Fig 1.
Model scheme of the standard DEB model.
Boxes: state variables. Arrows: mass/ energy fluxes associated with each of the processes specified on the scheme. Each process is quantified by a model parameter (see Table 1). The only process that is not represented here is aging, which is quantified by two parameters.
Table 1.
The 14 primary parameters of the std-DEB model in a time-length-energy frame [4].
The values are considered typical values among species at 20°C with maximum structural length for a dimensionless zoom factor z and
cm. Structural length is the cubic root of structural volume, which, together with reserve, contributes to body mass. The Gompertz stress coefficient is almost zero for ectotherms and around 0.1 for endotherms, while the Weibull aging acceleration varies greatly between species.
Fig 2.
Top figure: Architecture of the AmP procedure. The rounded (green) rectangles represent files that are set by the user: run_my_pet, mydata_my_pet, pars_init_my_pet, predict_my_pet. The specific inputs for each file are specified (see text for details). For example: the user must set data and weights in mydata_my_pet etc. The (orange) rectangles represent the output. All the estimated parameters are stored in results_my_pet.mat. Figures can be stored as .png. Parameters in the .mat file can be used to generate an html file which reports all types of DEB quantities. The estimation procedure (blue oval) is further detailed in the bottom figure. Bottom: conceptual overview of the four elements underlying the AmP parameter estimation procedure (blue ovals). The information that is used as input for each of the elements is represented in rounded (green) rectangles. The elements represented by the ovals perform the computations. The output of the estimation procedure (the parameters) is represented by the (orange) rectangle. The results returned by the procedure can then be used to restart the procedure (continuation method described in the text).
Table 2.
Default weight settings for the pseudo-data according to loss functions “sb” or “su”, see Methods (loss function) for their definition.
Fig 3.
The number of species added to AmP in time and the relative frequency of taxa.
Fig 4.
Goodness of fit of the DEB model to empirical data in the AmP collection: The survival function of MRE (blue) and SMSE (red), with their median values and the relationship between MRE and SMSE.
See Methods for the definition of the MRE and SMSE.
Fig 5.
Model performance (in terms of mean relative error, MRE) with respect to the level of completeness of the data all parameters were estimated with loss function ‘sb’.
Fig 6.
Statistics of the AmP collection: The relative frequency of standard-like (s-), acceleration (a-), and insect-like (h-) models.
Fig 7.
The metabolic acceleration factor in animal taxa, quantified as the ratio of length after and before acceleration.
The font colors of the taxa names indicate the values of the acceleration factor among their species: less than 2 (black), 5 (blue), 10 (magenta) or more than 10 (red).
Fig 8.
The acceleration factor, i.e. the ratio of lengths after and before acceleration, for the various taxa of Crustacea (top) and Mollusca (bottom).
The font colors indicate acceleration by a factor less than 2 (black), 5 (blue), 10 (magenta) or more than 10 (red).
Fig 9.
The acceleration factor, i.e. the ratio of lengths after and before acceleration, for the various taxa of the ray-finned fish.
The font colors indicate acceleration by a factor less than 2 (black), 5 (blue), 10 (magenta) or more than 10 (red).