Length-type conversion ratios from FishBase POPLL
get_length_conversions.RdFishBase tags every published length-weight pair with the length type the
original study measured. For tunas, billfish and several carangids that is
fork length, because FL is the standard measurement in those fisheries –
not because anything is wrong with the record. Zanzibar's enumerators
measure total length, so an FL-fitted (a, b) cannot be applied directly:
a 200 cm TL swordfish is 186.5 cm FL, and feeding the TL straight into the
FL relationship weighs it 105.7 kg against 83.7 kg, 26% too heavy.
The conversions are published data, in FishBase's POPLL table. This reads
them and reduces each to a single scaling factor ratio such that
L_type ~= ratio * TL, which is what convert_lw_to_tl() needs.
Details
POPLL stores a linear fit, Length2 = a + b * Length1. Treating it as
proportional (dropping the intercept) is what makes a power-law conversion
possible, and it is well supported: of the 23,921 TL-to-FL/SL rows in
release 25.04, 21,029 have an intercept of exactly zero and 23,222 are
below 1 cm. Rows with a larger intercept are not proportional and are
discarded rather than approximated. Where several fits exist for the same
species and type, the median ratio is used.
The resulting ratios are physically sensible – median FL/TL 0.962, SL/TL 0.831 – and validate against the 632 species that carry both a native TL pair and an FL one: converting halves the median error in predicted weight (15.6% against 27.5% for using the FL pair as-is), and on the 1,186 species with both TL and SL pairs it cuts it fourfold (16.8% against 70.6%). The residual is the scatter between independent published studies, not conversion error.