Modelling detection probabilities to evaluate management and control tools for an invasive species
- 29 January 2010
- journal article
- Published by Wiley in Journal of Applied Ecology
- Vol. 47 (1) , 106-113
Abstract
Summary: 1. For most ecologists, detection probability (p) is a nuisance variable that must be modelled to estimate the state variable of interest (i.e. survival, abundance, or occupancy). However, in the realm of invasive species control, the rate of detection and removal is the rate‐limiting step for management of this pervasive environmental problem. 2. For strategic planning of an eradication (removal of every individual), one must identify the least likely individual to be removed, and determine the probability of removing it. 3. To evaluate visual searching as a control tool for populations of the invasive brown treesnake Boiga irregularis, we designed a mark–recapture study to evaluate detection probability as a function of time, gender, size, body condition, recent detection history, residency status, searcher team and environmental covariates. 4. We evaluated these factors using 654 captures resulting from visual detections of 117 snakes residing in a 5‐ha semi‐forested enclosure on Guam, fenced to prevent immigration and emigration of snakes but not their prey. Visual detection probability was low overall ( = 0·07 per occasion) but reached 0·18 under optimal circumstances. 5. Our results supported sex‐specific differences in detectability that were a quadratic function of size, with both small and large females having lower detection probabilities than males of those sizes. There was strong evidence for individual periodic changes in detectability of a few days duration, roughly doubling detection probability (comparing peak to non‐elevated detections). Snakes in poor body condition had estimated mean detection probabilities greater than snakes with high body condition. Search teams with high average detection rates exhibited detection probabilities about twice that of search teams with low average detection rates. Surveys conducted with bright moonlight and strong wind gusts exhibited moderately decreased probabilities of detecting snakes. 6. Synthesis and applications. By emphasizing and modelling detection probabilities, we now know: (i) that eradication of this species by searching is possible, (ii) how much searching effort would be required, (iii) under what environmental conditions searching would be most efficient, and (iv) several factors that are likely to modulate this quantification when searching is applied to new areas. The same approach can be use for evaluation of any control technology or population monitoring programme.Keywords
This publication has 25 references indexed in Scilit:
- Microhabitat Use by Brown Treesnakes (Boiga irregularis): Effects of Moonlight and PreyJournal of Herpetology, 2008
- Response of Brown Treesnakes to Reduction of Their Rodent PreyThe Journal of Wildlife Management, 2007
- Detection Probability and Sources of Variation in White‐Tailed Deer Spotlight SurveysThe Journal of Wildlife Management, 2007
- Characteristics of Brown Treesnakes Boiga irregularis removed from Guam's transportation networkPacific Conservation Biology, 2004
- ESTIMATING DETECTION PROBABILITY PARAMETERS FOR PLETHODON SALAMANDERS USING THE ROBUST CAPTURE–RECAPTURE DESIGNThe Journal of Wildlife Management, 2004
- VARIABILITY IN GRASSLAND BIRD COUNTS RELATED TO OBSERVER DIFFERENCES AND SPECIES DETECTION RATESOrnithology, 2003
- THE ROLE OF INTRODUCED SPECIES IN THE DEGRADATION OF ISLAND ECOSYSTEMS: A Case History of GuamAnnual Review of Ecology and Systematics, 1998
- The Effect of Multiple Search Items and Item Abundance on the Efficiency of Human SearchersJournal of Herpetology, 1998
- Evidence that humidity influences snake activity patterns: a field study of the Malayan pit viper Calloselasma rhodostomaEcography, 1998
- On the statistical analysis of capture experimentsBiometrika, 1989