fmichonneau / sok-marine-biodiversity Goto Github PK

to get a sense of how much the new records contribute to the number of species that are in databases but not in the species lists, it should be possible to get the year for first collection and therefore get a sense of how much these new records contribute to this number

on the idigbio_records and related, it would be good to check using the WoRMS classification (and not the taxonomy from iDigBio) that all species are actually inverts (it seems that a few fish may have still make their through the data clean up process).

it seems the logic is simplified enough that I don't need to rely on this anymore

the numbers for the rare phyla are possibly exacerbated by low diversity in some phyla (e.g. only 10 spp of phoronids occur in the US). Need to control for diversity, could use phylum estimates from Appeltans et al 2012.

maybe try sequential and cumulative approaches; it should highlight how little collecting there has been in recent years

from JAM:

I would probably start also with something on the rate of biodiversity loss -- or lack of knowledge in the oceans or something... highlighting the importance of this work. If you can put some sort of number there and then say this is likely a huge under-estimate due to the fact that we dont really know what is there.

the add_ prefix suggests we are adding info to the original dataset, while (at least in its current form), this function gets the stats for each species in the original dataset

the trend that appears would however reflect more the decrease in sampling effort rather than a possible saturation...

it seems there is a citation associated with each record. Double check it's the case...

currently, I only use iDigBio, could be more interesting to also include species records from OBIS (and potentially GBIF) to see if they include other species

in the estimation of the proportion of species that include cryptics, remove singletons, as we can't use them if there is a single record for that species

for iDigBio, a quick manual search on the 750+ species of mulluscs and arthropods that are listed as "not in database", are actually in idigbio but do not have GPS coordinates associated with them (just generic info: e.g., Cuba, Florida, Texas) that could place them in the list of the Gulf when searching on geographic coordinates.

Maybe search by species names to estimate how many that could represent.

currently, only using iDigBio, see if possible to use other data sources.

maybe status-invertebrates-us or us-marine-biodiversity

in discussion: important to digitize older records/improve their geographic info; however, potentially can't be used for DNA (low success rate, and most specimens fixed in formalin).

to test for the idea that the number of institutions that are contributing to the knowledge of marine biodiversity is decreasing.

use estimates from WoRMS/BOLD to get total number of BINs compared to estimated number of species per phylum to get more accurate estimates than 2012 and 2009 papers...

in intro add paragraph about syntheses and other similar studies that have been published in the past

From JAM

if filtering on presence of geographic coordinates before doing the date query, I might be biasing the results by excluding older records for which no georeferrencing info is done/available.

mention (after double checking) that the FWRI collections are probably the most comprehensive for the Gulf of Mexico but they haven't been digitized yet.

splitting the map into regions, would allow me to compare richness along the coast. Maybe a simple similarity index + tree could do the job

US EEZ should be on top followed by GoM and PNW

Update is_in_gulf_of_mexico and is_in_pwn to use the real coordinates from felder_gom_map and kozloff_map to check if the points fall within these geographic ares.

the code seems better now, and I probably can find a better way to address this

a minimal dashboard that shows:

• for species name:
  • percent that have a match in WoRMS
  • percent identified at species level
  • misspellings
• for dates:
  • outside valid range
  • empty
• for geographic coordinates
  • empty
  • wrong?

for easy archival and sharing (it's not quite reasonable to archive the storr folders)

figure and commentary are currently missing, they need to be added

make sure that filtering the idigbio results on the phylum list from the gulf of mexico doesn't remove records that are otherwise valid (because a different classification was used)

use the online version of kozloff as data source

check proportion of records identified at:

• species
• genus
• family
• order

levels through time. An increase in proportion of identification at higher level could indicate waning of taxonomic expertise.

they could be fetched from the idigbio data store, but there are probably hundreds of sources... might worth to try to include them as normal citations though

e.g. Ctenophora, Amphitrite, etc..

to make the differences of percentages stand out more

• this study did an audit of GBIF records and rise some good points about data quality
• the numbers also include invasive species

• rename koz/kozloff to pnw (or simiar for pacific north west)

From JAM:

I would also talk about the issue of identifying species and needing DNA. I think that most people wont get this. I would indicate that this is one of the only methods for finding cryptic species and all sorts of diversity. Give some numbers on 'perceived species' due to morphlogical analysis and how that changed with DNA identification of some areas ... if you can. That would help to highlight some of major hurdles in this group.

see plot_samples_vs_spp_std and friends in time_plot.R for starters

I should double check that the queries for the chordata are general enough that I'm including records for institutions that may have recorded have as urochordata, etc...

instead of wonky spreadsheet, use CSV files downloaded from http://e-gulf.org/

• all the underscore functions should be updated
• use pull where needed

and loads: extrafonts, etc...