U.S. Geological Survey Home Page USGS Patuxent Wildlife Research Center Science Meetings; dedicated to Chandler S. Robbins USGS Patuxent Wildlife Research Center; Science Meetings dedicated to Chandler S. Robbins Dedicated to Chan Robbins USGS Patuxent Wildlife Research Center; Science Meetings dedicated to Chandler S. Robbins Poster Abstracts from USGS Patuxent Wildlife Research Science Meetings, October, 2006
Patuxent Science Meeting 2006 Poster Abstract

Uncertainty in detection bias: Hidden management costs of index-

based population monitoring

Moore CT, Kendall WL

Managers of wildlife populations commonly rely on simple counts of detected animals in making

decisions regarding conservation, harvest, or control. The main appeal in using such indirect

counts is their low material expense compared to methods that estimate the undetected

population (and therefore the entire population). However, their correct use rests on the rarely-

tested but often-assumed premise that they proportionately reflect population size, that is, that

they constitute a population index. We investigated forest management for the endangered

Red-cockaded Woodpecker (Picoides borealis) and the Wood Thrush (Hylocichla mustelina) at

the Piedmont National Wildlife Refuge in central Georgia (USA). We derived optimal decision

policies for a joint species objective under each of two alternative models of Wood Thrush

population dynamics. We simulated the optimal policies under each model for three scenarios

of bias for observed Wood Thrush densities: (1) unbiasedness, (2) consistent negative bias (i.e.,

a valid population index), and (3) habitat-dependent negative bias. Differences in simulation

outcomes between biased and unbiased detection scenarios provided the expected loss in

resource objectives (here, forest habitat and birds) through the basing of decisions on biased

population counts. Under the models and objective function we used, expected losses were as

great as 11%. For applications such as endangered species management, such a degree of

loss may not be trivial and could be far greater under different model assumptions. Our analysis

demonstrates that costs of uncertainty about the relationship between the population and its

observation can be measured in units of the resource, costs which may offset apparent savings

achieved by collecting uncorrected population counts.

Friday, September 22, 2006



Return to Posters page