The author uses source-sink population models to test the ability to detect grizzly bear population declines as a result of habitat degradation.

Interpretation and Results
"The important point for conversation that emerges from this simple model is that very small amounts of habitat degradation can lead to dramatic changes in population growth rate, depending upon the severity of degradation and the ease or rate of movement between degraded and undisturbed habitat areas. For example, the additional conversion of 15% of an area from good habitat to bad can shift a population from stability to a 4% annual decline This result raises serious questions about our ability to predict the consequences of habitat degradation Thus, even what is seen as very mild and incremental degradation (e.g., the building of roads or selective forestry) must be viewed with caution and with knowledge that severe consequences could result from incremental changes in the proportion of good and bad habitat. Although this concept is not new, my results quantify the risks posed by degradation and emphasize that arguably insignificant changes may have serious results." (1373-1374)

Predicting Population Response to Human Encroachment
"[There is a] time lag between the crucial point in degradation from which population decline will begin and when degradation effects might first be detected. For the first four scenarios simulated this time lag ranges from 8 to 13 years. By the time population decline would be detectable, the remaining amount of good habitat - even if no more degradation occurred past that point - would result in annual declines of from 3 to 6%. Thus, the general conclusion is that census data may provide extremely poor measures of population safety or health under conditions of ongoing habitat degradation. As Taylor and Gerrodette (1993) forcefully argue, statistical problems will often make detection of population declines from census data very unlikely. My results extend their argument by showing that by the time declines are detected, it will probably be too late to prevent critical amounts of habitat degradation from occurring." (1377)

Discussion
"This exercise in simple modeling provides several insights into the phenomenon of habitat degradation in general and the current management of grizzly bears in particular. First, habitat degradation can have highly nonlinear effects on population growth rates, with small amounts of degradation leading to large decreases in overall population growth. Thus to assume the safety of incremental degradation is not possible, very small losses could result in rapid declines of previously stable populations.
Second, our ability to detect crucial degrees of habitat degradation are imperfect at bestThus, an assurance of even rigorous population monitoring while degradation continues is not a justifiable substitute for an a priori analysis of the probable consequences of development.
Third, the models point out that movement rates and habitat-specific demographic rates are crucial variables to understand and measure if we are to predict the critical amounts of degradation that a population can tolerate. These data are rarely available; the model therefore suggests a very conservative approach in assessing the dangers of habitat degradation. This result is particularly relevant to the management of the remaining grizzly bears in the lower 48 states
By demonstrating the great effects that slow degradation can have on threatened populations and the difficulty in accurately detecting these effects with simple population monitoring, my results bolster concerns about continuing habitat loss and degradation careful analyses to predict future viability should be carried out before allowing habitat degradation and loss to continue and certainly before populations can justifiably be declared recovered." (1378)