Effectiveness of GIS in assesinggrizzly


In the Central Coast of Brittish Columbia.

Project Design

Factors and constraints
MCE Analysis

grizzly with salmon     The project design focused on three themes.  These were the creation of a grizzly bear habitat capability model, the performing of multiple multi-criteria analyses (MCE) and comparing the results of the MCE.
  To illustrate the ways in which institutional objectives might influence the decision set, three differently weighted multi criteria analysis were performed.  The MCE technique, offered in the Decision Support module of Idrisi, enables several different spatial data criteria to be analyzed.  Habitat quality is best described in terms of a range of suitability, not in terms of crisp boundaries.  For example, old forest is more desirable than selectively logged forest, which is in turn more desirable than logged forest.  For this reason, all factors were described using weighted linear combination (WLC) where factors were standardized to a common numeric range from 0-255 using fuzzy logic (Ronald, 1999). 
    With a MCE evaluation in Idrisi, regions that rank high in one factor can balance low performance in another.  Additionally, the degree to which certain factors are allowed to influence the final decision can be manipulated.  This process of weighting criteria differently was used in an attempt to mimic how different approaches to performing a multi-criteria evaluation can use similar data sets to arrive at different decisions.  The computer analyzed each cell by multiplying the value of a factor by its weighted importance and then summed the total from all of the factors to arrive at a new value that accounts for that cell's total performance. The final analysis of constraints served to mask out all undesirable areas.  The final result is a suitability map depicting a range of values where each cell is ranked according to its performance for each weighted factor. 

                                                                  salmno clearsut In creating a grizzly bear habitat capability model, key factors and one constraint
 were identified and data sets obtained.  Criteria are defined as parameters, or evidence, in relation to a decision that can be measured and evaluated.  Two types of criteria are factors, which can either detract or enhance from the suitability of an alternative, and constraints, which limit or prevent that alternative.  The only constraint identified in this analysis was large bodies of salt water including fjords and the ocean.  Key factors were selected based upon well documented needs of grizzly bears seasonally related to food sources, ideal vs. altered habitat, and the influence of human elements of the landscape.  This study highlighted habitat structures of low elevation old growth valleys, avalanche chutes, estuaries and sedge grass.  It recognized the importance of salmon bearing streams as a key component of grizzly diet.  It also emphasized the negative effects of human settlements and regions of human influence, including habitat alteration such as logging. 

Grizzly Bears in the Khutzeymateen study spent most of their active period on lower slopes or valley bottoms and in all seasons, actively selected habitats for feeding and bedding though these choices varied seasonally (MacHutchon, 1993).  The preferred and most heavily used habitats were consistently lower slopes and valley bottoms including both non-forested and forested habitats.  Flood-plain old growth, wetlands and estuaries were all heavily used.  The Bone Creek study (Saxena, 1999) reported that the species' most limiting life requisite for modeling purposes is the presence of spring habitat when bears descend to low-elevation wetlands and open forests for foraging needs.  This study further emphasizes the importance of late summer and fall foraging habitats and the availability of high-energy forage available in higher elevation study areas adjacent to the study site.  Denning activity was anticipated to be focused at higher elevations.   Numerous studies recognize the importance of human alteration of the landscape and mortality on habitat selection by grizzly bears.  Primm (1996) argues that the human caused mortality issues are the most important factor for understanding the conflict over grizzly bear management, though they also be among the most complex and difficult to analyze.  

