Inferring the absence of a species: A case study of snakes

Though the presence of a species can be unequivocally confirmed, its absence can only be inferred with a degree of probability. I used a model to calculate the minimum number of unsuccessful visits to a site that are necessary to assume that a species is absent. The model requires the probability of detection of the species per visit to be known. This probability may vary depending on habitat, year, season, the area surveyed, the population size of the species, and the observer. I studied 3 European snake species-asp viper (Vipera aspis), smooth snake (Coronella austriaca), and grass snake (Natrix natrix)-over a 5-yr (1994-1998) interval, and made 645 visits to 87 sites during their activity periods. I used a generalized logistic regression approach with random effects for years and sites to (1) estimate the probability of detection of these species from sites known to be occupied, (2) test factors affecting it, and (3) compute the minimum number of times that a site must be visited to infer the absence of the particular species. Probability of detection for all species was heavily influenced by an index of population size. For V. aspis, probability of detection increased from 0.23 to 0.50 and 0.70 in small, medium, and large populations, respectively. Similarly, probability of detection increased from 0.09 to 0.45 and 0.56 in small, medium, and large populations of C. austriaca, respectively, and from 0.11 in small to 0.25 in medium and large populations of N. natrix. Probability of detection also varied across months for all 3 species, among habitat types (C. austriaca only), and from year to year (N. natrix only). Sites with unknown occupancy status conservatively may be assumed to be occupied by small populations. I calculated that such sites need to be surveyed 12, 34, and 26 times for V. aspis, C. austriaca, and N. natrix, respectively, before assuming with 95% probability that the site is unoccupied. These results suggest that some species may be more wide-spread than thought. However, to ascertain the presence of such species at a site, search efforts need to be intensive.