvouchered, but sometimes very large, fish datasets compiled by state agencies. These data complement the museum data in many ways and we propose that vetting for possible identification errors by comparison to the verifiable museum specimen-based records can serve to improve overall data quality and future fish identifications by agency staff. Data in the freely available online database can be queried in diverse ways, mapped, and records downloaded. The records are also linked to a large set of high quality fish images, original field notes, specimen photos and other useful information, such as accounts of species' biology and ecology and digital identification keys. By allowing users to comment on records and upload images and field notes, we hope that crowdsourcing and citizen science will contribute to improvement of the overall quality of the data over time.