Moore Notes 1 5 11

Group Call

• Organizing group calls for next 6 months
  • group calls every week, unless PIs call a PI-only meeting
  • announce topics ahead of time so people can selectively participate

• Will there be a final in person meeting?
  • Folks found them useful
  • Question of how to pay for it
  • Maybe small groups will get together some time this spring
  • Josh visiting UO next week
  • Steve may visit UCSF this spring

• Niche mapping
  • Statistical community has issues with maxEnt
    • theoretically inconsistent
    • some times really wrong in simulations
    • but can be very accurate
    • SSU-rRNA data is different from intended use (museum samples) in some ways
  • GLM and maxEnt results - comparison
    • which to emphasize?
    • compare quantitatively at per taxon basis? qualitative will be much easier!
    • cite papers on GLMs for species distribution modeling
    • can use GLM to get occurrence estimates at different sampling efforts
      • avoids issue of choosing suitability threshold for "present" using maxEnt
      • need to compare maxEnt and GLM for several maxEnt suitability thresholds (maybe pick one that agrees with GLM)
      • could also calculate relative abundances (e.g., Simpson's and Shannon's index)
  • Want to use multinomial logistic regression rather than separate logistic regressions per taxon - Josh is working on some technical issues in R
  • What taxonomic level should we do richness at?
    • higher levels are more thoroughly sampled
    • interpreting results is harder at higher levels
    • does it make sense to do richness for all bacteria?
      • usually phylum level for eukaryotes
      • could make separate maps for each bacteria phylum
      • when do maps start looking different as you go down in phylogeny?
      • Alex will work on this
  • Outline/main messages (Word document)
    • richness / alpha diversity
    • richness vs. latitude
    • beta diversity
      • maybe exclude figure and just report result
      • could show distance decay instead (James is working on this)
    • Rapaport's rule (range breadth, seasonal variability)
    • maps for specific taxa
  • Methods document for next week - Katie will help
  • Tuning maxEnt parameters? Maybe hold off if GLM looks similar
  • Revisiting any ideas from previous calls