derive_tcrit() returns the rate-multiplier critical temperature T_crit: the
temperature at which the thermal-damage rate falls to a chosen low floor
rate. It is the maximum-likelihood analogue of the bayesTLS
extract_tdt() absolute-family T_crit, and follows directly from the fitted
CTmax and z:
$$T_{crit} = CTmax + z \, \log_{10}(rate / 100).$$
Because rate < 100 makes log10(rate / 100) < 0 and z > 0, T_crit sits
below CTmax: it is the lower thermal threshold at which damage becomes
negligible (the temperature cutoff a heat-injury accumulation model treats as
"no damage").
Arguments
- object
A
profile_tlsfit fromfit_tls().- rate
Damage-rate floor(s), a percentage of the lethal dose per hour (strictly positive). A scalar or a vector; default
1.- group
Optional single group level (grouped fits only; required when the fit is grouped).
Details
rate is a damage-rate floor expressed as a percentage of the lethal dose
per hour; bayesTLS brackets observed breakpoints with a default range of
0.1–1 %/hour. Unlike the Bayesian path, which samples rate to fold an
operational choice into the posterior, freqTLS treats rate as a fixed
input and returns the deterministic transform of the fitted CTmax and z
(combine their confidence intervals if you need to propagate uncertainty).
For a random-effects fit this is a population-level derived quantity; it does
not add a group BLUP.
T_crit assumes a lethal endpoint: it is a damage-accumulation concept, so
for sublethal endpoints (knockdown, photosynthetic failure) the steeper z
drives it implausibly low. derive_tcrit() says so, once per call.
See also
derive_ctmax() for the absolute-threshold critical temperature.
Examples
d <- simulate_tls(family = "binomial", CTmax = 36, z = 4, seed = 1)
fit <- fit_tls(d, y = survived, n = total, time = duration, temp = temp,
family = "binomial", tref = 1)
derive_tcrit(fit, rate = c(0.1, 1)) # lower thermal thresholds
#> `T_crit` assumes a lethal endpoint; for sublethal data its steeper `z` makes it
#> implausibly low.
#> [1] 23.93177 27.92980