Moore Notes 2 23 11

From OpenWetWare
Jump to navigationJump to search

Group Call - Niche Mapping subgroup

  • Josh and Katie working on manuscript
    • Getting text and figures from James, Tom, and Alex
    • James will finish figure today. Second paragraph of his text is for methods section.
    • Goal: circulate manuscript on Friday or Monday for full group to comment.
  • Alex: phylum level analysis plots
    • Some phyla are richer at tropics, others in temperate latitudes
    • Adjusting for number of genera per phylum
      • when look at 20 random subsets of 5 Proteobacteria genera, the dip in richness at equator is significantly reduced
      • Steve: may need more draws. Alex has this data, and can check.
      • Josh will check that genera not in subset of 5 are being added back to the baseline "unclassified" group
    • Tom: Why doesn't cyanobacteria richness map match chlorophyll concentration?
      • Didn't use primary productivity as a predictor, so interesting that we see this pattern independently
      • Should check abundance instead of richness for this question
      • Steve: some diverse places don't have a ton of biomass (Mediterranean) - need to deflect criticisms of genus-level analysis being meaningless
  • Reviewers
    • Biodiversity people over microbial oceanographers
    • Josh will check process at Science
  • Josh: bias in GLMs
    • Getting higher probabilities than we should
      • No zeros
      • Jensen's inequality means we are overestimating the probability of occurrence of each genus
    • What sampling depth should we use to draw the richness maps?
      • Bias is highest at intermediate levels of reads (10,000 reads)
      • Could potentially adjust this to compensate for bias
    • Observed data may underestimate presence, so complex