add function to show presence/absence by depth bin?

[iantaylor-NOAA]

The figure below was created for the 2019 Big Skate assessment as a simple illustration of the depth distribution of the species, where the width of each bar is proportional to the number of tows within that depth bin. In this case of Big Skate it was useful to see that the modal depth was likely near or shallower than the 55m minimum of the survey.

Would it be useful to generalize this for application to any species which occurs in the WCGBT Survey?
If so, I would provide inputs to choose the latitude range, depth range, and depth bin size and any other refinements that anyone suggests.

Code is at https://github.com/iantaylor-NOAA//BigSkate_Doc/blob/master/R/Depth_explorations.R

[chantelwetzel-noaa]

I think this type of addition would be great. I have a few other plots that I have been working on for survey visualization. One of them is a similar to your above (not nearly as pretty) which shows the percent positive tows by year. I think we could also do a similar plot to your but have it across latitudes. I will not be able to incorporate any of this new plotting functionality in the next couple of months but that could give us time to identify a needed suite of plots and how best to create the code (e.g. something similar to SS_plots that would do a full set of plots calling separate plotting functions internally).

[iantaylor-NOAA]

New function PlotPresenceAbsence.fn() added in commit 253248d.

Example commands below from ?PlotPresenceAbsence.fn produces the figures below

catch.WCGBTS.ling <- nwfscSurvey::PullCatch.fn(Name = "lingcod",
                                               SurveyName = "NWFSC.Combo")
PlotPresenceAbsence.fn(catch.WCGBTS.ling, dim = "lat")
PlotPresenceAbsence.fn(catch.WCGBTS.ling, dim = "depth")

[chantelwetzel-noaa]

Thank you @iantaylor-NOAA. I used the example code that you sent to create plots for Dover sole and it was super informative. Based on my playing around with Dover sole, I have some ideas of options and features that will create reasonable plots across species. This give us a create start for development. To help foster additional ideas here are the modified plots that I created for Dover sole (with Ian's help I can make great plots too!):

[iantaylor-NOAA]

Nice Dover plots, Chantel.
I should have noted that I have not tested this on anything other than Lingcod, so please modify as you wish for other species and other features.

[chantelwetzel-noaa]

Quick question. I inferred that the width of the bins in your example are related to the cpue by bin (e.g., higher cpue would result in wider bin). Is this correct? I standardized everything for Dover sole because the numbers of observations across depths and the size of positive tows was a bit messy since it is the extreme end for trawl species. I think both options would be useful.

[iantaylor-NOAA]

I believe widths are proportional to the number of tows within each bin, which I think is also useful information, but I could see how it could add confusion for a species like Dover that spans all depths everywhere.

100% of the plotting code in this function is done in the single line
https://github.com/nwfsc-assess/nwfscSurvey/blob/5f160c1736832141f9cb369aa5a9504e3cd0ad84/R/PlotPresenceAbsence.fn.R#L127
where using plot() on a 2-dimensional table calls plot.table() which then apparently calls mosaicplot(). I have no idea how you change the widths of the bars or anything else (other than the color, I guess).

[kellijohnson-NOAA]

@iantaylor-NOAA and @chantelwetzel-noaa would you mind if I changed the x-axis bin labels to be on an angle because the values often overlap (e.g., see Ian's example above and the example below).

[kellijohnson-NOAA]

Found that las = 3 can easily change the labels to always be vertical, which we could do by just adding ... to the args to allow people to pass arguments to plot, e.g., las = 3 which makes the following and does not force people to use vertical labels.

[chantelwetzel-noaa]

That is great. Feel free to commit this change. I plan on working more on this plotting function today or next week to incorporate additional revisions that I explored when creating a figure for Dover sole to make it more versatile.

[kellijohnson-NOAA]

I created a branch plot_presence that has a function that can {ggplot2} versions these figures. I pushed a wrapper as well that saves the CPUE by depth and latitude and we can easily add sex to the wrapper but I just haven't yet. Below are the CPUE figures. Let me know what you think. @chantelwetzel-noaa they don't have white backgrounds because then you cannot see the gridlines, which I think are helpful. I can change this though, no problem.

[chantelwetzel-noaa]

These figures look good. In my limited experience creating barplots with ggplot2, I had issues where there was white space between the figure and the axes but you have solved that issue. Have you successfully create these similar figures with equally sized bars across the x-axis? I think having that option is fairly important for species with wide depth distributions (e.g., Dover sole, sablefish, thornyheads).

[kellijohnson-NOAA]

This can be accomplished now with type = c("vspine", "hbar") which is some ggmosaic::geom_mosaic(divider = ) black magic.

[chantelwetzel-noaa]

Can you give me and example call for this? I just tested this out by using:

plot_bin_factor( data = bio_nwfsc_combo %>% dplyr::mutate(Sex = codify_sex(Sex)), col_factor = Sex, col_bin = Depth_m, type = c("vspine", "hbar"), width = 50, boundary = 0 ) + scale_fill_manual(values = c("red", "blue", "gray"))
which produced this figure:

This is a fine way to look at this but I would also like to be able to produce the sex ratio where the proportions by sex sum to 1.