Fit nonlinear regression models to data representing how development rate changes
with temperature (known as Thermal Performance Curves), based on
nls.multstart::nls_multstart() approach to development rate data across temperatures.
The fitting procedure is built upon previous packages for starting values estimation,
namely rTPC and devRate.
Source
The dataset used in the example was originally published in Satar & Yokomi (2022) under the CC-BY-NC license
Arguments
- temp
a vector of temperatures used in the experiment. It should have at least four different temperatures and must contain only numbers without any missing values.
- dev_rate
a vector of estimated development rates corresponding to each temperature. These rates are calculated as the inverse of the number of days to complete the transition from the beginning of a certain life stage to the beginning of the following at each temperature. It must be numeric and of the same length as
temp.- model_name
a string or a vector that specifies the model(s) to use for fitting the Thermal Performance Curves. Options include "all" or specific models listed in available_models. These models typically exhibit a common unimodal, left-skewed shape.
Value
A table in tibble format with estimates and standard errors
for each parameter of the models specified by the user that have adequately
converged. Models are sorted based on their Akaike Information Criterion (AIC) values,
with the best fitting models shown first. Fitted models are also provided in list format
in the model_list column and can be accessed using get_fitted_model() for
for further inspection.
It is important to consider ecological criteria alongside statistical information.
For additional help in model selection,
we recommend using plot_devmodels() and consulting relevant literature.
References
Angilletta, M.J., (2006). Estimating and comparing thermal performance curves. J. Therm. Biol. 31: 541-545. (for model selection in TPC framework)
Padfield, D., O'Sullivan, H. and Pawar, S. (2021). rTPC and nls.multstart:
A new pipeline to fit thermal performance curves in R. Methods Ecol Evol. 12: 1138-1143.
Rebaudo, F., Struelens, Q. and Dangles, O. (2018). Modelling temperature-dependent
development rate and phenology in arthropods: The devRate package for R.
Methods Ecol Evol. 9: 1144-1150.
Satar, S. and Yokomi, R. (2002). Effect of temperature and host on development of Brachycaudus schwartzi (Homoptera: Aphididae). Ann. Entomol. Soc. Am. 95: 597-602.
See also
nls.multstart::nls_multstart() for structure of model fitting approach
devRate::devRateEqList() for information on several equations
browseVignettes("rTPC") for model names, start values searching workflows and
bootstrapping procedures using both rTPC and nls.multstart packages.
Examples
data("aphid")
fitted_tpcs <- fit_devmodels(temp = aphid$temperature,
dev_rate = aphid$rate_value,
model_name = c("lactin2", "briere2", "mod_weibull")
)
#> fitting model lactin2
#> fitting model briere2
#> fitting model mod_weibull
fitted_tpcs
#> # A tibble: 12 × 8
#> model_name param_name start_vals param_est param_se model_AIC model_BIC
#> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 lactin2 a 0.119 0.0692 0.0463 -42.2 -42.5
#> 2 lactin2 b -0.254 -0.283 0.402 -42.2 -42.5
#> 3 lactin2 tmax 32.5 41.3 7.75 -42.2 -42.5
#> 4 lactin2 delta_t 7.5 13.5 6.98 -42.2 -42.5
#> 5 briere2 tmin 15 10.0 2.33 -42.6 -42.9
#> 6 briere2 tmax 32.5 36.5 3.40 -42.6 -42.9
#> 7 briere2 a 0.0002 0.0000416 0.0000534 -42.6 -42.9
#> 8 briere2 b 3 1.12 0.594 -42.6 -42.9
#> 9 mod_weibull a 0.112 0.143 0.00495 -44.7 -45.0
#> 10 mod_weibull topt 25 26.8 0.520 -44.7 -45.0
#> 11 mod_weibull b 17.5 30.0 16.2 -44.7 -45.0
#> 12 mod_weibull c 4 4.27 2.60 -44.7 -45.0
#> # ℹ 1 more variable: model_fit <list>