Restate Length-Weight Coefficients on a Total-Length Basis

Rescales a so that a length-weight pair published against fork, standard or another length type reads as a pair against total length, using the length-length conversions from the same expansion.

Usage

convert_lw_to_tl(length_weight, length_length, max_intercept = 1)

Arguments

length_weight: Length-weight rows as returned by get_length_weight_coeffs(). Must carry species_found, server, Type, a and b.
length_length: Length-length rows as returned by get_length_length_coeffs(), from the same expansion so that species_found and server are keyed consistently. Build them with length_types = NULL: the default c("TL", "FL") keeps no standard-length pair, and SL is where the correction is largest.
max_intercept: Numeric. Conversions whose intercept aL exceeds this in absolute value are discarded before the median is taken. Defaults to 1 (cm); a large intercept means the fit is not well approximated by the pure ratio applied here.

Value

length_weight with a restated and Type set to "TL" for every row that had a usable conversion. Rows with no conversion are returned unchanged, still carrying their original Type, so nothing is silently dropped. Filter on Type == "TL" afterwards if only restated pairs are wanted.

Details

Given W = a * L_type^b and L_type ~= ratio * TL, substitution gives W = a * ratio^b * TL^b. So b is unchanged and only a moves.

The ratio comes from the POPLL table, which fits Length1 = aL + bL * Length2 — the second column is the predictor. A row with Length2 == "TL" therefore gives ratio = bL for Length1, and a row with Length1 == "TL" gives ratio = 1 / bL for Length2. Reading that fit backwards inverts every ratio and moves weight by roughly (1/r)^b / r^b — about 1.1x for fork length and 1.8x for standard length — so the direction is asserted in the tests rather than left to the reader. Where several published conversions exist for a species and type, the median is taken.

Restating rather than filtering matters for coverage: keeping only Type == "TL" was measured to discard more than half the matched species for several taxon codes. Which rows to feed in is the caller's decision — pass everything to restate every convertible pair, or pass only the taxa that have no native total-length pair to leave the rest untouched.

Examples

if (FALSE) { # \dontrun{
m <- get_taxa_morphometrics(taxa, length_types = NULL)
lw_tl <- convert_lw_to_tl(m$length_weight, m$length_length)

# Restate only the taxa that have no native total-length pair
covered <- unique(m$length_weight$alpha3_code[m$length_weight$Type == "TL"])
lw_tl <- convert_lw_to_tl(
  dplyr::filter(m$length_weight, !alpha3_code %in% covered),
  m$length_length
)
} # }