These 7 factors were imported into Idrisi, converted from Vector to Raster format and then reclassified.  In some instances multiple themes were overlayed and again reclassified to create Boolean images representing the desired elements.  With the exception of the slope factor, which was analyzed using Surface, a distance operation was performed on each boolean image.  Fuzzy from the Decision Support Module then used to create fuzzy based logic classifications that rated each pixel cell according to its suitability for grizzly bear habitat. 
Top of Page

Factors and Constraints :

  Click here to see Data Model                                           Factor 1: Land Use

  landvalue model                                                         This factor in the analysis attempted to classify the landscape based on an estimation of how suitable a raster cell was in terms of habitat quality to grizzly bears.  The original data was the BTM data that identified 19 land use classifications in the region.  These were then reclassed according to an estimation of how suitable each general land type would be for habitat purposes.    Land types ranking high in terms of grizzly bear habitat (old forest, sub alpine avalanche, estuaries and wetlands) were given the highest values of 255.  Land types that provided some habitat or connectivity corridors, such as alpine and young forest, were ranked moderately at 150.  Land types that represented little use in terms of habitat, but did not represent negative factors were given the value 100 (selectively logged forest and barren surfaces).  Land types that represented a barrier to grizzly bear movement or habitat were given the low value of 50 (recently burned, glaciers and snow, recently logged).  Finally, land types representing a constraint to grizzly bear habitat were given a nil value (urban, residential, agriculture, mining, range and  water). 
  Click here to see descriptions of BTM classification Values

                                                                                        Factor 2: Old Forest                                                                          Click here to see data model  
old forest Old growth forest, with its structurally diverse and open canopy, provides an abundance of food sources for grizzly bear (Jeo, 1998, MacHutchon, 1993, Saxena, 1999). This factor attempted to emphasize the importance of old growth structure for grizzly bear habitat, foraging and denning using BTM data's old forest classification.  Unfortunately, the government's old forest classification does not delineate old growth structure from old forest structure.  This is a serious limitation to this factor's usefulness as the old growth characteristics most useful for grizzly bear habitat are not distinguished from those only marginally useful.  Low elevation forested riparian zones contain the majority of bear activity due to high species diversity, presence of fish and opportunity for day bedding.
       *   Lup BTM layer was reclassed so that 9 = 1 and all other values = 0.
     *  Disance and then Fuzzy was run using a sigmoidally decreasing set where c=150m and d=1000.                        

   Click here to see data mode                                          Factor 3: Human Settlement

hum set   This factor included data from two layers of the BTM data.  The LUP layer detailing land use designations was used to indicate the location of all urban settlements greater than 15ha.  This data was combined with the CLUP point file listing the location of all villages, towns and buildings which included locations smaller than the minimum mapping unit of the LUP layer.
   In order for people to kill grizzly bears, they need to be able to get to them.  Roads, camps, villages and other sites that bring people close to grizzly bears generally lead to contact, conflict and bear deaths (Primm, 1996).   Implicit in this factor is the access opportunities relevant to the issue of hunting (McLellan, 1999 & Wielgus, 2001).  Between 55-75% of grizzly deaths are caused by humans (Jeo, 1998).  This number reflects the hunting (89%), animal control (8%), illegal poaching (2%) and road kill(1%).  (Austin, 2002)  In regions where there is human-bear contact there are even higher numbers of  'problem bears' killed (McLellan, 1999). These interactions are considered in the following three factors as detracting from the suitability of the area.           The Round River (Jeo, 1998)  suggests that bears avoid 400-2000m of human sites and up to 5 km in areas known for high                                                                 hunting mortality.  
                                                   * Lup BTM layer reclassified so that 13 & 16 = 1.            All Clup values were reclassified to 1.
       l                                                                * These two layers were overlayed and reclassed to form a boolean image
                                                              * Disance and then Fuzzy was run using a sigmoidally increasing set where a=750m and b=5000m   

                                                                                     Factor 4: Human Influence                                                               Click here to see data mode l
This factor included data from three layers of the BTM data.  The LUP layer detailing land use designations was used to indicate the location of all recreation areas, recently logged and selectively logged areas, and mining locations greater than 15ha.  This data was combined with the TRNL line file listing trails and cart tracks in the region.  This data was also combined with the TRNP point file that listed the location of all airstrips and sea anchorages.  There are numerous studies documenting the extreme under use of modified habitats (Jao, 1998. McLellan, 1999) and problems associated with increasing encroachment by humans into grizzly bear habitat (Hood, 2001). Hood suggests that impact should be evaluated up to 500m.
      * Lup BTM layer reclassified so that 11,12,15 = 1, Trnl layer was reclassified so that 1,4=1, trnp was reclassified to 1.  
              * These three layers were overlayed and reclassed to form a boolean image.
     * Distance and then Fuzzy was run using a sigmoidally increasing set where a=250m and b=1000m.                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    

