DUPED BY FALSE-NEGATIVE ERRORS: A CASE STUDY WITH THE MEDITERRANEAN GECKO

Understanding habitat associations and organismal activity patterns can help scientists and managers gain insight to the invasive potential of a species; however, false-negative errors are common in detecting species within an area. A false negative error often takes the form of a question: was the species absent or did it just go undetected? We investigated how the assumption of perfect detection influences interpretation of habitat associations and activity patterns of the Mediterranean Gecko, Hemidactylus turcicus (Reptilia: Gekkonidae), which has been introduced to the Southeastern U.S. We conducted nocturnal surveys in Starkville, Mississippi, USA, and detected the Mediterranean Gecko at 17 of 22 sites on at least one occasion. We found that models that do (Single-season Single-species Occupancy Model) and do not (Logistic Regression) account for imperfect detection had a 15% difference in estimates of occupancy and were not dissimilar in the significance of covariates. Inference from our Occupancy Model indicated that well-defined eaves, minutes after sunset, and pedestrian traffic all influence detection probability, but no covariates were associated with Mediterranean Gecko occupancy. In contrast, results from the Logistic Regression model indicated that well-defined eves were of significance to the presence of Mediterranean Gecko. Interpretations of habitat associations and activity patterns can be misleading when imperfect detection goes unaccounted. We hope that more herpetologists take approaches to account for imperfect detection, focusing on sampling and survey methods that can confidently assess the distributional status, habitat associations and activity patterns, and eradication effectiveness of invasive species.