search

mgoldgisser / desert-fires-impact-biodiversity

This repo represents the data I used for my master's thesis evaluating the impacts of fire disturbance in the North American southwest desert on biodiversity.
https://mgoldgisser.github.io/desert-fires-impact-biodiversity/
1 stars 0 forks source link

Stats ideas for writeup #3

Closed zenrabbit closed 1 year ago

zenrabbit commented 2 years ago

Main effects of fire

Fire in Mojave and Sonoran correlate with higher NDVI. Higher NDVI correlates with more species occurrences reported. But ANOVA of ALL shows there is no sig diff in standardized species occurrences reported between burned and control.

If you take birds out, only the Sonoran had a sig diff in species reported in burned area vs control (higher in burned than in control).

zenrabbit commented 2 years ago

Marina

1) what is standardized occurrences standardized to total area (e.g. # of occurrences/1,000km^2) to account for the significant difference in area of burned area vs. control 2) are you 'sure' with your sense of the data that more birds in burned areas makes sense or it is a stats handling issue? just to clarify, no sig diff between burned and control when looking at all (birds included), when birds excluded greater number of occurrences in burned vs control (because of the Sonoran). Looked into Sonoran, it looks like bighorn sheep, puma and pupfish sightings in recent years skewing data. Only 13 sheep and 11 pumas total but the burned area's relatively small density (standardized obs) is quite large. All fires took place 5+ years before sightings.

zenrabbit commented 2 years ago

Chris

  1. what if you do not divide but use as offer in GAMS? same outcome?
  2. understood - sounds like a species specific effect - NOT birds?