COS 74-1 - Integrating local knowledge into regional assessment of habitat use by threatened species: Case study of the jaguar in Central America

Thursday, August 11, 2016: 8:00 AM
Floridian Blrm D, Ft Lauderdale Convention Center
Lisanne Petracca1,2, Jacqueline L. Frair2, Jonathan Cohen2, Ana Patricia Calderón2, Javier Carazo-Salazar3, Franklin Castañeda3, Cordelia Che3, Daniel Corrales-Gutierrez3, Sandra Hernandez-Potosme3, Luis Herrera3, Melva Olmos3, Sandy Pereira3, Howard Quigley3, Nathaniel Robinson1, Roberto Salom-Perez3, Yahaira Urbina3 and Hugh Robinson1, (1)Landscape Analysis Lab, Panthera, New York, NY, (2)Department of Environmental and Forest Biology, State University of New York College of Environmental Science and Forestry, Syracuse, NY, (3)Panthera, New York, NY
Background/Question/Methods

Surveying at large scales is critical for the conservation of wide-ranging species, yet survey methods often employ differing protocols at localized scales. We propose here the use of interviews with local people, coupled with a Bayesian occupancy model that accounts for false positives and incorporates random effects across space and time, to estimate species habitat use at a regional scale. The current conservation strategy of the jaguar (Panthera onca) is to maintain range-wide habitat connectivity, which necessitated a cost-effective, large-scale assessment to confirm species presence and identify areas of conservation concern. We collected interviews with local people in grid cells across 12 jaguar corridors of Central America to ascertain the presence of jaguars and four large-bodied prey species: white-lipped peccary (Tayassu pecari), collared peccary (Tayassu tajacu), red brocket deer (Mazama americana), and white-tailed deer (Odocoileus virginianus). We evaluated the performance of these data in estimating site-specific jaguar habitat use as a function of land cover, landscape productivity, prey richness, human disturbance, land protection status, and topography. We also tested hypotheses that probability of detection (p) would increase with interviewee effort (time per year spent in the sampling unit) and that probability of false detection (fp) would increase as a function of the number of interviews conducted in the sampling unit.

Results/Conclusions

We conducted 3,863 interviews with local people across 44,168 km2 in six Central American countries. We used Markov chain Monte Carlo (MCMC) simulation to determine that the main drivers of jaguar habitat use were percent agriculture and pasture (-), percent canopy cover (+), prey richness (+), distance from settlement (+), distance from protected area (-), and elevation (-). An interviewee’s ability to detect a jaguar increased with more time spent in the sampling unit. False positive rates ranged from 7.8-9.4% and increased with the number of interviews conducted in the unit. We show that detection varied across space and time, representing a source of heterogeneity that needs to be accounted for in future large-scale surveying and monitoring programs. Interviews targeting knowledgeable local people, when incorporated in a Bayesian occupancy framework, can serve as a robust, cost-effective means of surveying a threatened species over large scales and can help guide immediate conservation action.