Click here to see data model                                            Factor 5: Roads
 roads                                                               This factor attempts to recognize the negative impact roads have on grizzly bear habitat.  Ideally, a data set indicating the density of roads,  including logging roads, would be the most useful.  However this form of data is linked to polygon shapes, generalized from regions, and was unavailable for this study.  Therefore a layer indicating the location of all gravel and paved roads was used, TNRL, from the BTM data.   Landscape fragmentation is greatly influenced by the presence of roads and road density is a good indicator of the ecological value of an area.  For this reason, roads and road density are referred to as "keystone disturbances" (Jao, 1996).  
      * Trnl  BTM layer reclassified so that 2,3,5 = 1.
      * Disance and then Fuzzy was run using a sigmoidally increasing set where a=250m and b=2000m.
 salmon                                                                                                             Factor 6: Salmon                                                             Click here to see data model   
            Salmon are the most important source of meat for grizzly bears and the availability of this food source greatly influences habitat quality for grizzly bears at both the individual level and the population level  (Hilderbrand et al, 1999).  Salmon are described as a keystone species, a species that plays a disproportionately large role in the ecosystem function  (Jao, 1996).  Salmon are also increasingly related to the nutrient cycling and trophic interactions in forested ecosystems adjacent to salmon streams (ibid).  Salmon provide an essential food source for the grizzly bear when they are spawning.  This food source is particularly important for its high fat content prior to denning for the winter.
                                    * Evzl FISS layer was reclassified so that 1 = 1, all others reclassed to 0
                  * Disance and then Fuzzy was run using a sigmoidally decreasing set where c=150m and d=10,000m

    Click here to see data model                                                Factor 7: Slopes

 Grizzly bear prefer broad forested valleys although they often will use steeper slopes for denning sites.  Slope was calculated from the DEM and areas less than 15 degrees were given highest values with a sigmoidally decreasing function up to 45 degrees.  This was a difficult factor to model because grizzly bears also require steep broken terrain for denning caves (>30 degrees).  
            * Surface was run on the DEM.
        * Fuzzy was run using a sigmoidally decreasing set where c = 15 and d = 45.     


Constraint: Water

The only constraint applied in this data model was a boolean image of locations of major fjords and the ocean in order to eliminate non-land surfaces
* Lup BTM layer reclassified so that only 19 = 1

                                                                                                Top of page
 Click here to see data model  

Performing Multiple Multi-Criteria Evaluations

    Click here to see Data Model
data maodel   Analysis # 1 Human influence bias: The first analysis weighted heavily the factors that were seen to negatively influence grizzly bear habitat.  The factors of human settlement, human influence and roads were therefore heavily weighted in the analysis.   Land use values were not rated highly because they were interpreted to contribute to habitat quality. 

Analysis # 2 Neutral position:  This analysis attempted to adopt a more neutral position and to allow for selected factors to stand more evenly against one another.  A few differences were highlighted, such as the relative importance of human settlements over human influence and the importance of salmon over all other factors. 

Analysis # 3  Habitat bias: This analysis weighted heavily the factors that were seen to have a strong positive contribution                                        to grizzly bear habitat. 

1 2 3  



    Click to enlarge pariwise comparison file #1                 Click to enlarge pariwise comparison file#2            Click to enlarge pariwise comparison file #3

Conceptual idea
   Introduction  Data Collection      Project Design      Spatial Analysis  Conclusions   Problems


Back to Index