What did John Lennon Say About Plans…

mussel biomarkers
map making
spatial analysis in GIS/ QGIS
Published

February 5, 2026

Oh, that’s right, he said “Life is what happens to you while you’re busy making other plans.” Well, life today definitely didn’t agree with my plans.

Projects Touched Today

  • mussel biomarkers

New Knowledge

  • So, in my quest to wrap up mapping, I found a mistake in my spatial analysis of the biomarker metrics and indices. When I mapped my spatial results, I started with the geographic clusters designated using K-Means clustering that I verified with variance and significance testing ensure the groups were statisically different enough to be meaningful. When I threw the groups in the map, it looked like I just threw out a handful of fruit loops and said ‘ok’…
    • First, I went back and double checked the sites and assigned color palette were correct. Next I went to the data table to verify the values were in the correct format and not mistakenly overwritten. Finally, I moved over to check the code I used to analyze the clusters.
    • One seven letter word got me- degrees. I completed the SPATIAL analysis in degrees, not distance. I made a note in the R code, and moved back over to GIS to redo the analysis.
  • To redo the analysis, I did the following:
    • Imported a clean table of the metrics and identifiers, and formatted the columns from characters to actual numbers for next steps
    • Next we need to link the table to the map, so I created a point layer to link the Puget Sound map. This takes the lat/ long data and converts it from coordinates (degrees) to actual distances. I confirmed the projection and moved to confirm the results.
    • To verify that worked, beyond the points showing up on the map, I ran a distance matrix summary (by site_id, k=5) to verify the point layer and to guide the parameters I select in the Moran’s I analysis. The summary gives you a quick table of minimum and mean distances and the standard deviation; these guide the distance bands (scale) for Moran’s.
    • I confirmed the distance matrix by running a linear matrix that shows if the distances are actually varying, there aren’t any self-comparisons, and verify both tests returned consistent results.
    • Happy with the confirmation, I ran the Moran’s I (spatial autocorrelation- regional or global) for p450, mapped the result, and then moved to sod.
      • P450 showed a few significant clusters, SOD did not. This aligns with the KW comparisons and the Spearman’s correlations results. Thank goodness.
  • Tomorrow, I will return to the Moran’s I analysis of the remaining metrics (18 of 20 total), the LISA analysis (all 20), and begin mapping the results one metric at a time.

Old Tricks

The log of mapping decisions I may need to return to:

  • Returned to the sampling site map, added insets for Seattle and Tacoma, and threw it in canva to clean up the format.

  • I chose to add the following city labels:

    • Olympia, Tacoma, Seattle, Everett, Penn Cove, and San Juan Islands

  • The insets have the city label only for Tacoma and both the city label and Bainbridge Island for the Seattle one because of the geography captured in the inset

  • Next maps will most likely not need the Canva step since I figured out how to do adjust borders independently in QGIS, but we’ll see.

    • I created a specific color palette for reporting areas in QGIS to be used in all plotting/ mapping for consistency. Hex codes follow a pound sign with no spaces, and are as follows:

      • 6 - 0f6f6c (dark teal)

      • 7 - 6a4c93 (plum)

      • 8.1 - c7352d (brick red)

      • 8.2 - d4a437 (gold)

      • 9 - e07a5f (coral/ salmon)

      • 10 - 4fb3a2 (light teal)

      • 11 - ff7f00 (orange)

      • 12 - 33a02c (yellow)

      • 13 - 1f3c5b (navy blue)

    • My sampling site basemap is in purple with black borders

      • Seattle inset is framed in black

      • Tacoma inset is framed in white - may need to change to yellow or something brighter

    • My sites by reporting area map is just like the sites map, except color coded as outlined above.

      • The legend is framed in the same weight as the other frames, and has no background so that it doesn’t stand out so starkly.

      • This map feels VERY busy, feedback will help for refining

Questions or Concerns?

  • I made a quick protocol for converting non-geometric data into geometric data in QGIS. Mainly to help myself not miss a step, but secondarily to make sure I am creating mapping layers in the right projections with the right formats before I start mapping, rather than get stuck in the middle because nothing is working correctly.

  • This and the maps will be in my weekly wrap-up.

Tomorrow’s Plan

  • Tomorrow (2/6), I will return to the Moran’s I analysis of the remaining metrics (18 of 20 total), the LISA analysis (all 20), and begin mapping the results one metric at a time.
  • The documentation and analysis of the mussel pilot experiment and resazurin trials will get put back in the queue behind finishing the biomarker work.