wolves in 4 packs, we simulated the 5-year population growth under different adult mortality rates (10 to 50%). The rates of population growth
were negatively correlated with mortality and suggested that population
size stabilized with a mortality rate of about 35%, similar to the conclusion
of Fuller (1989) based on field studies. Additional model tests and sensitivity analyses are reported in Cochrane (2000).
The software for the wolf simulation model was written by and is available from the two senior authors. Versions of the source code were written
in FORTRAN and Visual BASIC. The applications were performed on an
IBM300PL and other personal computers. We have used this type of model
for other social carnivores, including the San Joaquin kit fox (Vulpes
macrotis mutica) in California (Haight et al. 2002) and the African lion
(Panthera leo) (Starfield et al. 1981). Population models with similar territorial and dispersal mechanisms were used for northern spotted owl (Strix
occidentalis caurina) recovery planning (Lande 1987; Lamberson et al.
1994).
2.3.2 Persistence of Wolves in
Human-Dominated Landscapes
Following protection under the Endangered Species Act in 1973, wolves
from northeastern Minnesota recolonized most of northern Minnesota and
parts of northern Wisconsin and northern Michigan (see Figure 2.1). The
landscape in this range was not wilderness but a mosaic of forest, agricultural, and developed land under a variety of public and private ownerships
(Mladenoff et al. 1995). Logging and agriculture had created extensive areas
of young forest that supported large populations of white-tailed deer, the
preferred prey of wolves in this region. Colonizing wolves first settled in
forested areas with few roads and little human settlement. Later, wolves
settled in forested areas with higher road and human population densities
(Fuller et al. 1992). The wolf populations in Wisconsin and Michigan were
separated from the larger source population in northern Minnesota by large
areas of less-favorable habitat and Lake Superior. Further, much of the wolf
mortality was human caused, whether intentional, accidental, or indirectly
caused through the transmission of disease from domestic dogs (Fuller
et al. 1992).
Because the management objectives of state agencies included protection of colonizing wolf populations, the agencies wanted to predict the fates
of small, disjunct populations under alternative assumptions about humancaused mortality. To address this question, Haight et al. (1998) used the wolf
model to simulate a hypothetical disjunct wolf population. The model
assumed a maximum of 16 wolf territories divided into core and peripheral
ranges. The average annual mortality rate in the core range was 20%,
whereas the mortality rate in the peripheral range was higher (40%)
because of human-caused deaths. Haight et al. (1998) conducted a set of
2. Modeling for Endangered-Species Recovery
31
were negatively correlated with mortality and suggested that population
size stabilized with a mortality rate of about 35%, similar to the conclusion
of Fuller (1989) based on field studies. Additional model tests and sensitivity analyses are reported in Cochrane (2000).
The software for the wolf simulation model was written by and is available from the two senior authors. Versions of the source code were written
in FORTRAN and Visual BASIC. The applications were performed on an
IBM300PL and other personal computers. We have used this type of model
for other social carnivores, including the San Joaquin kit fox (Vulpes
macrotis mutica) in California (Haight et al. 2002) and the African lion
(Panthera leo) (Starfield et al. 1981). Population models with similar territorial and dispersal mechanisms were used for northern spotted owl (Strix
occidentalis caurina) recovery planning (Lande 1987; Lamberson et al.
1994).
2.3.2 Persistence of Wolves in
Human-Dominated Landscapes
Following protection under the Endangered Species Act in 1973, wolves
from northeastern Minnesota recolonized most of northern Minnesota and
parts of northern Wisconsin and northern Michigan (see Figure 2.1). The
landscape in this range was not wilderness but a mosaic of forest, agricultural, and developed land under a variety of public and private ownerships
(Mladenoff et al. 1995). Logging and agriculture had created extensive areas
of young forest that supported large populations of white-tailed deer, the
preferred prey of wolves in this region. Colonizing wolves first settled in
forested areas with few roads and little human settlement. Later, wolves
settled in forested areas with higher road and human population densities
(Fuller et al. 1992). The wolf populations in Wisconsin and Michigan were
separated from the larger source population in northern Minnesota by large
areas of less-favorable habitat and Lake Superior. Further, much of the wolf
mortality was human caused, whether intentional, accidental, or indirectly
caused through the transmission of disease from domestic dogs (Fuller
et al. 1992).
Because the management objectives of state agencies included protection of colonizing wolf populations, the agencies wanted to predict the fates
of small, disjunct populations under alternative assumptions about humancaused mortality. To address this question, Haight et al. (1998) used the wolf
model to simulate a hypothetical disjunct wolf population. The model
assumed a maximum of 16 wolf territories divided into core and peripheral
ranges. The average annual mortality rate in the core range was 20%,
whereas the mortality rate in the peripheral range was higher (40%)
because of human-caused deaths. Haight et al. (1998) conducted a set of
2. Modeling for Endangered-Species Recovery
31