Linear Morphometrics
We modified our recent approaches (Pyron and Beamer 2023) based on
Bingham et al. (2018)’s strategy for multivariate assessment of
salamanders. We recorded 17 measurements at 0.01mm precision for 89
specimens from the 31 sites: SVL (snout-vent length), TL (tail length),
AG (axilla-groin length), CW (chest width), FL (femur length), HL
(humerus length), SG (snout-gular length), TW (tail width at rear of
vent), TO (length of third toe), FI (length of third finger), HW (head
width), ED (eye diameter), IN (internarial distance), ES (eye-snout
distance), ON (orbito-narial distance), IO (inter-orbital distance), and
IC (inter-canthal distance). We implemented the general recommendations
and statistical approaches of (Onn and Grismer 2021; Chan and Grismer
2022) for linear morphometrics, allometric size-correction, and
quantitative analyses. We employed the “common within-group pooling”
strategy, wherein distinct populations or localities within a species
(e.g., geographic lineages) have separate slopes but a shared intercept
representing overall body-size for the whole species. We tested both
pooled-group and multi-species approaches; they were qualitatively
identical, and we therefore present the more conservative pooled-group
analyses. After allometric size-correction, we performed
Linear-Discriminant Analysis (LDA to examine phenotypic differentiation
of populations. We identified variables with the heaviest loading across
the discriminant axes as a preliminary indicator of the morphological
trait with the greatest importance in differentiating populations.