Abstract:
A spatially explicit population model written in C++ programming language is used in
this study examining the population dynamics of the Desert Tortoise (Gopherus
agassizii) in the California Mojave Desert. The model is constructed with hexagonal
divisions of space, and with one-year increments of time. Four thematic maps: vegetation,
soil type, roads, and elevation, are used to generate hexagon grids classif'ing the status of
resources and roads in the modeled site. The grids are used as the input habitat data for
runs of the model. Simulations used a projection matrix for seven size classes. The matrix
contains the averages of demographic rates of tortoise population in different study sites
in the Western Mojave (Doak Ct al. 1994). Four landscapes with a total area of 2736.21
are sampled from the study site. Main events taking place in one time step include
movement, survival, and reproduction. The movement is modeled for eight size classes
based on the mobility characteristics of the species and the size-specific mobilities. The number of animals moving out of a hexagon in every time step is based on the
instantaneous resource gradient between that hexagon and its six neighbors. Reproduction
and survival are responses to the site-specific status of roads and rainfall. Because of the
serious lack of field data to support relationships hypothesized in the model, sensitivity
analysis of model parameters was conducted. This study also examined the elasticity of
elements of the projection matrix to see if the rankings of elasticities are different in a
spatial context.
The results show that the road effect is the most sensitive parameter in the model.
Differences in site-specific population dynamics can be well explained by local road
status. Effects from spatial and temporal variation of rainfall do not show any obvious
differences in population dynamics between sites. Simulations conducted at the finer
scale show the dynamics of population better than at the coarser scale. Elasticities of vital
rates in the projection matrix in the spatial model have the same ranking pattern as in a
non-spatial model.
