Predictors: Environment, sexual size dimorphism and sexual
dichromatism
Environmental information was extracted using the R package ALA4R
(Newman et al. 2020), which is an R implementation of the Atlas
of Living Australia spatial portal (Belbin 2011). We selected eight
variables that together characterise the environment of the species
studied. We focussed on variables related to vegetation productivity,
seasonality and climatic conditions during the warmest quarter as these
are most relevant to carotenoid availability, particularly during the
breeding season (Austral Spring) when lizards were sampled. The eight
variables were the annual mean growth index for C3 and C4 megatherm
plants, annual mean aridity index (the monthly ratio of precipitation to
potential evaporation and an indicator of dryness), radiation,
temperature and precipitation of the warmest quarter (Bioclim variables
26, 10 and 18 respectively), and temperature and precipitation
seasonality (Bioclim variables 04 and 15 respectively; details in
supplementary material Table S5). These variables were used in a
principal component analysis where the two first axes extracted
explained 47.8% and 33.6% of the total variation. The first axis (PC1)
was associated with growth index of C3 megatherm (tropical, broadleafed)
plants, aridity, radiation and temperature of the warmest quarter
(Figure S5; Table S5). In the figures we multiplied PC1 by -1 so that it
can be interpreted as overall productivity with high values indicating
environments that are more productive, less arid and with less extreme
summer radiation and temperatures. The second axis (PC2) is highly
related to growth index of C4 plants (mainly grasses), precipitation of
the warmest quarter and precipitation seasonality, with higher values
indicating wetter, seasonal grasslands (Figure S5; Table S5). The two
first axes were used as predictors in posterior analyses.
Measures of sexual dichromatism and size dimorphism were derived from
Chen et al. (2013). Briefly, sexual size dimorphism was calculated using
the index of Lovich and Gibbons (1992), where sexual dimorphism index
(SDI) = [(mean size of male)/(mean size of female)] - 1. Mean male
and female size (snout-vent length, SVL) measures were derived from the
literature and measured from museum specimens (Chen et al. 2012;
Chen et al. 2013). The index of sexual dichromatism was derived
from scores of sex differences in the hue or intensity of colour
patterns for each of 9 body regions, with 0 = no difference; 1 =
difference in colour intensity or pattern and 2 = entirely different
colour or difference in both colour and pattern (Ostman & Stuart-Fox
2011; Chen et al. 2012). Colours that may be generated by the
same mechanism (e.g. yellow, orange and red) or that may reflect
differences in descriptors used in field guides (e.g. cream, white) were
scored as differences in colour intensity (1). Scores for the nine body
regions were summed to derive a measure of overall sexual dichromatism
ranging from 0–18.