Weekly survival analysis - heat map plot

[adroghini]

The idea is to create a heat map that summarizes the number of taxa that can survive in the Bering Sea for every week of the year and for every 1° latitude. The final plot should be a heat map from low to high species richness across weeks and latitudes.

We are trying to decide how best to summarize the data across taxa, ROMS, and pixels (latitude). The following issues have come up (copy-pasted from Tony's e-mail):

• [x] we need to consider how best to summarize the pixel-wise weekly survival values across the Bering Shelf latitudinal bands (some % cut off = 1 ?? instead of the mean across the latitudinal band);
• [x] I need to then run this again with the retention of the year variable;
• [x] then we need to consider how best to summarize across years within a study period (mode or mean); and across taxa and models.

Re: Summarizing across latitude Currently summarizing cumulative taxa survival across latitude by using the mean. From Tony's e-mail: Could we have taken the mode instead?? Perhaps, however, this would have required a custom function to be applied. Given that we are summarizing across the longitudinal width of the study area within each latitudinal band (see figure below), this may be reasonable. We may also consider using some cutoff, such as casting it as a 1 if at least 10% of the latitudinal band allowed survival. This could be done by a simple rule to get us back to 1 (if > 10% of pixels in latitudinal band had survival, 0 otherwise).

[adroghini]

Current decision rules: 1) Summarizing weekly survival across latitude (within a taxon): if survival = 1 for at least 1 pixel in that latitudinal band, assign 1 (survival) for that week. From Tony: The idea that at least one pixel can be 'infected' follows a mantra from public health and infectious diseases. 2) Summarizing across years (within a study period): If weekly survival = 1 for at least 7/10 years, assign 1 for that latitudinal band. 7/10 threshold is the same threshold used in our yearly survival analysis. 3) Summarizing across ROMS: Average survival across the three ROMS. Thus, for a given taxon-week-latitude, survival can take one of four values: 0 (no survival across all 3 ROMS), 0.3 (1 model predicts survival), 0.6 (2/3 models), 1 (all 3 models predict survival). 4) For each study period, summarize across taxa by adding up the results of 3). This gives us a final metric of "Number of taxa", consistent with our yearly survival analysis.