Methodology

Data Aquisition and Manipulation

    All data concerned with BC commercial salmon licences was obtained from the Department of Fisheries and Oceans.  The data was received in tabular format and included a georeferenced location of the licence owner for each year from 1985 to 1999.  Where multiple licence owners existed for a single commercial fishing vessel, the owner providing contact information and residence was taken as the relevant owner for analyses.  The years 1995 and 1999 were used for this report in order to isolate our research to the effects of the 1996 Mifflin Plan.  Data manipulation and geocoding of unique census subdivision (CSD) codes was carried out using Minitab14 and Excel spreadsheets.  The finished product of this initial data manipulation was an attribute table of commercial salmon fishing community CSDs-each with an associated "Change in Licences (95-99)" percentage.  Licence change was normalized using the number of residents employed in the respective communities.  While one salmon licence cannot be equalized with one job- because fishing vessels (especially Seiners) employ multiple crew members- the relative weight of licence losses and gains is consistent throughout all communities.  Regardless, inferences based on the change in salmon licence changes must be attuned to this limitation. 
    The second source of data used is Statistics Canada's 1996 CSD attribute tables.  More specifically, tables concerned with the Average Dwelling Value ($CAD), Total Aboriginal Population and number of Primary Sector Jobs were used in our analyses.  In each analysis, as detailed below, the relevant statscan variable was joined with the CSD salmon licence table outlined above.
    Consistent with all of the maps and analyses are the following generalizations:

• BC Coastal Salmon Fishing Community is taken as any community bounded by a census definition and having greater than 5 commercial salmon licences in 1995.
• On the rare occassion where a coastal community is not given a unique census boundary, it is amalgamated with its closest neighbouring community (ex. Sointula and Area)
  
• All Statscan variables were sampled with respect to the 1996 census and originate from the DMTI database.
  
• All maps use a CSD basemap projected according to NAD 83 Albers.
  
• All commercial salmon licences owned by individuals or groups living outside of British Columbia, or at great distances from the BC coast, were excluded from analysis.
  

Schematic of Methodology and Reasoning

1. Change in Licences

Analysis 1: Projecting most severe losses and largest gains in commercial salmon licences for the years 1996-1999.
  
    A basemap was created using Ecotrust’s (2004) BC coastal definitions.  Each region was clipped and manually assigned a unique color as an identifier. After outlining the five major coastal regions (North Coast, North Island/Mainland, West Coast Vancouver Island, Central Island Mainland and Metropolitan Region) the top six coastal communities in terms of licences lost (as % of community employment) were mapped and classified from 'hot' red tags to 'cold' blue tags (Figure 1).  Figure 2 shows the exact opposite illustrating the six largest gains in commercial salmon licences in coastal communities.  A subsequent excel table was imported into both maps further demonstrating the relative change.  

2. Exploratory Analysis

Analysis 2: Determinining and projecting correlation between licence change and dwelling value

    Firstly, in order to add context and further emphasize the movement of licences amongst BC's coastal communities, the regional change in licences will be shown in Figure 3.  But, at this juncture, our research begins to question not only where licences are being bought and sold, but also what factors correlate with these changes.  Overlaying the regional map are bar charts indicating their residuals with respect to the correlation between regional licence change and average dwelling value. A regression analysis was carried out for the two variables “Licence Change” and “Average Dwelling Value” using Minitab’s analysis program.  The output map was then exported to Corel Draw where its properties were adjusted to make it compatible with ArcMap.  Finally, the plot was re-exported to ArcMap as a gif. file. 
    Two methodological problems should be noted. Firstly, when creating a average dwelling value for a region it should be recognized that regional value is not necessarily representative of each of its constituent CSDs that it contains.  Also, the CSDs used can be percieved as arbitrary political definitions.  If different boundaries were used the "average" could be dramatically different.  Secondly, it should be noted that Census Canada suppresses CSDs,  and these were excluded from our computations of "average dwelling value".
    This regional analysis is then complimented with two focused maps displaying the coastal salmon fishing communities and average CSD dwelling value, namely the North Coast (Figure 4) and the North and West Coast of Vancouver Island (Figure 5).  These two regions were chosen based on the previous regional map's illustration that these two regions have been the most severely impacted by licence losses.  After clipping the two regions (West Coast/North Island and North Coast), they were given a choropleth symbology of average dwelling value. The salmon icons were exported from corel draw 11.

Analysis 2.1: Determining and projecting correlation between licence change and aboriginal communities
   
    Showing the relationship between licence change and % aboriginal population required two analyses.  The first linear regression (using Minitab 14) indicated that aboriginal population, at first glance, does not predict the change in licences with much success (Figure 6).  This observation necessitated a second regression, but this time the two outliers (Sayward and Port Simpson) were removed (Figure 7). In both instances the regression analyses were symbolized using the residual values and categorized by their standard deviation.
    A symbology of graduated circles was used for both maps. Instead of using graduated size circles we used graduated colors because the latter better depicts contrast between the coastal communities.  Mapping small coastal communities on a relatively large coastline creates a cartographic predicament in showing contrast between communities.  It was concluded that color contrast with small circles was most appropriate for our purposes.
 
Analysis 2.2: Determining and projecting correlation between licence change and reliance on the primary sector
   
    Using the same regression analysis as discussed previously, licence change and % of the labour force employed in the primary sector were correlated and mapped with respect to their residual values (Figure 8).  Similarly to the previous analysis, the residuals were categorized by their standard deviation from the best-fit plot.
    And like Figures 6 and 7, a graduated color symbology was tailored to combat the challenge of mapping small coastal communities on an expansive BC coastline.  But, unlike the previous maps, only the top 15 communities in terms of their licences lost were mapped.  The rational for this was that we wanted to begin to narrow down our analysis to the most severely impacted coastal communities. 

3. Further Explanations and Implications

Analysis 3: Isolated commercial salmon fishing communities

    This final map, (Figure 9), is intended to demonstrate that community isolation has both contributed to licence change and exacerbated the fallout of licence losses.  Overlaying BC's major roads and highways on the map of BC, and using a 10km buffer, community isolation was posited based on community distance from these buffered arterial routes.  Spatially joining the coastal communities with the buffered street map illustrated those communities that lie outside the roads proximity.  If a community did not overlap with the buffered street map, it was considered to be in isolation.  Isolated communities were then symbolized using ArcMap's bar chart feature and with respect to distance from the closest major road or highway. All buffering operations were carried out in Arc catalog.  All buffers were dissolved where overlap occurred.