HABIT@

Wildlife Habitat Capability Modeling Program

Historic Range of Variability Analyses

This section provides a description of the HABIT@ analysis done for this application with reference to the species’ models described elsewhere, including a description of the input data generated from the RMLANDS simulation and a description of the methods employed to summarize changes in habitat capability over time.

      For purposes of this application, HABIT@ analysis was tightly coupled with the output of RMLANDS. The RMLANDS simulation produced a landscape snapshot for each time step and then HABIT@ was used to assess the habitat capability of each snapshot for each indicator species. We summarized the range of variation in each species’ habitat capability index to characterize the HRV.

Input Landscape Data

      The input landscape data for HABIT@ consisted of a suite of output grids generated by RMLANDS for each time step of the simulation. The specific RMLANDS output grids utilized varied among species, as given in the species’ model descriptions (see HABIT@ - Species’ Models), and included the following (note, HABIT@ is not limited to the use of these RMLANDS particular output grids):

      •    covercond       Cover-condition grid. This grid is a categorical map in which each unique combination of cover type and condition class has a unique class value.

      •    covercondt-1   Cover-condition grid from the previous timestep (10 years ago).

      •    cond                Condition grid. This grid is a categorical map in which each unique stand condition class has a unique class value. In many cases, the same stand condition class exists in multiple cover types.

      •    slope               Percent slope grid. This grid is a continuous surface map in which each cell takes on the value of the local slope, given in percent slope.

      •    wfashm           Years since high-mortality fire. This grid is a continuous surface map in which each cell takes on the value of the number of years since the last high-mortality fire.

      •    sbashm            Years since high-mortality spruce beetle outbreak. This grid is a continuous surface map in which each cell takes on the value of the number of years since the last high-mortality spruce beetle outbreak.

      •    sbaslm            Years since low-mortality spruce beetle outbreak. This grid is a continuous surface map in which each cell takes on the value of the number of years since the last low-mortality spruce beetle outbreak.

      •    fbashm            Years since high-mortality Douglas-fir beetle outbreak. This grid is a continuous surface map in which each cell takes on the value of the number of years since the last high-mortality Douglas-fir beetle outbreak.

      •    fbaslm             Years since low-mortality Douglas-fir beetle outbreak. This grid is a continuous surface map in which each cell takes on the value of the number of years since the last low-mortality Douglas-fir beetle outbreak.

Characterization of Habitat Capability Dynamics

      To characterize the dynamics in habitat capability for each indicator species under the HRV scenario, we simply plotted the landscape capability (LC) index generated by each species’ model over time and then summarized its statistical distribution. The LC index provides an index of the number of homeranges likely to be supported across the landscape given the pattern of habitat capability across the landscape. The LC index can be based on either the homerange capability (HRC) index or population capability (PC) index (see HABIT@ - Model Overview) and is appropriate for territorial species, although it can also be used as a general index for species that do not defend territories. Briefly, it randomly places non-overlapping homeranges, starting with highest-valued cells, until the landscape is saturated with potential homeranges. Then, homeranges are randomly kept or dropped with a probability equal to the PC or HRC for each homerange center (thus, a homerange centered on a cell with PC = 0.8 has an 80% chance of persisting). The result is the number of realized homeranges on the landscape. The random tiling process can be repeated several times to account for its stochastic nature. In practice, when the homerange size is relatively small compared to the landscape extent (as is the case for all indicator species we considered), the number of realized homeranges varies remarkably little among replicates. Nevertheless, we repeated the tiling process five times and reported the mean.

      In addition to the simple statistical summary of LC, we also generated LC “response curves” to evaluate the frequency at which the LC index dropped below a particular threshold value. Specifically, we plotted (as a line chart) the cumulative percentage of time (on the Y-axis) against the LC index (on the X-axis with the highest value at the origin). Thus, the y-intercept was always anchored at 100%, since the LC index was less than or equal to its maximum value 100% of the time over the 800-year simulation. The curve decreases in some fashion and approaches 0% at the minimum LC value. Thus, the response curve can be interpreted as the percentage of time over the course of a simulation that the LC index drops below a particular threshold level. If biologically significant thresholds in the LC index can be identified, then response curves can be used to determine the likelihood of exceeding these critical thresholds under a particular disturbance regime. Alternatively, in the absence of known biological relationships, response curves can still be used to compare the relative effect of alternative disturbance regimes on habitat capability.

      Unfortunately, due to intensive computer processing requirements of HABIT@, we were only able to complete the above analysis for the Columbine District, which encompasses 308,829 ha and represents 36% of the SJNF. In addition, we were only able to complete the analysis for a single simulation run. Hence, the statistical summary of LC and the response curves were based on 70 observations (70 snapshots at 10-year intervals from a single 800-year simulation run, after excluding the first 100 year equilibration period). Note, based on other analyses, we are confident that the basic habitat patterns are consistent across runs. Thus, the measured range of variation in each species’ habitat capability (LC index) likely would not differ from the results reported here had we been able to complete the analysis for multiple runs.