search

AquaAuma / FishGlob_data

Database and methods related to the manuscript "An integrated database of fish biodiversity sampled with scientific bottom trawl surveys"
Creative Commons Attribution 4.0 International
21 stars 7 forks source link

length-weight for species with length reported other than TL #27

Closed AquaAuma closed 9 months ago

AquaAuma commented 1 year ago

Four families of fish for which TL is not reported, but SL/PAFL/PSCFL for which converting factors need to be applied, all are now available and easily implementable in the DATRAS cleaning code.

See: https://github.com/AquaAuma/FishGlob_data/blob/main/QAQC/DATRAS/lengthtypes_family_concern.csv https://github.com/AquaAuma/FishGlob_data/blob/main/QAQC/DATRAS/lengthtypes.csv (not that the reported length type in DATRAS does not help here)

Modified code: https://github.com/AquaAuma/FishGlob_data/blob/a5623e1ce29512b1bcc798b051368343b970fbc2/cleaning_codes/get_datras.R#L657

Yet, not sure this is good because some of those specific taxa (which should be rare/small) seem to be outliers in the dataset, and appear as some of the most abundant, even without the conversion See summary of SP-ARSA and EVHOE https://github.com/AquaAuma/FishGlob_data/blob/main/summary/evhoe.pdf https://github.com/AquaAuma/FishGlob_data/blob/main/summary/sp-arsa.pdf

We might have to double check (ii) with known total lengths of fish

AquaAuma commented 9 months ago

image Checking the raw length data distribution for some of those species in multiple surveys, it seems that multiplying by the conversion factor will make the length data go way outside of the length range of the species, which makes no sense. This means the length data did not need a correction from the start, or that it has already been corrected by the member states.

In this case, the length distribution is exactly below the total length of the species (60cm) according to fishbase https://www.fishbase.se/summary/Malacocephalus-laevis therefore invalidating the need to convert the length of the species from PAFL to TL because the data show the length is already a classic good TL distribution for the species, with the common length at centre of distribution

(checked by Aurore & Daniël on zoom)

AquaAuma commented 9 months ago

all species checked here: https://github.com/AquaAuma/FishGlob_data/tree/main/QAQC/DATRAS/Length%20Types and since most of the time the red histogram is extended beyond the max length from fishbase by quite a lot (red dashed line), we do not apply the TL conversions to the surveys