Quantifying population declines based on presence‐only records for red‐list assessments

Abstract

Accurate trend estimates are necessary for understanding which species are declining and which are most in need of conservation action. Imperfect species detection may result in unreliable trend estimates because this may lead to the overestimation of declines. Because many management decisions are based on population trend estimates, such biases could have severe consequences for conservation policy. We used an occupancy-modeling framework to estimate detectability and calculate nationwide population trends for 14 Swiss amphibian species both accounting for and ignoring imperfect detection. Through the application of International Union for Conservation of Nature Red List criteria to the different trend estimates, we assessed whether ignoring imperfect detection could affect conservation policy. Imperfect detection occurred for all species and detection varied substantially among species, which led to the overestimation of population declines when detectability was ignored. Consequently, accounting for imperfect detection lowered the red-list risk category for 5 of the 14 species assessed. We demonstrate that failing to consider species detectability can have serious consequences for species management and that occupancy modeling provides a flexible framework to account for observation bias and improve assessments of conservation status. A problem inherent to most historical records is that they contain presence-only data from which only relative declines can be estimated. A move toward the routine recording of nonobservation and absence data is essential if conservation practitioners are to move beyond this toward accurate population trend estimation. AbstractAccurate trend estimates are necessary for understanding which species are declining and which are most in need of conservation action. Imperfect species detection may result in unreliable trend estimates because this may lead to the overestimation of declines.As many management decisions are based on population trend estimates, such biases could have severe consequences for conservation policy. We used an occupancy modelling framework to estimate detectability and calculate nationwide population trends for 14 Swiss amphibian species both accounting for and ignoring imperfect detection. Through application of IUCN Red List criteria to the different trend estimates, we assessed whether ignoring imperfect detection could impact on conservation policy. Imperfect detection was seen in all species and detection varied substantially among species, which led to the overestimation of population declines when detectability was ignored. Consequently, accounting for imperfect detection lowered the red-list risk category for 5 of the 14 species assessed. We demonstrate that failing to consider species detectability can have serious consequences for species management, and that occupancy modelling provides a flexible framework to account for observation bias and improve assessments of conservation status. A problem inherent to most historical records is tha...

Quantifying data quality in a citizen science monitoring program: False negatives, false positives and occupancy trends

“…This pattern can be seen in our results on tree-frogs (H. arborea); this species is typically easily detectable from calls (Cruickshank et al, 2016;), yet our standard occupancy model (FNO model) produced detection estimates substantially lower than is typical for this species. However, the information used to estimate detection probabilities changes; now 6 surveys are used to calculate detection probability, and the estimated probability declines from 0.66 to 0.50 ( 3 6 ).…”

“…The per-survey true detection probability p 11 was assigned a typical Uniform(0,1) prior, therefore this specification ensured that, on average p 11 > p 10 ( Figure S2). This prior choice for patch colonization and local extinction rates is based on the knowledge that patch persistence (the complement to local extinction) is greater than patch colonization rates for the species included in this study throughout Switzerland (Cruickshank, Ozgul, Zumbach, & Schmidt, 2016). This prior choice for patch colonization and local extinction rates is based on the knowledge that patch persistence (the complement to local extinction) is greater than patch colonization rates for the species included in this study throughout Switzerland (Cruickshank, Ozgul, Zumbach, & Schmidt, 2016).…”

Antifungal treatment of wild amphibian populations caused a transient reduction in the prevalence of the fungal pathogen, Batrachochytrium dendrobatidis

“…This species has suffered Bd -driven population declines and extirpations and is red-listed in some European countries 27–29 . The tadpoles of this species overwinter so that in the summer there are often multiple cohorts of tadpoles present in the same pond.…”

Exploiting opportunistic observations to estimate changes in seasonal site use: an example with wetland birds

“…1 km2; van Strien et al 2013; or 100 km2; Kamp et al 2016) or selected habitat patches (Cruickshank et al 2016) and closure periods, often a breeding season of two months or more (Kendall et al 2013;van Strien et al 2013;Kamp et al 2016;Cruickshank et al 2016). In such an annual occupancy model occupancy is then defined as the proportion of occupied sites or grid squares at a landscape or regional scale during each season.…”